Power system reliability evaluation method and system considering endogenous uncertainty
Through a hybrid model-data method, the M-BPNN architecture and endogenous uncertainty model are used to solve the problem of computational complexity and dynamic changes in power system evaluation, achieving more efficient and accurate reliability evaluation, revealing that The impact of operational decisions on reliability.
Patent Information
- Application Number
- CN202410871847.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-01
- Publication Date
- 2024-11-12
AI Technical Summary
The existing power system operating reliability evaluation technology has problems such as high computational complexity, difficulty in capturing dynamic changes, difficulty in weighing speed and accuracy, and ignoring the impact of endogenous uncertainty.
A model-data-dipped hybrid power system operation reliability evaluation method is proposed, and the multi-back propagation neural network (M-BPNN) architecture is used to construct endogenous considerations through non-sequential Monte Carlo simulation and endogenous uncertainty model. Uncertainty power system reliability evaluation model, real-time data input to obtain reliability indicators.
A faster and possibly more accurate power system operation reliability assessment is achieved, and the optimization model can be solved efficiently, revealing the mechanism of impact of operation decisions on the system reliability level, and improving the accuracy and efficiency of evaluation.
Smart Images

Figure CN118940936A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of power system operation, and in particular to a method and system for evaluating the reliability of a power system taking into account inherent uncertainty. Background Art
[0002] Modern power systems rely heavily on their reliability to ensure safe and adequate supply of electrical energy. Traditional power system reliability assessments use historical data and assume static failure rates of equipment such as generators and transmission lines. However, recent studies have emphasized the temporal variation of equipment performance over operational timescales. This calls for operational reliability assessments, which have gained tremendous attention in recent decades in order to address short-term reliability issues under dynamic conditions.
[0003] The increasing complexity of power systems requires a closer connection between operational optimization and reliability assessment. This recognition has led to the incorporation of operational reliability indicators as objectives or constraints in key decision problems. These problems include unit commitment (UC), economic dispatch, reserve optimization, and power market settlement.
[0004] Previous studies have focused on exogenous uncertainties as the main factor affecting reliability indicators. These uncertainties include variability in demand and renewable energy, as well as random equipment failures. It is usually assumed that exogenous uncertainties have an independent nature from the decision-making process. However, this approach ignores the key role of endogenous uncertainties (EU) in operational reliability.
[0005] EU represents the uncertainty affected by operational optimization decisions. There is increasing evidence that operating conditions significantly affect factors such as equipment failure rates. Operational decisions directly determine these operating conditions, so operational decisions also affect the uncertainty associated with the equipment state, such as the uncertainty associated with the equipment state, as shown in the appendix to the manual. Figure 1 As shown, Figure 1 It is shown how decision variables determine the operating conditions of generation and transmission equipment, which in turn affect the equipment's force outage rates (FORs).
[0006] The presence of EU has a significant impact on operational reliability, thus requiring an integrated approach that considers optimization and assessment as interrelated processes. By incorporating endogenous uncertainty into our assessment of operational reliability, we can achieve more accurate and robust decision making, ultimately improving the safety and reliability of the power system.
[0007] Traditionally, assessing power system reliability is a time-consuming process. This is because optimal power flow (OPF) calculations are performed on an exponentially growing number of system states. To address this complexity, DC power flow models are often used as an alternative to computationally simpler AC power flow models. However, even with such simplifications, reliability assessments of provincial systems can still take several to tens of hours. The inclusion of endogenous uncertainties significantly increases the number of possible system states, further exacerbating this computationally complex challenge, making the assessment computationally more expensive. In addition, system fluctuations caused by renewable energy sources lead to dynamic changes in reliability over time. This dynamic nature makes it challenging to ensure timely assessments of operational reliability. Therefore, it is becoming increasingly important to develop faster and more efficient reliability assessment methods.
[0008] In current research, several approaches have been proposed to address the computational complexity challenges associated with operational reliability assessment. Techniques such as quick sort algorithms, state space decomposition, and dimensionality reduction can minimize the number of states considered. Similarly, reliability assessment using Monte Carlo simulation (MCS) has made advances through state space pruning, population-based intelligent search, and advanced sampling techniques such as Latin hypercube sampling. These methods mainly focus on speeding up the simulation process by reducing the number of system states explored.
[0009] Lagrange multiplier-based state enumeration (LMSE), multi-parametric linear program (MPLP), and load feasible region (LFR) methods bypass the need to perform a full optimal power flow (OPF) calculation for each state by adopting alternative methods to determine the load reduction requirements. While existing methods provide faster reliability assessments by reducing state exploration and analysis, they struggle to keep up with the dynamic characteristics of renewable energy sources. Any fluctuation requires a comprehensive reassessment, hindering real-time early warning of risks.
[0010] In order to solve the problem of the difficulty in balancing the speed and accuracy of operational reliability assessment, this patent proposes a new method: a model-data hybrid driven power system operational reliability assessment method. A multi-backpropagation neural network (M-BPNN) architecture is used to achieve faster and potentially more accurate assessments. M-BPNN utilizes an offline training phase, in which a non-sequential Monte Carlo simulation (NMCS) is used to calculate the system reliability index under various operating conditions. After training, the system status is classified and a separate BP neural network is trained specifically for each category. During online operation, real-time data, such as wind speed, solar radiation, and load, are input into the corresponding pre-trained neural network to obtain reliability indicators. This M-BPNN method has the potential to surpass traditional machine learning algorithms in terms of the accuracy of operational reliability assessment. Summary of the invention
[0011] The present invention is proposed in view of the fact that the accurate operational reliability assessment of the existing power system is difficult to keep up with the dynamic characteristics caused by renewable energy, resulting in the need for a comprehensive reassessment when facing system fluctuations, thereby hindering the real-time risk warning.
[0012] Therefore, the problem to be solved by the present invention is that the existing power system operation reliability assessment technology has the defects of high computational complexity, difficulty in capturing dynamic changes, difficulty in balancing assessment speed and accuracy, and neglect of the impact of endogenous uncertainty.
[0013] In order to solve the above technical problems, the present invention provides the following technical solutions: a method for evaluating the operation reliability of a power system driven by a hybrid model and data that takes into account inherent uncertainty is proposed.
[0014] In the first aspect, an embodiment of the present invention provides a method for evaluating the reliability of a power system taking into account intrinsic uncertainty, which includes defining intrinsic uncertainty and constructing an intrinsic uncertainty model of the power system; constructing a reliability evaluation model of the power system taking into account intrinsic uncertainty based on the intrinsic uncertainty model of the power system; designing a reliability evaluation algorithm based on the M-BPNN model and data-driven operation according to the intrinsic uncertainty model of the power system and the reliability evaluation model of the power system taking into account intrinsic uncertainty, and evaluating the operational reliability of the power system.
[0015] As a preferred solution of the power system reliability assessment method taking into account the intrinsic uncertainty of the present invention, wherein: the construction of the power system intrinsic uncertainty model includes calculating the condition-dependent failure rate of the electrical components, and the calculation formula of the condition-dependent failure rate is as follows:
[0016]
[0017] Among them, F(I) represents the probability of forced shutdown of electrical equipment, I represents the real-time current, I0 represents the rated current of the equipment, I1 represents the tripping setting current, and α and β are parameters. It indicates the initial value of FOR of the electrical equipment when the current is less than I0; the calculation formula of FOR is as follows:
[0018]
[0019] Among them, F0 represents the probability of forced shutdown of electrical equipment when it leaves the factory, λ represents the failure rate, and μ represents the repair rate of electrical equipment; the formula for calculating parameters α and β based on historical data is as follows:
[0020]
[0021] Among them, α and β are parameters, F0 represents the probability of forced shutdown of electrical equipment when it leaves the factory, I0 represents the rated current of the equipment, and F max represents the maximum forced outage probability within the allowable range, I1 represents the trip setting current; the actual value of the current I is obtained according to the optimal power flow calculation, the optimal power flow depends on the operation decision of the power system, and the FOR of the i-th power-consuming equipment based on endogenous uncertainty is recorded as F i EU , F i EU The equivalent formula is as follows:
[0022] F i EU =F i EU (I i )=F i EU (Q(x,y)) (5)
[0023] I i =Q(x,y) (6)
[0024] Among them, F i EU represents the EU-based FOR of the i-th electrical equipment, I1 represents the tripping setting current, I i represents the actual operating current of the ith electrical equipment, x is a binary variable, y is a continuous variable, and Q(·) represents the state function between the binary variable x and the continuous variable y. In addition to the inherent constraints, the FOR of the power component is further affected by external uncertainties, including aging-related failure factors. Both the EU effect and the aging effect are incorporated into the reliability model. Assuming that the service age of an electrical equipment is t0, the FOR in the ith cycle is recorded as The calculation formula is as follows:
[0025]
[0026] in, represents the FOR in the i-th cycle, t0 represents the service age, t represents the time, H(t) represents the function of aging-related failures with respect to t, and i represents the number of the service cycle; there is a dependency between the power outage caused by EU problems and the power outage caused by equipment aging, and Formula 1 is updated to a general form, and the specific formula is as follows:
[0027]
[0028] Among them, F(·) represents the function of FOR, α age and β age Respectively represent the F0 update in equation 3 and equation 4 as The obtained parameters, I represents the real-time current, I0 represents the rated current of the equipment, and I1 represents the trip setting current. represents the FOR in the ith cycle, T i Indicates the service life.
[0029] As a preferred solution of the power system reliability assessment method taking into account endogenous uncertainty of the present invention, wherein: constructing the power system reliability assessment model taking into account endogenous uncertainty includes generating typical scenarios of endogenous uncertainty, and the typical scenario generation of endogenous uncertainty includes, assuming that in the process of power system operation reliability assessment, the scenario probability is independent of the operation decision, and the solution process precedes the equipment operation decision, then the calculation formula of the scenario probability is as follows:
[0030]
[0031] Among them, π s represents the probability of the scene occurring, φ s,up Represents the set of available devices corresponding to the scene, φ s,dn represents the set of unavailable equipment, F(·) represents the probability function of forced equipment outage, I represents the real-time current, T1 represents the first service cycle, and i represents the power system component number; the scenario probability is explicitly modeled as a function of the operation decision variable. Equation 9 represents the decision-related scenario probability P of period t. t (s), the decision-related scenario probability of period t is given by the operational decision variable and A specific combination of
[0032] π s,t =π(x,y) (10)
[0033] Among them, x represents the probability of normal operation of power system equipment, y represents the probability of failure of power system equipment, and π s,t represents the scenario probability; suppose two of the N devices fail, and equation 10 is rewritten as:
[0034]
[0035] Among them, π s,t represents the scene probability, I i,t ,I j,t They represent the currents flowing through the two faulty devices, k represents the device number of the normal operating device set, s represents the scenario, and I k,t Indicates the current flowing through the device during normal operation.
[0036] As a preferred solution of the power system reliability assessment method taking into account endogenous uncertainty described in the present invention, wherein: constructing the power system reliability assessment model taking into account endogenous uncertainty also includes reliability assessment including random unit combinations, and the reliability assessment including random unit combinations includes that the set of all results of uncertain parameters is obtained through a discrete scenario set S and its corresponding probability π s,t To indicate that D ts Denotes the electricity demand achieved in each scenario s∈S in time period t. The availability of generation capacity includes the uncertainty of power outages and intermittent sources. The first stage determines the unit commitment decision, and the second stage determines the dispatch variables. The two-stage random unit commitment problem incorporates uncertainty through a scenario-based approach. In the case of uncertain demand, after the load is known in the second stage, D ts Determining the generation level, in contrast, the combination decision must be made before this information is available and is classified as a first-stage decision, with the objective function f g Expressed as the first stage cost function Second stage cost function and demand for cost reduction c cur is the cost coefficient of demand reduction, is the total demand reduction, and the objective function of the entire two-stage planning problem is as follows:
[0037]
[0038] in, represents the cost function of the first stage of the unit commitment model, represents the cost function of the second stage of the unit commitment model, u g,t-1 represents the unit commitment decision at time t-1, u g,t represents the unit commitment decision at time t, represents the unit output decision at time t, represents the demand reduction cost; the first stage cost function The startup cost of the generator set and downtime costs The formula for calculating the sum is as follows:
[0039]
[0040] in, represents the cost function of the first stage, u g,t-1 represents the unit commitment decision at time t-1, u g,t represents the unit commitment decision at time t, represents the startup cost of the generator set, represents the shutdown cost of the generator set; the second-stage cost function is the operating cost of the generator set, and the calculation formula of the second-stage cost function is as follows:
[0041]
[0042] in, represents the cost function of the second stage, represents the scheduling variable, c g represents the operating cost coefficient of the g-th unit; the constraints of the unit combination model are as follows:
[0043]
[0044] Among them, TO represents the shortest downtime, u g,t-1 represents the combined decision at time t-1, u g,t represents the combined decision at time t, u g,k represents the combined decision at time k, TS represents the shortest running time, represents the startup cost of the generator set, represents the downtime cost of the generator set, H g and J g They represent the startup and shutdown cost coefficients of unit g respectively.
[0045] As a preferred solution of the power system reliability assessment method taking into account the intrinsic uncertainty described in the present invention, the constraints of the second stage include forcing power balance on each bus, controlling the power flow through the line, ensuring that the line power flow remains within the limit range, ensuring that the generator output remains within the limit range, limiting the upward and downward adjustment of power output, limiting the degree of demand reduction, setting phase angle boundaries and introducing reliability-specific constraints.
[0046] As a preferred solution of the power system reliability assessment method taking into account endogenous uncertainty described in the present invention, wherein: designing a reliability assessment algorithm based on the M-BPNN model and data-driven operation includes organizing the BPNN into interconnected artificial neuron layers, the artificial neuron layers including an input layer, a hidden layer and an output layer; the input layer receives raw data representing the state of the power system and performs reliability assessment, the raw data including power generation, power demand of each grid point and the operating state of the components; the hidden layer performs core information processing; the output layer outputs a single value representing the probability of system failure or outputs a set of values representing the reliability of different grid components; designing a reliability assessment algorithm based on the M-BPNN model and data-driven operation also includes that during forward propagation, the activation calculation of the hidden neurons is as follows:
[0047] y j =f J (μ j +ω j y m )
[0048] Among them, y j Indicates that the jth hidden layer receives input from the previous layer, f J (·) represents the activation function of each neuron in the jth hidden layer, ω j represents the weights connecting the previous layer to the neurons in the jth hidden layer, μ j represents the threshold of each neuron in the jth hidden layer; the expression of the relationship between the current hidden layer and the next hidden layer is as follows:
[0049] y j+1 =f J (μ j+1 +ω j+1 y j )
[0050] Among them, y j+1 represents the output of the j+1th hidden layer, f J (·) represents the activation function of each neuron in the jth hidden layer, ω j represents the weights connecting the previous layer to the neurons in the jth hidden layer, μ j represents the threshold of each neuron in the jth hidden layer; the expression of the relationship between the hidden layer and the output layer is as follows:
[0051] y t =f o (u o +ω o y j,max )
[0052] Among them, y tIndicates that the model predicts the value at each time point t, f o (·) represents the activation function, u o represents the threshold applied to the output layer, ω o represents the weight connecting the final hidden layer and the output layer, y j,max represents the maximum predicted output of the jth hidden layer; the calculation formula for the error between the predicted output and the actual output is as follows:
[0053] δ y =y t -y
[0054] Among them, δ y Represents the error between the predicted output and the actual output, y t It indicates that the model predicts the value at each time point t, and y represents the actual output value; the formula for calculating the unit error term in the jth hidden layer is as follows:
[0055]
[0056] Among them, δ j represents the unit error in the jth hidden layer, represents the weight of the jth hidden layer, δ j+1 represents the unit error in the j+1th hidden layer, represents the normalized activation function of each neuron in the Jth hidden layer, f J represents the activation function of each neuron in the Jth hidden layer; the weight coefficient of hidden layer j is updated as:
[0057]
[0058] Among them, ω j represents the weight coefficient of hidden layer j, δ j+1 represents the unit error in the j+1th hidden layer, Represents the predicted value of the hidden layer j of the model.
[0059] As a preferred solution of the power system reliability assessment method taking into account endogenous uncertainty described in the present invention, the step of using an adaptive algorithm to find the optimal architecture of the M-BPNN model includes: inputting training and test data sets into the system; setting the maximum number of hidden layers and units per layer; using a cyclic process to iteratively evaluate various configurations and the number of elements in the layer while keeping them within predefined limits; cyclically using the selected hidden layer and element configuration to iteratively train the back propagation neural network; using the BP neural network to calculate reliability indicators and errors; cyclically iterating until all possible combinations of hidden layers and hidden elements are enumerated; after evaluating all unit combinations, finding the combination that results in the minimum error; and calculating the reliability index using the identified combination of hidden layers and hidden elements.
[0060] In the second aspect, an embodiment of the present invention provides an electric power system reliability assessment system taking into account intrinsic uncertainty, which includes a definition module for defining intrinsic uncertainty and constructing an intrinsic uncertainty model of the electric power system; a model construction module for constructing an electric power system reliability assessment model taking into account intrinsic uncertainty based on the electric power system intrinsic uncertainty model; an algorithm design module for designing a reliability assessment algorithm based on the M-BPNN model and data-driven operation according to the electric power system intrinsic uncertainty model and the electric power system reliability assessment model taking into account intrinsic uncertainty, and evaluating the operation reliability of the electric power system.
[0061] In a third aspect, an embodiment of the present invention provides a computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: when the computer program instructions are executed by the processor, the steps of the method for evaluating the reliability of a power system taking into account intrinsic uncertainties as described in the first aspect of the present invention are implemented.
[0062] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program instructions are executed by a processor, the steps of the method for evaluating the reliability of a power system taking into account intrinsic uncertainties as described in the first aspect of the present invention are implemented.
[0063] The beneficial effects of the present invention are as follows: the traditional power system reliability assessment model only considers the impact of exogenous uncertainty and cannot characterize the impact mechanism of power system operation decisions on the system reliability level. The present invention incorporates endogenous uncertainty into the power system operation reliability assessment process, reveals the mechanism of operation decisions on the system reliability level; establishes a reliability-constrained unit combination model to assess the impact of endogenous uncertainty on operation reliability; proposes an M-BPNN-based model and a data-driven operation reliability assessment algorithm, which can efficiently solve the power system operation reliability assessment optimization model. BRIEF DESCRIPTION OF THE DRAWINGS
[0064] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0065] Figure 1 The figure is a comparison between the traditional power system reliability assessment and the power system reliability assessment considering the inherent uncertainty.
[0066] Figure 2 This is the EU-based power system operation reliability assessment diagram.
[0067] Figure 3 This is the BPNN structure diagram.
[0068] Figure 4 This is a comparison chart of the absolute errors of the four methods and the standard method. DETAILED DESCRIPTION
[0069] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the accompanying drawings.
[0070] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.
[0071] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.
[0072] Example 1
[0073] Reference Figure 1 to Figure 3 , which is the first embodiment of the present invention, and provides a method for evaluating the reliability of a power system taking into account endogenous uncertainties, comprising:
[0074] S1: Define endogenous uncertainty and construct an endogenous uncertainty model for the power system.
[0075] Specifically, a clear distinction is made between endogenous uncertainty and its counterpart, exogenous uncertainty, which refers to situations where there is inherent, substantial uncertainty. Exogenous uncertainty can be based on historical data or expert knowledge to establish the underlying probability distribution of random parameters. In contrast, endogenous uncertainty arises when decisions made at a specific point in time significantly affect the level of uncertainty encountered later.
[0076] Furthermore, in the field of endogenous uncertainty, there are two main categories of models. The first category: decisions can change the probability distribution of uncertain parameters. Although the decision-maker's behavior cannot completely eliminate uncertainty, it can increase the probability of some results and reduce the probability of other results. The second category: decision-makers can control existing uncertainty through decisions.
[0077] Furthermore, the present invention delves into the complexity of the first category, a specific type of endogenous uncertainty model. Although endogenous uncertainty has received attention in power system expansion planning, its application in operational reliability assessment remains less studied. Existing studies have explored the role of EU (endogenous uncertainty) in solar and concentrated solar power plant investment planning. However, there is a lack of comprehensive understanding of the impact of EU on operational reliability assessment. Ignoring these multifaceted influences may lead to overly optimistic or pessimistic assessments, which may mislead decisions on power system operation and lead to suboptimal results.
[0078] Furthermore, EU models the relationship between decision variables and the FOR (Forced Outage Rate) of electrical equipment. The service life of equipment usually shortens with the increase of operating temperature and reaction rate. Components subjected to continuous high temperature will shorten their service life and increase the probability of failure accordingly. This is because the increase of current flowing through the component usually causes its temperature to rise. The calculation formula of the condition-dependent failure rate (CDFR) of electrical components is as follows:
[0079]
[0080] Among them, F(I) represents the forced outage rate (FOR) of the electrical equipment, I represents the real-time current, I0 represents the rated current of the equipment, I1 represents the tripping setting current, α and β are parameters, Indicates the initial value of FOR of the electrical equipment when the current is less than I0.
[0081] Furthermore, the calculation formula for the forced shutdown probability of electrical equipment when it leaves the factory is as follows:
[0082]
[0083] Among them, F0 represents the probability of forced shutdown of electrical equipment when it leaves the factory, λ represents the failure rate, and μ represents the repair rate of electrical equipment, which is set according to historical data.
[0084] Furthermore, the parameters α and β are calculated based on historical data as follows:
[0085]
[0086] Among them, α and β are parameters, F0 represents the probability of forced shutdown of electrical equipment when it leaves the factory, I0 represents the rated current of the equipment, and F max It represents the maximum forced outage probability within the allowable range, and I1 represents the tripping setting current;
[0087] Furthermore, the actual value of the current I can be obtained according to the optimal power flow calculation. The optimal power flow depends on the operation decision of the power system, such as the decision of the unit combination. Therefore, the EU-based FOR of the i-th power device is denoted as F i EU , F i EU The equivalent formula is as follows:
[0088] F i EU =F i EU (I i )=F i EU (Q(x,y)) (5)
[0089] I i =Q(x,y) (6)
[0090] Among them, F i EU represents the EU-based FOR of the i-th electrical equipment, I1 represents the tripping setting current, I i represents the actual operating current of the ith electrical equipment, x is a binary variable, which represents the decision of UC, y represents a continuous variable, which is related to the power generation plan, etc., and Q(·) represents a binary variable x and the continuous variable y.
[0091] Furthermore, in addition to the inherent constraints, the FOR of power components is further affected by external uncertainties. These uncertainties include factors such as aging-related failures. Therefore, both EU effects and aging effects are incorporated into the reliability models of these devices. Assuming that the service age of an electrical device is t0, the FOR in the i-th cycle is denoted by The calculation formula is as follows:
[0092]
[0093] in, represents the FOR in the i-th cycle, t0 represents the service age, t represents the time, H(t) represents the function of aging-related failures with respect to t, and i represents the number of the service cycle.
[0094] Furthermore, there may be a dependency between the power outages caused by EU problems and those caused by equipment aging. Equipment with a longer service life tends to be more sensitive to operational changes, thereby increasing the sensitivity to problems caused by endogenous uncertainties. Therefore, formula (1) can be restated in a more general form. The specific formula is as follows:
[0095]
[0096] Where F(·) represents the forced outage probability function, α age and β age The F0 in equations (3) and (4) is updated as The obtained parameters, I represents the real-time current, I0 represents the rated current of the equipment, and I1 represents the trip setting current. represents the FOR in the ith cycle, T i Indicates the service life.
[0097] S2: Based on the intrinsic uncertainty model of the power system, a power system reliability assessment model considering the intrinsic uncertainty is constructed.
[0098] Specifically, we analyze the probability density functions (PDFs) of the decision variables in the optimization model for the randomness of control equipment failures. The PDFs of the decision variables for the randomness of control equipment failures will directly affect the overall reliability of the power system. By directly incorporating operational reliability into the optimization objective, we achieve a closer coordination between power system operation and its reliability assessment.
[0099] Furthermore, the instruction manual is attached Figure 2 An operational reliability assessment framework that includes endogenous uncertainty is proposed. The framework addresses the interaction between operational decisions and reliability assessment by introducing reliability indicators as objectives and constraints. In addition, the inherent uncertainty of equipment status is captured to obtain more accurate assessment results and optimal sequential unit commitment (SUC) decisions. System reliability indicators from various scenarios are used as input data for the neural network training process.
[0100] Furthermore, a multiple back-propagation neural network (M-BPNN) is used to establish a highly nonlinear mapping between operation scenarios and system reliability indicators. This can significantly reduce the computational time required for power system operation reliability assessment, especially when renewable energy penetration is high.
[0101] Furthermore, based on the SUC-EU model, the interaction between operation optimization and reliability assessment is explored in more depth. Considering “Nk” faults (where k is greater than 1) is conducive to generating a wider range of potential system operation scenarios.
[0102] Furthermore, the typical scenario generation of endogenous uncertainty includes: In order to ensure the accuracy of the scenario generation process involving random unit combinations, the present invention assumes that in the process of power system operation reliability assessment, the scenario probability is independent of the operation decision, and the solution process precedes the equipment operation decision. The calculation formula of the scenario probability is as follows:
[0103]
[0104] Among them, π s represents the probability of the scene occurring, φ s,up Represents the set of available devices corresponding to the scene, φ s,dn represents the set of unavailable equipment, F(·) represents the probability function of forced outage of equipment, I represents the real-time current, T1 represents the first service cycle, and i represents the number of power system components.
[0105] Furthermore, within this framework, scenario probabilities are explicitly modeled as operational decision variables (with d s,t (9) captures the concept of decision-dependent uncertainty, where the likelihood of a particular scenario unfolding is affected by the choices made during the planning process. t (s) is a simplified form of the probability of operating the decision variable and is determined by a specific combination of
[0106] π s,t =π(x,y) (10)
[0107] Among them, x represents the probability of component normal operation, y represents the probability of component failure, and π s,t The scenario probability can be expressed in a form similar to that of formula (9). For example, if two of the N devices fail, that is, the "N-2" event, formula (10) can be rewritten as:
[0108]
[0109] Among them, π s,t represents the scene probability, I i,t ,I j,t They represent the currents flowing through the two faulty devices, k represents the device number of the normal operating device set, s represents the scenario, and I k,t Indicates the current flowing through the device during normal operation.
[0110] Furthermore, considering the reliability assessment of random unit commitment, the set of all possible outcomes of uncertain parameters can be approximated by a discrete set of scenarios S and their corresponding probabilities s, given the known underlying distribution of the uncertainty sources. In this framework, D ts represents the electricity demand achieved in each scenario s∈S during time period t. The availability of generation capacity includes the uncertainty of power outages and intermittent sources, further denoted by X its This symbol indicates that the capacity of each generating unit depends on the specific time period and the scenario considered.
[0111] Furthermore, the reliability assessment problem considering random unit combinations is considered as a two-stage planning problem, where the combination variable u it must be determined in the first stage. After the known uncertain parameters are realized, the scheduling variables x are then carried out in the second stage its .
[0112] Furthermore, the two-stage random unit combination problem incorporates uncertainty through a scenario-based approach. ts The power generation level can be determined by x its In contrast, the combined decision (with u it ) must be made before this information is available and is therefore classified as a first-stage decision. The objective function f g It can be expressed as the first stage cost function Second stage cost function and demand for cost reduction where c cur is the cost coefficient of demand reduction, is the total demand reduction, and the objective function of the entire two-stage planning problem is as follows:
[0113]
[0114] in, represents the cost function of the first stage of the unit commitment model, represents the cost function of the second stage of the unit commitment model, u g,t-1 represents the start-stop decision of the unit at time t-1, u g,t represents the start and stop decision of the unit at time t, represents the unit output decision at time t, Indicates the need to cut costs.
[0115] Furthermore, the first stage cost function It can be expressed as the startup cost of the generator set and downtime costs The formula for calculating the sum is as follows:
[0116]
[0117] in, represents the cost function of the first stage of the unit commitment model, u g,t-1 represents the start-stop decision of the unit at time t-1, u g,t represents the start and stop decision of the unit at time t, represents the startup cost of the generator set, Represents the downtime cost of the generator set.
[0118] Furthermore, the second-stage cost function is the operating cost of the generator set, which is calculated as follows:
[0119]
[0120] in, represents the cost function of the second stage, represents the scheduling variable, c g Represents the operating cost coefficient of the g-th unit.
[0121] Furthermore, the first stage incorporates constraints on the unit commitment decision. Equations (15) and (16) ensure that each generator unit g meets the minimum downtime and minimum uptime requirements. Constraints (17) and (18) address the costs associated with unit startup and shutdown. g and J g They represent the startup and shutdown cost coefficients of unit g respectively.
[0122] The constraints of the first stage of the unit commitment problem are as follows:
[0123]
[0124] Among them, TO represents the shortest downtime, u g,t-1 represents the combined decision at time t-1, u g,t represents the combined decision at time t, u g,k represents the combined decision at time k, TS represents the shortest running time, represents the startup cost of the generator set, represents the downtime cost of the generator set, H g and J g They represent the startup and shutdown cost coefficients of unit g respectively, and t represents time.
[0125] Furthermore, the second stage contains various constraints, expressed by equations (19) to (28), considering the relationship with the generator (represented by the matrix C g ) and lines (represented by the matrix C ft (19) enforces power balancing on each bus. and represents the demand of each bus in period t and the corresponding reduction. Equation (20) limits the power flow through the line, where represents the voltage phase angle difference between the buses connected to line l at time t. Constraint (21) ensures that the line power flow remains within the limit, where represents the availability of line l in scenario s, Ω lrepresents the set of all lines. Furthermore, constraint (22) ensures that the generator output remains within the limits, where represents the availability of unit g in scenario s, Ω g represents the set of all units. Constraints (23) and (24) address the ramp capability, limiting the upward and downward adjustments of power output. g,u and R g,d Indicates the rising and falling limits respectively. In addition, ST g,u and SD g,d denote the limits for starting and shutting down power, respectively. Constraint (25) limits the extent of demand reduction, while constraint (26) sets the boundary of the phase angle. The set of availability states of all elements is denoted by Z s Indicates that each element and is the set Z s Finally, Equations (27) and (28) introduce reliability-specific constraints.
[0126] Specifically, the constraints of the second stage of the unit commitment problem are as follows:
[0127]
[0128]
[0129] Among them, C g represents the node unit-node connection matrix, represents the node generator output, C ft represents the node-node connectivity matrix, represents the transmission line transmission power, Indicates load, Indicates the amount of cutting load, Indicates the power angle difference, X l Indicates line reactance, P l min Indicates the minimum active power of the line. represents the impedance of the transmission line l, P l max Indicates the minimum active power of the line. Indicates the minimum output of the generator, Indicates the available status of the line, u g,t represents the unit start-stop decision at time t, represents the maximum output of the generator, represents the unit output decision at time t-1, u g,t-1 represents the unit start-stop decision at time t-1, R g,u Indicates the unit's upward climbing rate, ST g,u Indicates the rate at which the unit starts. represents the unit output decision at time t-1, R g,d Indicates the downward climbing rate of the unit, SD g,d Indicates the rate at which the unit is shut down. represents the demand reduction at time t, represents the maximum demand reduction at time t, θ min Indicates the minimum value of the voltage phase angle, represents the voltage phase angle at time t, θ max Indicates the maximum value of the voltage phase angle, EENS t represents the power loss in period t, π s represents the scene probability, Indicates the average value of power loss.
[0130] Furthermore, reliability indicators used to quantify operational reliability include expected energy not supplied (EENS) and loss of load probability (LOLP). EENS refers to the amount of electricity demand that is expected to be unsatisfied within a specific time frame. t The calculation formula is as follows:
[0131]
[0132] Among them, EENS t represents the expected power shortage at time t, Δt represents the time range considered, and π s represents the probability of the scenario occurring, Indicates the load reduction corresponding to scenario s.
[0133] Furthermore, LOLP measures the probability that the power system will not be able to meet power demand within a year. The calculation formula for LOLP is as follows:
[0134]
[0135] Among them, L s,t is a binary variable. If the demand reduction occurs in scenario s, then L s,t is equal to 1, otherwise L s,t is equal to 0.
[0136] Furthermore, L s,t The expression is as follows:
[0137]
[0138] Among them, L s,t represents the demand reduction indicator variable, Indicates the amount of load being cut.
[0139] Furthermore, operational reliability assessments often use scenario generation techniques based on Nk contingencies, where N represents the total number of system components and k represents the number of allowed simultaneous failures. Increasing the value of k allows for a more comprehensive assessment by incorporating the impact of multiple component outages. However, this approach faces a trade-off between accuracy and computational burden. As k grows, the number of potential contingencies grows exponentially, significantly increasing the computational complexity required for the analysis.
[0140] S3: According to the power system intrinsic uncertainty model and the power system reliability assessment model considering intrinsic uncertainty, a reliability assessment algorithm based on the M-BPNN model and data-driven operation is designed, and the power system operation reliability is evaluated.
[0141] Preferably, due to the huge size and interconnectedness of modern power grids, assessing the operational reliability of power systems is a complex task. For large systems, traditional methods such as Nk contingency analysis become computationally expensive because they require simulating a large number of outage scenarios. Back propagation neural network (BPNN) provides a promising alternative by reducing the computational burden associated with reliability.
[0142] Furthermore, BPNN provides a solution that can continuously update data, thereby being able to adapt to changes and maintain the accuracy of assessing operational reliability over time. BPNN is organized into interconnected artificial neuron layers. These layers include: input layer, hidden layer and output layer, as shown in the attached manual. Figure 3 shown.
[0143] Furthermore, the input layer receives raw data representing the state of the power system for reliability assessment, examples of such data include power generation, power demand at various grid points, and operating status of components.
[0144] Further, the hidden layers perform the core information processing. A BPNN can have one or more hidden layers, each containing multiple interconnected neurons that apply mathematical functions to transform the received data. The number of hidden layers and neurons significantly affects the ability of a BPNN to learn complex relationships in the data.
[0145] Furthermore, the output layer produces the predictions or outputs of the network, which in the context of reliability assessment can be a single value representing the probability of system failure or a set of values representing the reliability of different grid components.
[0146] Furthermore, information flows through connections between neurons in adjacent layers, rather than directly between neurons in the same layer. Each connection has an associated weight that determines the strength of the transmitted signal. These weights are crucial to the learning process of BPNN.
[0147] Furthermore, during the forward propagation, the activation of the hidden neuron is calculated as follows:
[0148] y j =f J (μ j +ω j y m ) (32)
[0149] Among them, y j Indicates that the jth hidden layer receives input from the previous layer, f J (·) represents the activation function of each neuron in the jth hidden layer, ω j represents the weights connecting the previous layer to the neurons in the jth hidden layer, μ j represents the threshold of each neuron in the jth hidden layer.
[0150] Furthermore, the relationship between the current hidden layer and the next hidden layer can be expressed as:
[0151] y j+1 =f J (μ j+1 +ω j+1 y j ) (33)
[0152] Among them, y j+1 Indicates that the j+1th hidden layer receives input from the previous layer, f J (·) represents the activation function of each neuron in the jth hidden layer, ω j represents the weights connecting the previous layer to the neurons in the jth hidden layer, μ j represents the threshold of each neuron in the jth hidden layer.
[0153] Furthermore, the relationship between the hidden layer and the output layer can be expressed as:
[0154] y t =f o (u o +ω o y j,max ) (34)
[0155] Among them, y t Indicates that the model predicts the value at each time point t, f o (·) represents the activation function, u orepresents the threshold applied to the output layer, ω o represents the weight connecting the final hidden layer and the output layer, y j,max Represents the maximum predicted output of the jth hidden layer. Usually, the hidden layer activation function is a logistic function or a hyperbolic function, while the output layer usually uses a linear activation function.
[0156] Furthermore, in order to evaluate the accuracy of the model, the error between the predicted output and the actual output is calculated using the following formula:
[0157] δ y =y t -y (35)
[0158] Among them, δ y Represents the error between the predicted output and the actual output, y t It indicates that the model predicts the value at each time point t, and y indicates the actual output value.
[0159] Furthermore, the formula for calculating the unit error term in the jth hidden layer is as follows:
[0160]
[0161] Among them, δ j represents the unit error term in the jth hidden layer, represents the transpose of the weights of the previous layer connected to the neurons in the jth hidden layer, δ j+1 represents the unit error term in the j+1th hidden layer, represents the updated activation function of each neuron in the jth hidden layer, f J represents the activation function of each neuron in the jth hidden layer.
[0162] Furthermore, the weight coefficient of hidden layer j can be updated as:
[0163]
[0164] Among them, ω j represents the weight coefficient of hidden layer j, δ j+1 represents the unit error term in the j+1th hidden layer, represents the transpose of the j-th hidden layer output.
[0165] The above partial equations (32)-(37) describe one iteration of training a multiple back propagation neural network (M-BPNN). By repeatedly adjusting its learning parameters over multiple iterations, the M-BPNN gradually improves its accuracy.
[0166] Furthermore, a novel approach, namely the multiple back-propagation neural network (M-BPNN) architecture, is introduced to improve the prediction accuracy of system reliability indicators under fluctuating operating conditions. The traditional BPNN model directly maps operating conditions to reliability indicators. However, the impact of different operating points on reliability varies with the changing conditions. This limitation may lead to inaccurate predictions from a single BPNN architecture.
[0167] Furthermore, M-BPNN addresses this limitation by introducing a classification step before network training. System operation data is first classified according to selected attributes, such as mean or quantile values. This allows the creation of multiple BPNN sub-models, each trained on data with similar characteristics. During the prediction process, the input data in the test set is first classified and then directed to the corresponding sub-model for prediction. Finally, the results of all sub-models are combined to form a complete prediction.
[0168] Furthermore, the effectiveness of BPNN is greatly affected by the number of hidden layers and neurons in these layers. To address this problem, an adaptive algorithm is proposed to find the optimal architecture of M-BPNN. The algorithm iteratively evaluates different network configurations and selects the one that provides the most accurate prediction results.
[0169] The adaptive algorithm consists of the following steps:
[0170] Feed the training and testing datasets into the system;
[0171] Set the maximum value, the network architecture has a predetermined maximum number of hidden layers and units per layer;
[0172] Enumeration, the model uses a recurrent procedure to iteratively evaluate various configurations of hidden layers and the number of elements within those layers while staying within predefined limits;
[0173] BPNN training, iteratively trains the back-propagation neural network using the selected hidden layer and element configuration;
[0174] After training the network, the reliability index and error are calculated using the BP neural network;
[0175] The loop iterates until all possible combinations of hidden layers and hidden elements have been enumerated;
[0176] Minimum error selection, after evaluating all unit combinations, the adaptive algorithm will find the combination that leads to the minimum error;
[0177] Reliability Index Calculation,The reliability index is calculated using the combination of the identified hidden layers and hidden elements.
[0178] Furthermore, this embodiment also provides a power system reliability assessment system taking into account endogenous uncertainty, comprising:
[0179] Definition module, used to define endogenous uncertainty and build an endogenous uncertainty model for power systems;
[0180] A model building module is used to build a power system reliability assessment model taking into account inherent uncertainty based on the power system inherent uncertainty model;
[0181] The evaluation module is used to design a reliability evaluation algorithm based on the M-BPNN model and data-driven operation according to the power system intrinsic uncertainty model and the power system reliability evaluation model considering intrinsic uncertainty, and to evaluate the power system operation reliability.
[0182] This embodiment also provides a computer device, which is suitable for the case of a power system reliability assessment method taking into account inherent uncertainties, including a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute computer-executable instructions to implement the power system reliability assessment method taking into account inherent uncertainties as proposed in the above embodiment.
[0183] The computer device may be a terminal, and the computer device includes a processor, a memory, a communication interface, a display screen and an input device connected via a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device may be a touch layer covering the display screen, or a key, trackball or touchpad provided on the housing of the computer device, or an external keyboard, touchpad or mouse, etc.
[0184] This embodiment also provides a storage medium having a computer program stored thereon, which, when executed by a processor, implements the power system reliability assessment method taking into account the intrinsic uncertainty as proposed in the above embodiment.
[0185] In summary, the traditional power system reliability assessment model only considers the impact of exogenous uncertainty and cannot characterize the impact mechanism of power system operation decisions on the system reliability level. The present invention incorporates endogenous uncertainty into the power system operation reliability assessment process, reveals the mechanism of operation decisions on the system reliability level; establishes a reliability-constrained unit combination model to assess the impact of endogenous uncertainty on operation reliability; proposes an M-BPNN-based model and a data-driven operation reliability assessment algorithm, which can efficiently solve the power system operation reliability assessment optimization model.
[0186] Example 2
[0187] Reference Figure 4 , which is the second embodiment of the present invention, and this embodiment provides a method for evaluating the reliability of a power system taking into account endogenous uncertainties. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.
[0188] Specifically, a case study is conducted through numerical experiments to verify the advantages of this method in improving the efficiency of power system operation reliability assessment. The present invention considers five power system reliability assessment methods. Method 1 (M1): non-sequential Monte Carlo simulation method. Method 2 (M2): method proposed by the present invention. Method 3 (M3): method based on radial basis function (RBF) neural network. Method 4 (M4): method based on generalized regression neural network. Method 5 (M5): genetic algorithm.
[0189] Preferably, three test systems are considered, including the RTS·79 system, the RBTS system and the RTS·96 system.
[0190] Further, the evaluation process first uses M1 to calculate the system reliability index under various operating conditions. These calculated indices are then used as input and output data for M2-M4. Next, the data is divided into a training set and a test set in a ratio of 9:1. The training set is used to train the neural network model, and the test set is used to verify the accuracy of the model.
[0191] Furthermore, the unit combination results of the RBTS system are taken as an example. The RBTS system consists of 6 busbars, 9 transmission lines, and generators including 4 thermal generators and 7 hydroelectric generators. It meets the needs of 5 demand centers with a total installed capacity of 240 MW, exceeding the peak demand of 185 MW. The daily demand curve of IEEE-RBTS provides information for the demand model. For ease of management, the maximum number of equipment that fails simultaneously is limited to 3. Table 1 lists the capacity and reliability parameters of the system components in detail. A total of 1351 scenarios were considered in the analysis, covering all possible "N-3" contingencies. For reference, Table 2 provides a complete list of "N-1", some "N-2" and "N-3" scenarios.
[0192] Table 1 Reliability parameters of RBTS system electrical equipment
[0193] Electrical Equipment Capacity(MW) Failure rate (occ. / a) Repair rate (occ. / a) Probability of forced outage Generator 1 40 2.0 194.67 0.0300 Generator 2 40 2.0 194.67 0.0300 Generator 3 10 4 194.67 0.0200 Generator 4 20 2.4 159.27 0.0250 Generator 5 5 2.4 159.27 0.0100 Generator 6 5 2.4 159.27 0.0100 Generator 7 40 2.4 159.27 0.0200 Generator 8 20 5.0 194.67 0.0150 Generator 9 20 3.0 146.00 0.0150 Generator 10 20 6.0 194.67 0.0150 Generator 11 20 6.0 194.67 0.0150 Transmission Line 1 45 1.5 876.00 0.0017 Transmission Line 2 40 5.0 876.00 0.0057 Transmission Line 3 40 4.0 876.00 0.0045 Transmission Line 4 71 1.0 876.00 0.0011 Transmission Line 5 71 1.0 876.00 0.0011 Transmission Line 6 45 1.5 876.00 0.0017 Transmission Line 7 42 5.0 876.00 0.0057 Transmission Line 8 71 1.0 876.00 0.0011 Transmission Line 9 71 1.0 876.00 0.0011
[0194] Table 2 System event probability table considering endogenous uncertainty
[0195]
[0196] Furthermore, Table 3 compares the performance of each method M1-M5 in three test systems (including RTS 79 system, RBTS system, and RTS 96 system). M1 provides accurate results but is time-consuming. Methods M2-M5 do not provide optimal results but are faster than M1. By comparing M2 with M3-M5 in each test system, we can evaluate the trade-off between accuracy and computational efficiency.
[0197] Further, the table header shows how the results of each method (EENS and LOLP) compare to the standard method M1 in the same test system. Take the results of the RBTS system as an example. M1 gives an EENS baseline value of 1,056MWh / a. The EENS value of M2 (1,021MWh / a) is lower than that of M1, indicating that it is closer. The EENS value of M3 (1,098MWh / a) is higher than that of M1, with a larger deviation. The EENS value of M4 (998MWh / a) is lower than that of M1, but the magnitude of the difference may be large compared to M2. The EENS value of M5 (1,007MWh / a) is lower than that of M1, but the magnitude of the difference may be greater than that of M2.
[0198] Table 3 Reliability evaluation results of M1-M5
[0199]
[0200] Furthermore, Figure 4The absolute errors of the four approximation methods M2, M3, M4 and M5 were compared with the standard method M1 to evaluate the reliability of the three test systems. The y-axis of the graph represents the absolute error and the x-axis represents the evaluation method. Each method is represented by a different color line in the graph. There are two graphs in the figure, one for each indicator, the absolute error of EENS and the absolute error of LOLP.
[0201] Furthermore, the absolute errors of method M2 in EENS are generally lower compared to M3, M4, and M5 on the RBTS, RTS-79, and RTS-96 systems, indicating that M2 provides a better approximation to the standard method M1 in terms of EENS on these systems.
[0202] Furthermore, on the RBTS, RTS-79, and RTS-96 systems, the LOLP absolute error of method M2 is also lower than that of M3, M4, and M5. This indicates that M2 also provides a better approximation to the standard method M1 in terms of LOLP on these systems.
[0203] Overall, Figure 4 It shows that for EENS and LOLP of RBTS, RTS-79 and RTS-96 systems, the approximation method M2 provides the closest approximation to the standard method M1, while the other methods M3, M4 and M5 perform worse in comparison. This shows that M2 performs closest to the standard method among the methods evaluated.
[0204] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. A method for evaluating the reliability of a power system taking into account endogenous uncertainties, characterized in that: include, Define endogenous uncertainty and construct an endogenous uncertainty model for power systems; Based on the power system endogenous uncertainty model, a power system reliability assessment model considering endogenous uncertainty is constructed; According to the inherent uncertainty model of the power system and the power system reliability assessment model considering the inherent uncertainty, a reliability assessment algorithm based on the M-BPNN model and data-driven operation is designed, and the power system operation reliability is evaluated.
2. The method for evaluating the reliability of a power system taking into account endogenous uncertainties according to claim 1, characterized in that: The construction of the power system endogenous uncertainty model includes: The condition-dependent failure rate of the electrical component is calculated, and the calculation formula of the condition-dependent failure rate is as follows: Among them, F(I) represents the probability of forced shutdown of electrical equipment, I represents the real-time current, I0 represents the rated current of the equipment, I1 represents the tripping setting current, and α and β are parameters. It indicates the initial value of the electrical equipment FOR when the current is less than I0; The calculation formula of FOR is as follows: Among them, F0 represents the probability of forced shutdown of electrical equipment when it leaves the factory, λ represents the failure rate, and μ represents the repair rate of electrical equipment; The formula for calculating parameters α and β based on historical data is as follows: Among them, α and β are parameters, F0 represents the probability of forced shutdown of electrical equipment when it leaves the factory, I0 represents the rated current of the equipment, and F max It represents the maximum forced outage probability within the allowable range, and I1 represents the tripping setting current; The actual value of the current I is obtained by calculating the optimal power flow. The optimal power flow depends on the operation decision of the power system. The FOR of the i-th power-consuming device based on the endogenous uncertainty is denoted as F i EU , F i EU The equivalent formula is as follows: F i EU =F i EU (I i )=F i EU (Q(x,y)) (5) I i =Q(x,y) (6) Among them, F i EU represents the EU-based FOR of the i-th electrical equipment, I1 represents the tripping setting current, I i represents the actual operating current of the ith electrical equipment, x is a binary variable, y represents a continuous variable, and Q(·) represents the state function between the binary variable x and the continuous variable y; In addition to inherent constraints, the FOR of power components is further affected by external uncertainties, including aging-related failure factors. EU effect and aging effect are both incorporated into the reliability model. Assuming the service age of an electrical device is t0, the FOR in the i-th cycle is recorded as The calculation formula is as follows: in, represents the FOR in the i-th cycle, t0 represents the service age, t represents the time, H(t) represents the function of aging-related failures with respect to t, and i represents the number of the service cycle; There is a dependency between the power outages caused by EU issues and the power outages caused by equipment aging. Formula 1 is updated to a general form. The specific formula is as follows: Among them, F(·) represents the function of FOR, α age and β age Respectively represent the F0 update in equation 3 and equation 4 as The obtained parameters, I represents the real-time current, I0 represents the rated current of the equipment, and I1 represents the trip setting current. represents the FOR in the ith cycle, T i Indicates the service life.
3. The method for evaluating the reliability of a power system taking into account endogenous uncertainties according to claim 2, characterized in that: Constructing the power system reliability assessment model considering endogenous uncertainty includes generating typical scenarios of endogenous uncertainty, and the typical scenarios of endogenous uncertainty include: Assuming that in the process of power system operation reliability assessment, the scenario probability is independent of the operation decision, and the solution process precedes the equipment operation decision, the calculation formula of the scenario probability is as follows: Among them, π s represents the probability of the scene occurring, φ s,up represents the set of available devices corresponding to the scene, φ s,dn represents the set of unavailable equipment, F(·) represents the probability function of forced equipment outage, I represents the real-time current, T1 represents the first service cycle, and i represents the number of the power system component; The scenario probability is explicitly modeled as a function of the operational decision variables. Equation 9 represents the decision-related scenario probability P at period t. t (s), the decision-related scenario probability of period t is given by the operational decision variable and A specific combination of p s,t =π(x,y) (10) Among them, x represents the probability of normal operation of power system equipment, y represents the probability of failure of power system equipment, and π s,t represents the probability of the scenario; Assume that two of the N devices fail, and equation 10 can be rewritten as: Among them, π s,t represents the scene probability, I i,t ,I j,t They represent the currents flowing through the two faulty devices, k represents the device number of the normal operating device set, s represents the scenario, and I k,t Indicates the current flowing through the device during normal operation.
4. The method for evaluating the reliability of a power system taking into account endogenous uncertainties according to claim 3, characterized in that: Constructing the power system reliability assessment model considering endogenous uncertainty also includes reliability assessment including random unit combination, and the reliability assessment including random unit combination includes: The set of all results of uncertain parameters is represented by the discrete scenario set S and its corresponding probability π s,t To indicate that D ts represents the electricity demand achieved in each scenario s∈S during time period t, with the availability of generation capacity including the uncertainty of power outages and intermittent sources; The first stage determines the unit commitment decision, and the second stage determines the dispatch variables; The two-stage random unit commitment problem incorporates uncertainty through a scenario-based approach. When the demand is uncertain, after the second-stage load is known, D ts Determining the generation level, in contrast, the combination decision must be made before this information is available and is classified as a first-stage decision, with the objective function f g Expressed as the first stage cost function Second stage cost function and demand for cost reduction c cur is the cost coefficient of demand reduction, is the total demand reduction, and the objective function of the entire two-stage planning problem is as follows: in, represents the cost function of the first stage of the unit commitment model, represents the cost function of the second stage of the unit commitment model, u g,t-1 represents the unit commitment decision at time t-1, u g,t represents the unit commitment decision at time t, represents the unit output decision at time t, Indicates the need to cut costs; First stage cost function The startup cost of the generator set Downtime costs The formula for calculating the sum is as follows: in, represents the cost function of the first stage, u g,t-1 represents the unit commitment decision at time t-1, u g,t represents the unit commitment decision at time t, represents the startup cost of the generator set, represents the downtime cost of the generator set; The second stage cost function is the operating cost of the generator set. The calculation formula of the second stage cost function is as follows: in, represents the cost function of the second stage, represents the scheduling variable, c g represents the operating cost coefficient of the g-th unit; The constraints of the unit commitment model are as follows: Among them, TO represents the shortest downtime, u g,t-1 represents the combined decision at time t-1, u g,t represents the combined decision at time t, u g,k represents the combined decision at time k, TS represents the shortest running time, represents the startup cost of the generator set, represents the downtime cost of the generator set, H g and J g They represent the startup and shutdown cost coefficients of unit g respectively.
5. The method for evaluating the reliability of a power system taking into account endogenous uncertainties according to claim 4, characterized in that: The second stage constraints include enforcing power balance on each bus, controlling the power flow through the line, ensuring that the line power flow remains within limits, guaranteeing that generator output remains within limits, limiting upward and downward adjustments in power output, limiting the extent of demand reduction, setting bounds on phase angles, and introducing reliability-specific constraints.
6. The method for evaluating the reliability of a power system taking into account endogenous uncertainties according to claim 5, characterized in that: The reliability assessment algorithm designed based on M-BPNN model and data-driven operation includes: Organizing the BPNN into interconnected artificial neuron layers, the artificial neuron layers comprising an input layer, a hidden layer, and an output layer; The input layer receives raw data representing the state of the power system and performs reliability assessment, wherein the raw data includes power generation, power demand of each grid point and operating status of components; The hidden layer performs core information processing; The output layer outputs a single value representing the probability of system failure or outputs a set of values representing the reliability of different power grid components; The reliability assessment algorithm designed based on the M-BPNN model and data-driven operation also includes: During the forward propagation, the activation of the hidden neuron is calculated as follows: y j =f J (m j +oh j y m ) Among them, y j Indicates that the jth hidden layer receives input from the previous layer, f J (·) represents the activation function of each neuron in the jth hidden layer, ω j represents the weights connecting the previous layer to the neurons in the jth hidden layer, μ j represents the threshold of each neuron in the jth hidden layer; The expression of the relationship between the current hidden layer and the next hidden layer is as follows: y j+1 =f J (m j+1 +oh j+1 y j ) Among them, y j+1 represents the output of the j+1th hidden layer, f J (·) represents the activation function of each neuron in the jth hidden layer, ω j represents the weights connecting the previous layer to the neurons in the jth hidden layer, μ j represents the threshold of each neuron in the jth hidden layer; The expression of the relationship between the hidden layer and the output layer is as follows: y t =f o (u o +ω o y j,max ) Among them, y t Indicates that the model predicts the value at each time point t, f o (·) represents the activation function, u o represents the threshold applied to the output layer, ω o represents the weight connecting the final hidden layer and the output layer, y j,max represents the maximum predicted output of the jth hidden layer; The error between the predicted output and the actual output is calculated as follows: δ y =and t -and Among them, δ y Represents the error between the predicted output and the actual output, y t It indicates that the model predicts the value at each time point t, and y indicates the actual output value; The formula for calculating the unit error term in the jth hidden layer is as follows: Among them, δ j represents the unit error in the jth hidden layer, represents the weight of the jth hidden layer, δ j+1 represents the unit error in the j+1th hidden layer, represents the normalized activation function of each neuron in the Jth hidden layer, f J represents the activation function of each neuron in the Jth hidden layer; The weight coefficient of hidden layer j is updated as: Among them, ω j represents the weight coefficient of hidden layer j, δ j+1 represents the unit error in the j+1th hidden layer, Represents the predicted value of the hidden layer j of the model.
7. The method for evaluating the reliability of a power system taking into account endogenous uncertainties according to claim 6, characterized in that: The steps of using the adaptive algorithm to find the optimal architecture of the M-BPNN model include: Feed the training and testing datasets into the system; Set the maximum number of hidden layers and units per layer; Utilize a looping process to iteratively evaluate various configurations and numbers of elements within a layer while staying within predefined limits; Iteratively train the back-propagation neural network using the selected hidden layer and element configuration; Use BP neural network to calculate reliability index and error; The loop iterates until all possible combinations of hidden layers and hidden elements have been enumerated; After evaluating all unit combinations, find the combination that results in the minimum error; The reliability index is calculated using the combination of the identified hidden layers and hidden elements.
8. A power system reliability assessment system taking into account endogenous uncertainty, based on the power system reliability assessment method taking into account endogenous uncertainty according to any one of claims 1 to 7, characterized in that: Also includes, Definition module, used to define endogenous uncertainty and build an endogenous uncertainty model for power systems; A model building module is used to build a power system reliability assessment model taking into account inherent uncertainty based on the power system inherent uncertainty model; The algorithm design module is used to design a reliability assessment algorithm based on the M-BPNN model and data-driven operation according to the power system intrinsic uncertainty model and the power system reliability assessment model considering intrinsic uncertainty, and to evaluate the power system operation reliability.
9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the method for evaluating the reliability of a power system taking into account the intrinsic uncertainty described in any one of claims 1 to 7 are implemented.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for evaluating the reliability of a power system taking into account the intrinsic uncertainty described in any one of claims 1 to 7 are implemented.
