A risk assessment method for power system nodes under extreme high temperature conditions

Abstract

The present invention discloses a method for assessing the risk of power system nodes under extremely high temperatures. The method includes: establishing a temperature-dependent power load linear regression model to output the power system's power load; inputting the model into a power load forecasting model to obtain a load Gaussian distribution curve, and using the Monte Carlo sampling method to perform multiple sampling to select preselected load values for power nodes; inputting the model into a DC power flow optimization model to output the final load of the power node; obtaining the power node's power generation cost parameter, obtaining the power node's risk assessment index, and performing a risk assessment on the power node under extremely high temperatures. The method of the present invention can effectively identify the risk distribution of power nodes under extreme weather conditions, providing effective guidance for establishing risk prevention and control strategies for power systems under extreme weather conditions.

Description

Technical Field

[0001] The present invention relates to a node risk assessment method, and in particular to a node risk assessment method for a power system under extremely high temperatures. Background Art

[0002] In recent years, with the frequent occurrence of extreme heat waves, the power system has faced widespread blackouts and prolonged peak cost parameters. In extreme heat, the demand for air conditioning and cooling drives a sharp surge in power load, leading to a supply shortage and increased line congestion in the power system. This impacts the normal power consumption of residential users, who experience the dual pressures of forced power outages and soaring electricity cost parameters. Therefore, in response to extreme heat waves, it is necessary to conduct a comprehensive assessment of the node risks in the power system. Summary of the Invention

[0003] In order to solve the problems existing in the background technology, the present invention provides a method for risk assessment of power system nodes under extreme high temperatures.

[0004] The technical solution adopted in the present invention is:

[0005] The power system node risk assessment method of the present invention comprises the following steps:

[0006] 1) Establish a temperature-power load linear regression model for the power system under extreme high temperatures. The temperature under extreme high temperatures is input into the temperature-power load linear regression model, and the temperature-power load linear regression model outputs the power load of the power system under extreme high temperatures.

[0007] 2) Establish an electric load forecasting model for the power system under extreme high temperatures, input the electric load of the power system under extreme high temperatures into the electric load forecasting model, and the electric load forecasting model obtains the load Gaussian distribution curve of each power node of the power system under extreme high temperatures. For each power node, the load Gaussian distribution curve of the power node is normalized and discretized and then divided into seven Gaussian distribution areas. The power load forecasting model outputs the median load value in each Gaussian distribution area; for each power node in the power system, the Monte Carlo sampling method is used to perform multiple sampling to select the median load value in one of the Gaussian distribution areas as the pre-selected load value of the power node.

[0008] 3) While satisfying power constraints, a DC power flow optimal model for the power system under extreme high temperatures is established. For each power node in the power system, the preselected load of the power node selected after each sampling in step 2) is input into the DC power flow optimal model. The DC power flow optimal model outputs the optimal load of the power node. This is performed until the number of samplings exceeds a preset number of samplings and the DC power flow optimal model is optimal. The optimal load of the power node output by the DC power flow optimal model after the current sampling is obtained as the final load of the power node. The optimal DC power flow optimal model is optimal when the sum of the power system's generation cost and load reduction cost is minimized under all sampling conditions.

[0009] 4) For each power node in the power system, the power node's generation cost parameter is calculated based on the final load of the power node obtained in step 3). A risk assessment index for the power node is then obtained based on the power node's generation cost parameter. Based on the risk assessment index, a risk assessment is performed on power nodes exposed to extreme high temperatures. This allows for the identification of several power nodes with high loads, prompting early warnings and the implementation of appropriate measures. The cost is specifically a quantity related to power generation.

[0010] In step 1), the temperature power load linear regression model of the power system under extreme high temperature is specifically as follows:

[0011] ln(Load(t))=ln(Load NR (t))+α ED (τ(t)-τ NR (t))

[0012] Among them, Load(t) and Load NR (t) represent the preset reference temperature τ(t) under extreme high temperature and normal summer day, respectively. NR (t) the power load of the power system in the next period t; α ED Represents the parameters of the power load linear regression model.

[0013] The temperature τ(t) under extreme high temperature is input into the temperature power load linear regression model, and the temperature power load linear regression model outputs the power load Load(t) of the power system under the extreme high temperature τ(t) during time period t.

[0014] In step 2), the power load prediction model for the power system under extreme high temperature conditions is established as follows:

[0015]

[0016] in, represents the median load value in one of the Gaussian distribution areas of the load Gaussian distribution curve of the i-th power node in the power system under extreme high temperature during period t; N() represents Gaussian distribution, Load i (t) represents the predicted load of the i-th power node in the power system during period t under extreme high temperature; η ED represents the error coefficient of the power load forecasting model; Load(t) represents the power load of the power system in time period t under extremely high temperature τ(t).

[0017] The power load forecasting model is a multi-state model, and the load value of the power node is the median of the load value in one of the Gaussian distribution areas. Probability is the area of the corresponding Gaussian distribution region.

[0018] In step 3), an optimal DC power flow model of the power system under extreme high temperature is established while satisfying power constraints, as follows:

[0019]

[0020] Among them, f l represents the sum of the power generation cost and load reduction cost of the power system under the lth sampling; the power system includes several generators, each generator is located at its own power node, n g and n ED Respectively represent the total number of generator sets and the total number of power nodes in the power system; C G,m represents the power generation cost of the mth generator set in the power system, C D,i represents the load reduction cost of the i-th power node in the power system; represents the active output of the mth generator set in the power system during the time period t under the lth sampling period, It represents the load reduction of the i-th power node in the power system during the time period t under the l-th sampling.

[0021] The power constraints are as follows:

[0022] a) Power balance constraints:

[0023]

[0024] in, It represents the load of the i-th power node in the power system during the time period t under the l-th sampling.

[0025] b) Load reduction constraints:

[0026]

[0027]

[0028] in, Represents the load Gaussian distribution curve under the lth sampling period The median load value in the jth Gaussian distribution area selected in is the preselected load of the i-th power node in the power system in time period t under the l-th sampling; r represents a random number sequence that obeys the standard normal distribution; and They represent the median load values in the j-1th and j+1th Gaussian distribution regions, respectively.

[0029] c) Generator output constraints:

[0030]

[0031] in, It represents the maximum active output of the mth generator set in the power system during the period t under the lth sampling.

[0032] d) Line flow constraints:

[0033]

[0034] Among them, F ik represents the flow of the power line between the i-th power node and the k-th power node in the power system, and They represent the minimum flow and maximum flow of the power line between the i-th power node and the k-th power node in the power system, respectively.

[0035] The preselected load of the power node selected after each sampling in step 2) is input into the DC power flow optimal model, and the DC power flow optimal model outputs the optimal load of the power node, that is, the preselected load of the i-th power node in the power system in time period t under the l-th sampling Input the DC power flow optimal model, and the DC power flow optimal model outputs the load of the i-th power node in the power system during the period t under the l-th sampling period.

[0036] In step 4), for each power node in the power system, the power generation cost parameter of the power node is calculated based on the final load of the power node obtained in step 3), as follows:

[0037]

[0038] Among them, ρ i,l (t) represents the power generation cost parameter of the i-th power node in the power system during the time period t under the l-th sampling; Ls represents the Lagrangian function operator; It represents the load of the i-th power node in the power system during the time period t under the l-th sampling.

[0039] In step 4), the risk assessment index of the power node is obtained according to the power generation cost parameter of the power node. The risk assessment index of the power node includes an expected node cost parameter index and a node cost parameter standard deviation index, which are as follows:

[0040] a) Expected node cost parameter index:

[0041]

[0042] in, is the standard deviation index of node cost parameter; p l (t) is the probability of the lth sampling in time period t. When the number of sampling is large, it can be approximately considered as 1 / L, where L is the total number of sampling; r i,l (t) represents the power generation cost parameter of the i-th power node in the power system during time period t under the l-th sampling.

[0043] b) Node cost parameter standard deviation index:

[0044]

[0045] Among them, σ i (t) represents the standard deviation index of the node cost parameter; L represents the total number of sampling times.

[0046] The risk assessment of power nodes under extreme high temperatures is performed according to the risk assessment index of the power nodes. That is, the larger the expected node cost parameter index and the node cost parameter standard deviation index are, the greater the risk of the i-th power node in the power system.

[0047] The beneficial effects of the present invention are:

[0048] This method can effectively identify the node risk distribution under extreme weather conditions and provide effective guidance for the power system to establish risk prevention and control strategies under extreme weather conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 The expected node cost parameter distribution of a certain period of time in the present invention;

[0050] Figure 2 Flow chart of the method of the present invention. DETAILED DESCRIPTION

[0051] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0052] The power system node risk assessment method of the present invention is as follows Figure 2 The following steps are shown:

[0053] 1) Establish a temperature-power load linear regression model for the power system under extreme high temperatures. The temperature under extreme high temperatures is input into the temperature-power load linear regression model, and the temperature-power load linear regression model outputs the power load of the power system under extreme high temperatures.

[0054] In step 1), the temperature power load linear regression model of the power system under extreme high temperature is established as follows:

[0055] ln(Load(t))=ln(Load NR (t))+α ED (τ(t)-τ NR (t))

[0056] Among them, Load(t) and Load NR (t) represent the preset reference temperature τ(t) under extreme high temperature and normal summer day, respectively. NR (t) the power load of the power system in the next period t; α ED Represents the parameters of the power load linear regression model.

[0057] The temperature τ(t) under extreme high temperature is input into the temperature power load linear regression model, and the temperature power load linear regression model outputs the power load Load(t) of the power system under the extreme high temperature τ(t) during time period t.

[0058] 2) Establish an electric load forecasting model for the power system under extreme high temperatures, input the electric load of the power system under extreme high temperatures into the electric load forecasting model, and the electric load forecasting model obtains the load Gaussian distribution curve of each power node of the power system under extreme high temperatures. For each power node, the load Gaussian distribution curve of the power node is normalized and discretized and then divided into seven Gaussian distribution areas. The power load forecasting model outputs the median load value in each Gaussian distribution area; for each power node in the power system, the Monte Carlo sampling method is used to perform multiple sampling to select the median load value in one of the Gaussian distribution areas as the pre-selected load value of the power node.

[0059] In step 2), the power load forecasting model for the power system under extreme high temperature conditions is established as follows:

[0060]

[0061] in, represents the median load value in one of the Gaussian distribution areas of the load Gaussian distribution curve of the i-th power node in the power system under extreme high temperature during period t; N() represents Gaussian distribution, Load i (t) represents the predicted load of the i-th power node in the power system during period t under extreme high temperature; η ED represents the error coefficient of the power load forecasting model; Load(t) represents the power load of the power system in time period t under extremely high temperature τ(t).

[0062] The power load forecasting model is a multi-state model, and the load value of the power node is the median of the load value in one of the Gaussian distribution areas. Probability is the area of the corresponding Gaussian distribution region.

[0063] 3) Under the condition that power constraints are satisfied, a DC power flow optimal model of the power system under extremely high temperatures is established. For each power node in the power system, the preselected load of the power node selected after each sampling in step 2) is input into the DC power flow optimal model. The DC power flow optimal model outputs the optimal load of the power node until the number of samplings exceeds the preset number of samplings and the DC power flow optimal model is optimal. The optimal load of the power node output by the DC power flow optimal model after the current sampling is obtained as the final load of the power node. The optimal situation of the DC power flow optimal model is that the sum of the power generation cost and the load reduction cost of the power system obtained in all sampling situations is minimized.

[0064] In step 3), the optimal DC power flow model of the power system under extreme high temperature is established while satisfying the power constraints, as follows:

[0065]

[0066] Among them, f l represents the sum of the power generation cost and load reduction cost of the power system under the lth sampling; the power system includes several generators, each generator is located at its own power node, n g and n ED Respectively represent the total number of generator sets and the total number of power nodes in the power system; C G,m represents the power generation cost of the mth generator set in the power system, C D,i represents the load reduction cost of the i-th power node in the power system; represents the active output of the mth generator set in the power system during the time period t under the lth sampling period, It represents the load reduction of the i-th power node in the power system during the time period t under the l-th sampling.

[0067] The power constraints are as follows:

[0068] a) Power balance constraints:

[0069]

[0070] in, It represents the load of the i-th power node in the power system during the time period t under the l-th sampling.

[0071] b) Load reduction constraints:

[0072]

[0073]

[0074] in, Represents the load Gaussian distribution curve under the lth sampling period The median load value in the jth Gaussian distribution area selected in is the preselected load of the i-th power node in the power system in time period t under the l-th sampling; r represents a random number sequence that obeys the standard normal distribution; and They represent the median load values in the j-1th and j+1th Gaussian distribution regions, respectively.

[0075] c) Generator output constraints:

[0076]

[0077] in, It represents the maximum active output of the mth generator set in the power system during the period t under the lth sampling.

[0078] d) Line flow constraints:

[0079]

[0080] Among them, F ik represents the flow of the power line between the i-th power node and the k-th power node in the power system, and They represent the minimum flow and maximum flow of the power line between the i-th power node and the k-th power node in the power system, respectively.

[0081] The preselected load of the power node selected after each sampling in step 2) is input into the DC power flow optimal model, and the DC power flow optimal model outputs the optimal load of the power node, that is, the preselected load of the i-th power node in the power system in time period t under the l-th sampling Input the DC power flow optimal model, and the DC power flow optimal model outputs the load of the i-th power node in the power system during the period t under the l-th sampling period.

[0082] 4) For each power node in the power system, the power node's generation cost parameter is calculated based on the final load of the power node obtained in step 3). A risk assessment index for the power node is then obtained based on the power node's generation cost parameter. Based on the risk assessment index, a risk assessment is performed on power nodes exposed to extreme high temperatures. This allows for the identification of several power nodes with high loads, prompting early warnings and the implementation of appropriate measures. The cost is specifically a quantity related to power generation.

[0083] In step 4), for each power node in the power system, the power generation cost parameter of the power node is calculated based on the final load of the power node obtained in step 3), as follows:

[0084]

[0085] Among them, ρ i,l (t) represents the power generation cost parameter of the i-th power node in the power system during the time period t under the l-th sampling; L s represents the Lagrangian function operator; It represents the load of the i-th power node in the power system during the time period t under the l-th sampling.

[0086] In step 4), the risk assessment index of the power node is obtained according to the power generation cost parameter of the power node. The risk assessment index of the power node includes the expected node cost parameter index and the node cost parameter standard deviation index, which are as follows:

[0087] a) Expected node cost parameter index:

[0088]

[0089] in, is the standard deviation index of node cost parameter; p l (t) is the probability of the lth sampling in time period t. When the number of sampling is large, it can be approximately considered as 1 / L, where L is the total number of sampling; ρ i,l (t) represents the power generation cost parameter of the i-th power node in the power system during time period t under the l-th sampling.

[0090] b) Node cost parameter standard deviation index:

[0091]

[0092] Among them, σ i(t) represents the standard deviation index of the node cost parameter; L represents the total number of sampling times.

[0093] The risk assessment of power nodes under extreme high temperatures is performed according to the risk assessment index of the power nodes. That is, the larger the expected node cost parameter index and the node cost parameter standard deviation index are, the greater the risk of the i-th power node in the power system.

[0094] The embodiments of the present invention are as follows:

[0095] 1. A modified IEEE-39 node power system was used for verification analysis. Assuming that the average temperature in a certain area during normal summer days is 28°C, which is the preset reference temperature, and when extremely high temperatures occur, 35°C, 39°C, and 42°C were selected for research. A power load prediction model was established based on these meteorological parameters.

[0096] 2. Based on the Monte Carlo sampling method, set the maximum number of sampling times N = 10000, sample the power load status, and determine the actual load status. Based on the optimal model of DC power flow in the power system, finally calculate the node risk assessment index, and the distribution of the power generation cost parameters of the expected node in a certain period of time is as follows: Figure 1 As shown in the figure, when extreme high temperatures occur, the expected node cost parameters all increase significantly compared to normal summer days. Furthermore, as the temperature continues to rise, the expected node cost parameters further increase. Nodes 3, 4, and 18 exhibit the highest risk. This is due to the high power loads these nodes carry, which can easily lead to load reduction costs. This method effectively identifies the node risk distribution under extreme high temperatures.

Claims

1. A method for risk assessment of power system nodes under extreme high temperatures, characterized by: The method comprises the following steps: 1) Establishing a temperature-power load linear regression model for the power system under extreme high temperatures. Inputting the temperature under extreme high temperatures into the temperature-power load linear regression model, the temperature-power load linear regression model outputs the power load of the power system under extreme high temperatures. 2) Establishing an electric load forecasting model for the power system under extreme high temperatures. Inputting the electric load of the power system under extreme high temperatures into the power load forecasting model, the power load forecasting model obtains the load Gaussian distribution curve of each power node in the power system under extreme high temperatures. For each power node, the load Gaussian distribution curve of the power node is normalized and discretized, and then divided into seven Gaussian distribution regions. The power load forecasting model outputs the median load value in each Gaussian distribution region. For each power node in the power system, the Monte Carlo sampling method is used to perform multiple sampling to select the median load value in one of the Gaussian distribution areas as the preselected load value of the power node; 3) Establishing a DC power flow optimal model for the power system under extreme high temperatures while satisfying power constraints. For each power node in the power system, input the preselected load of the power node selected after each sampling in step 2) into the DC power flow optimal model. The DC power flow optimal model outputs the optimal load of the power node. This is repeated until the number of samplings exceeds a preset number of samplings and the DC power flow optimal model is optimal. The optimal load of the power node output by the DC power flow optimal model after each sampling is obtained as the final load of the power node. 4) For each power node in the power system, calculate a power generation cost parameter for the power node based on the final load of the power node obtained in step 3), obtain a risk assessment index for the power node based on the power generation cost parameter, and perform a risk assessment on the power node under extreme high temperature conditions based on the risk assessment index for the power node; In step 1), the temperature power load linear regression model of the power system under extreme high temperature is specifically as follows: ln(Load(t))=ln(Load NR (t))+α ED (τ(t)-τ NR (t)) Among them, Load(t) and Load NR (t) represent the extreme high temperature τ(t) and the preset reference temperature τ in summer days, respectively. NR (t) the power load of the power system in the next period t; α ED represents the parameters of the linear regression model of power load; The temperature τ(t) under extreme high temperature is input into the temperature power load linear regression model, and the temperature power load linear regression model outputs the power load Load(t) of the power system under the extreme high temperature τ(t) in time period t; In step 2), the power load prediction model for the power system under extreme high temperature conditions is established as follows: in, represents the median load value in one of the Gaussian distribution areas of the load Gaussian distribution curve of the i-th power node in the power system under extreme high temperature during period t; N( ) represents Gaussian distribution, Load i (t) represents the predicted load of the i-th power node in the power system during period t under extreme high temperature; η ED represents the error coefficient of the power load forecasting model; Load(t) represents the power load of the power system in time period t under extremely high temperature τ(t).

2. The method for risk assessment of power system nodes under extreme high temperatures according to claim 1, characterized in that: In step 3), an optimal DC power flow model of the power system under extreme high temperature is established while satisfying power constraints, as follows: Among them, f l represents the sum of the power generation cost and load reduction cost of the power system under the lth sampling; the power system includes several generators, each generator is located at its own power node, n g and n ED Respectively represent the total number of generator sets and the total number of power nodes in the power system; C G,m represents the power generation cost of the mth generator set in the power system, C D,i represents the load reduction cost of the i-th power node in the power system; represents the active output of the mth generator set in the power system during the time period t under the lth sampling period, represents the load reduction of the i-th power node in the power system during the time period t under the l-th sampling period; The power constraints are as follows: a) Power balance constraints: in, represents the load of the i-th power node in the power system during the time period t under the l-th sampling period; b) Load reduction constraints: in, represents the median load value in one of the Gaussian distribution areas of the load Gaussian distribution curve of the i-th power node in the power system under extreme high temperature during period t; represents the median load value in the jth Gaussian distribution area selected in the lth sampling period of time period t, that is, the preselected load value of the i-th power node in the power system in the lth sampling period of time period t; r represents a random number sequence that obeys the standard normal distribution; and represent the median load values in the j-1th and j+1th Gaussian distribution regions respectively; c) Generator output constraints: in, represents the maximum active output of the mth generator set in the power system during time period t under the lth sampling period; d) Line flow constraints: Among them, F ik represents the flow of the power line between the i-th power node and the k-th power node in the power system, and denote the minimum flow and maximum flow of the power line between the i-th power node and the k-th power node in the power system respectively; The preselected load of the power node selected after each sampling in step 2) is input into the DC power flow optimal model, and the DC power flow optimal model outputs the optimal load of the power node, that is, the preselected load of the i-th power node in the power system in time period t under the l-th sampling Input the DC power flow optimal model, and the DC power flow optimal model outputs the load of the i-th power node in the power system during the period t under the l-th sampling period.

3. The method for risk assessment of power system nodes under extreme high temperatures according to claim 1, characterized in that: In step 4), for each power node in the power system, the power generation cost parameter of the power node is calculated based on the final load of the power node obtained in step 3), as follows: Among them, ρ i,l (t) represents the power generation cost parameter of the i-th power node in the power system during the time period t under the l-th sampling; L s represents the Lagrangian function operator; It represents the load of the i-th power node in the power system during the time period t under the l-th sampling.

4. The method for risk assessment of power system nodes under extreme high temperatures according to claim 1, characterized in that: In step 4), the risk assessment index of the power node is obtained according to the power generation cost parameter of the power node. The risk assessment index of the power node includes an expected node cost parameter index and a node cost parameter standard deviation index, which are as follows: a) Expected node cost parameter index: in, is the standard deviation index of node cost parameter; p l (t) is the probability of the lth sampling in time period t; ρ i,l (t) represents the power generation cost parameter of the i-th power node in the power system during the time period t under the l-th sampling period; b) Node cost parameter standard deviation index: Among them, σ i (t) represents the standard deviation index of the node cost parameter; L represents the total number of sampling times; The risk assessment of power nodes under extreme high temperatures is performed according to the risk assessment index of the power nodes. That is, the larger the expected node cost parameter index and the node cost parameter standard deviation index are, the greater the risk of the i-th power node in the power system.

Images