Power distribution network risk diagnosis method, device and equipment and readable storage medium

Abstract

The application discloses a power distribution network risk diagnosis method, device and equipment and a readable storage medium. The method can acquire a simulation model of a power distribution network, parameter prediction models of different subjects and current period time sequence state parameters; predict different subject prediction state parameters; generate predicted time sequence parameters of a next period of a subject based on the time sequence state parameters of the subject by using the parameter prediction model of the subject, evaluate whether the subject belongs to a state fluctuation subject, and determine a parameter fluctuation range; generate a plurality of state feasible sets based on the parameter fluctuation range of different state fluctuation subjects; obtain simulation results of each state feasible set; and determine a risk value of the power distribution network based on the simulation results of each state feasible set. It can be seen that the application can effectively avoid the fluctuation source ambiguity problem of traditional shallow data by multi-prediction model cooperation, reduce the missed judgment rate and the misjudgment rate, cover a multi-fluctuation subject combination scene, and further improve the accuracy of risk evaluation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power distribution networks, and more particularly to a power distribution network risk diagnosis method, device, equipment and readable storage medium. BACKGROUND

[0002] With the continuous development and complication of power systems, the safe and stable operation of power distribution networks, as a key link connecting power transmission networks and end users, is of great significance to the protection of social and economic development and people's living needs. However, the power distribution network has a complex network structure and numerous devices, and its operation state is affected by many factors, such as load fluctuation, device aging, natural disasters, etc. These factors can cause faults or risks in the power distribution network, thereby affecting the reliability and stability of power supply.

[0003] Traditional network risk assessment mostly uses shallow data-driven methods. Although these methods can rely on historical load fluctuation curves for statistical prediction and fitting of fault occurrence probability, they can only make simple inferences based on a single historical feature, resulting in low prediction accuracy. Ultimately, the risk prediction accuracy is not high, and it is difficult to identify hidden faults in the power distribution network in advance. SUMMARY

[0004] Therefore, the present application provides a power distribution network risk diagnosis method, device, equipment and readable storage medium to solve the problem of low risk prediction accuracy in existing power distribution network risk prediction technology.

[0005] In order to achieve the above purpose, the present scheme is as follows:

[0006] A power distribution network risk diagnosis method comprises:

[0007] obtaining a simulation model of a power distribution network, parameter prediction models of different subjects in the power distribution network, and time sequence state parameters of different subjects in the power distribution network in a current period;

[0008] using the simulation model to predict the predicted state parameters of different subjects at future time points after a target period based on the state parameters at the current time point in the current period;

[0009] For each subject, using the parameter prediction model of the subject to generate the predicted time sequence parameters of the subject in the next period based on the time sequence state parameters of the subject; based on the predicted state parameters of the subject and the predicted time sequence parameters, determining whether the subject belongs to a state fluctuation subject; and when the subject is a state fluctuation subject, determining the parameter fluctuation range based on the predicted state parameters and the predicted time sequence parameters;

[0010] Generate a plurality of state feasible sets based on the parameter fluctuation range of the state fluctuation subject in different states, each state feasible set containing feasible state values of each state fluctuation subject;

[0011] Import each state feasible set into the simulation model in turn to obtain a simulation result of each state feasible set;

[0012] Determine the risk value of the power distribution network based on the simulation result of each state feasible set.

[0013] Optionally, the evaluation of whether the subject belongs to a state fluctuation subject based on the predicted state parameter and the predicted timing parameter includes:

[0014] Obtain a state parameter comparison model;

[0015] Use the state parameter comparison model to evaluate whether the subject belongs to a state fluctuation subject based on the predicted state parameter and the predicted timing parameter.

[0016] Optionally, the state parameter comparison model includes:

[0017] Construct a deep learning model;

[0018] Obtain a plurality of training samples of different training subjects, each training sample containing a training state parameter and a training timing parameter, the training state parameter and the training timing parameter containing parameter values of a plurality of training parameter items;

[0019] Based on the training state parameter and the training timing parameter of each training sample, calculate the absolute difference of different training parameter items in the same training sample, and integrate each absolute difference of the same training sample to calculate the training fluctuation value of the corresponding training sample;

[0020] Based on each training sample and its corresponding training fluctuation value, train the deep learning model until the deep learning model meets the preset stopping condition, and the final deep learning model is the state parameter comparison model.

[0021] Optionally, the predicted state parameter and the predicted timing parameter contain parameter values of a plurality of state parameter items;

[0022] The determination of the parameter fluctuation range based on the predicted state parameter and the predicted timing parameter includes:

[0023] Compare the parameter values of the predicted state parameter and the predicted timing parameter corresponding to the same state parameter item to determine the parameter value range of the corresponding state parameter item;

[0024] By combining the parameter value ranges of each state parameter item, the parameter fluctuation range of the main state fluctuation entity is obtained.

[0025] Optionally, the generation of multiple feasible state sets based on the parameter fluctuation ranges of different state fluctuation subjects includes:

[0026] Gradient values ​​are obtained for the parameter fluctuation range of different state fluctuating entities to obtain multiple feasible state values ​​for different state fluctuating entities;

[0027] By randomly combining the feasible state values ​​of different state fluctuation subjects, multiple feasible state sets are obtained.

[0028] Optionally, determining the risk value of the distribution network based on the simulation results of each feasible set of states includes:

[0029] Based on the simulation results for each feasible state set, the elasticity risk index for the corresponding feasible state set is calculated.

[0030] The risk value of the distribution network is determined based on the elastic risk index of each feasible set of states.

[0031] Optionally, the calculation of the elasticity risk index for the corresponding feasible state set based on the simulation results for each feasible state set includes:

[0032] Based on the simulation results for each feasible state set, calculate the N-1 / N-2 pass rate, transient stability margin, frequency regulation margin, and reactive power compensation margin for the corresponding feasible state set.

[0033] A power distribution network risk diagnosis device, comprising:

[0034] The acquisition module is used to acquire the simulation model of the distribution network, the parameter prediction model of different entities in the distribution network, and the time-series state parameters of different entities in the distribution network in the current period.

[0035] The prediction module is used to use the simulation model to predict the predicted state parameters of different subjects at future times after the target period, based on the state parameters at the current time in the current period.

[0036] The evaluation module is used to, for each subject, utilize the subject's parameter prediction model to generate predicted time-series parameters for the next period based on the subject's time-series state parameters; evaluate whether the subject belongs to a state-fluctuating subject based on the predicted state parameters and the predicted time-series parameters; and determine the parameter fluctuation range based on the predicted state parameters and the predicted time-series parameters when the subject is a state-fluctuating subject.

[0037] The generating module is configured to generate a plurality of state feasible sets based on the parameter fluctuation range of the different state fluctuation subjects, and each state feasible set contains feasible state values of each state fluctuation subject;

[0038] The simulation module is configured to sequentially import each state feasible set into the simulation model to obtain a simulation result of each state feasible set.

[0039] The determining module is configured to determine the risk value of the power distribution network based on the simulation result of each state feasible set.

[0040] A power distribution network risk diagnosis device includes a memory and a processor.

[0041] The memory is configured to store a program.

[0042] The processor is configured to execute the program to implement each step of the power distribution network risk diagnosis method.

[0043] A readable storage medium has a computer program stored thereon, and the computer program, when executed by a processor, implements each step of the power distribution network risk diagnosis method.

[0044] It can be seen from the technical solutions that the power distribution network risk diagnosis method provided by the application can obtain a simulation model of a power distribution network, parameter prediction models of different subjects in the power distribution network, and time sequence state parameters of the different subjects in the power distribution network in a current period; the simulation model is used to predict predicted state parameters of the different subjects at future time points after a target period based on state parameters at a current time point in the current period; for each subject, the parameter prediction model of the subject is used to generate predicted time sequence parameters of the subject in a next period based on the time sequence state parameters of the subject; whether the subject belongs to a state fluctuation subject is evaluated based on the predicted state parameters and the predicted time sequence parameters of the subject; when the subject is a state fluctuation subject, a parameter fluctuation range is determined based on the predicted state parameters and the predicted time sequence parameters; based on this, the predicted state parameters reflect predicted parameters obtained based on a physical law of a power grid topology and physical characteristics of equipment, and the predicted time sequence parameters are data-driven predictions based on historical data trends. The two prediction methods respectively depict parameter changes from physical mechanisms and data laws, and differences between the prediction results of the two methods can expose potential uncertainties of parameter changes, thereby accurately locating state fluctuation subjects; prediction deviations caused by interference factors that are not covered by a single model are avoided, and missed judgments and misjudgments caused by traditional fuzzy judgments of a whole system are avoided, thereby reducing the accuracy of risk assessment; on this basis, the application qualitatively describes fluctuation changes of state fluctuation subjects as quantitative ranges through the parameter fluctuation range; subsequently, the application can generate a plurality of state feasible sets based on the parameter fluctuation ranges of different state fluctuation subjects, each state feasible set containing feasible state values of each state fluctuation subject; each state feasible set is sequentially imported into the simulation model to obtain simulation results of each state feasible set; a risk value of the power distribution network is determined based on the simulation results of the state feasible sets; based on this, the application performs multi-scenario simulation coverage on future operation of the power distribution network based on the quantitative ranges of the state fluctuation subjects, ensures that subsequent simulation can comprehensively verify power grid risks under different fluctuation combinations, and avoids risk omissions caused by traditional single-scenario verification. It can be seen that the application can effectively avoid the fluctuation source fuzzy problem of traditional shallow data through multi-prediction model cooperation, reduce the missed judgment rate and the misjudgment rate, cover multi-fluctuation subject combination scenarios, avoid the limitations of traditional single-scenario verification, ensure risk value determination without risk dead angles, and further improve the accuracy of risk assessment. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or the prior art description. Obviously, the drawings in the following description are only embodiments of the application, and those skilled in the art can obtain other drawings according to the provided drawings without any creative effort.

[0046] Figure 1 A flow chart of a power distribution network risk diagnosis method disclosed by an embodiment of the present application;

[0047] Figure 2 A structure block diagram of a power distribution network risk diagnosis device disclosed by an embodiment of the present application;

[0048] Figure 3 A hardware structure block diagram of a power distribution network risk diagnosis device disclosed by an embodiment of the present application. DETAILED DESCRIPTION

[0049] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0050] The embodiment of the present application provides a power distribution network risk diagnosis method, which can be applied to various power distribution network risk assessment systems or power distribution network monitoring systems, and can also be applied to various computer terminals or intelligent terminals. The execution subject can be a processor or a server of a computer terminal or an intelligent terminal.

[0051] Next, the power distribution network risk diagnosis method of the present application will be described in detail, including the following steps: Figure 1 The power distribution network risk diagnosis method of the present application will be described in detail, including the following steps:

[0052] Step S1, obtaining a simulation model of a power distribution network, parameter prediction models of different subjects in the power distribution network, and time sequence state parameters of different subjects in the power distribution network in a current period.

[0053] Specifically, the topological data and device parameters of the power distribution network can be collected, the device parameters and the topological data are mapped by using the IEC 61970 / 61968 standard, and a simulation model is constructed.

[0054] The topological data can include line parameters such as line length, line cross section, and line impedance, connection relationships of each bus, transformer, switch number, and connection table, tower longitude and latitude, device longitude and latitude, etc.

[0055] The topological data can be obtained through AD drawings or a GIS system.

[0056] The device parameters can include transformer nameplate parameters such as rated capacity, short-circuit impedance, and transformation ratio, cable model, overhead line model, cable conductor material, cable insulation grade, switch device rated current, breaking capacity, protection device action threshold, and measurement device accuracy.

[0057] The power distribution network is abstracted as a directed graph comprising a node set and a branch set; the node set comprises bus nodes, load nodes, and power source nodes; and the branch set comprises lines, transformers, and switches.

[0058] A node admittance matrix and a branch parameter matrix are constructed for power flow calculation and state estimation, and multi-scale modeling is performed, including a macroscopic layer and a microscopic layer.

[0059] The macroscopic layer takes a feeder or a substation as a unit, establishes a network connection relationship such as a radial type or a ring network structure, and labels the location and standby capacity of a tie line; and the microscopic layer can be refined to specific equipment, such as modeling a transformer as an excitation branch and a π-type equivalent circuit of winding leakage reactance, and modeling a cable as a distributed parameter line.

[0060] A transformer can adopt a flux-current equation to describe the excitation characteristics, and a winding temperature rise model is considered; an overhead line can consider the skin effect and adopt a frequency-dependent model; a cable can adopt a π-type equivalent circuit containing a ground capacitance, which affects the reactive power distribution; and a load can adopt a dynamic load model.

[0061] Based on the historical data of equipment failures, a remaining life prediction model is established, including the following contents:

[0062] Cable insulation aging can be based on the theory of electric tree growth to establish a Weibull distribution model of insulation resistance and operating time;

[0063] Switch contact wear can be calculated according to the number of on-off times, and a linear degradation model is used to calculate the growth rate of contact resistance.

[0064] The time characteristics of the overcurrent protection action of a protection device, the automatic reclosing timing, a photovoltaic inverter control model for a distributed power source, and the charge-discharge logic of a storage battery can be adopted.

[0065] The time sequence state parameter can comprise parameter values of different state parameter items at different time points in a current period.

[0066] The current period comprises a current time.

[0067] The parameter values of different state parameter items at the current time in the time sequence state parameter can be used as state parameters.

[0068] Each subject can be all equipment and all lines in the power distribution network.

[0069] The state parameter items of different types of subjects can be different, for example, the state parameter items of a transformer can be oil temperature, winding DC resistance, and partial discharge signal; and the state parameter items of a line can include current, voltage, partial discharge quantity, and insulation resistance.

[0070] Therefore, in order to improve the reliability of parameter prediction, different subjects can correspond to different parameter prediction models.

[0071] The parameter prediction model acquisition process of each subject can be as follows:

[0072] An LSTM model is built, and a training set, a validation set, and a test set are obtained, the training set, the validation set, and the test set containing historical state time series parameters of different cycle numbers and continuous historical cycles;

[0073] The LSTM model is trained using each training set, and the trained LSTM model is verified and tested using the validation set and the test set. When the verification and testing meet the stopping condition, the finally obtained LSTM model is the parameter prediction model corresponding to the subject.

[0074] The cycle number ratio of the training set, the validation set, and the test set can be set according to actual needs, such as a division ratio of 70:15:15.

[0075] Model performance is evaluated during the training process to assist in adjusting hyperparameters (such as learning rate and number of hidden layer neurons) and avoid overfitting; test set (15%): after model training is completed, independent data not involved in training is used to evaluate prediction accuracy and ensure generalization ability.

[0076] Step S2, using the simulation model, based on the state parameter of the current time in the current cycle, predicting the predicted state parameters of different subjects at future time after the target period.

[0077] Specifically, the state parameter of the current time in the current cycle can be imported into the simulation model for simulation. After the simulation model continues to simulate for the target period, the simulation is stopped. The state parameters of different subjects of the final simulation model are used as the predicted state parameters of different subjects at future time after the target period.

[0078] The length of the target period can be less than or equal to the cycle length, so that the future time corresponding to the predicted state parameter is located in the next cycle of the current cycle.

[0079] Step S3, for each subject, using the parameter prediction model of the subject, based on the time series state parameters of the subject, generating the predicted time series parameters of the next cycle of the subject; based on the predicted state parameters of the subject and the predicted time series parameters, evaluating whether the subject belongs to a state fluctuation subject; when the subject is a state fluctuation subject, based on the predicted state parameters and the predicted time series parameters, determining the parameter fluctuation range.

[0080] Specifically, the time series state parameters of each subject can be input into the corresponding parameter prediction model to obtain the predicted time series parameters of each subject in the next cycle.

[0081] The predicted state parameter and the predicted timing parameter of the subject are compared, and whether the corresponding subject has an abnormal fluctuation risk is evaluated, and whether it belongs to a state fluctuation subject is determined.

[0082] The parameter fluctuation range of the state fluctuation subject can be determined based on the predicted state parameter and the predicted timing parameter of the same state fluctuation subject.

[0083] Step S4, based on the parameter fluctuation range of different state fluctuation subjects, a plurality of state feasible sets are generated, and each state feasible set contains the feasible state value of each state fluctuation subject.

[0084] Specifically, based on each parameter fluctuation range, each feasible state value of each state fluctuation subject is determined.

[0085] Randomly combine each feasible state value to obtain a state feasible set.

[0086] Step S5, each state feasible set is sequentially introduced into the simulation model to obtain the simulation result of each state feasible set.

[0087] Specifically, each state feasible set can be introduced into the simulation model, and the running state of each state feasible set in the simulation model is updated, and the running condition of the simulation model is determined as the simulation result.

[0088] Step S6, based on the simulation result of each state feasible set, the risk value of the power distribution network is determined.

[0089] Specifically, the simulation results of each state feasible set can be integrated to calculate the risk value of the power distribution network.

[0090] Each simulation result can also be input into a risk prediction model to obtain the risk value of the power distribution network.

[0091] It can be seen from the technical solutions that the power distribution network risk diagnosis method provided by the application can obtain a simulation model of a power distribution network, parameter prediction models of different subjects in the power distribution network, and time sequence state parameters of different subjects in the power distribution network in a current period; the simulation model is used to predict the predicted state parameters of different subjects at a future time after a target period based on the state parameters at a current time in the current period; for each subject, the parameter prediction model of the subject is used to generate the predicted time sequence parameters of the subject in a next period based on the time sequence state parameters of the subject; whether the subject belongs to a state fluctuation subject is evaluated based on the predicted state parameters and the predicted time sequence parameters of the subject; when the subject is a state fluctuation subject, the parameter fluctuation range is determined based on the predicted state parameters and the predicted time sequence parameters; based on this, the predicted state parameters reflect the predicted parameters obtained based on the physical laws of the power grid topology and the physical characteristics of equipment; and the predicted time sequence parameters are data-driven predictions based on historical data trends. The two prediction methods respectively depict parameter changes from the physical mechanism and the data law, and the difference between the prediction results of the two methods can expose the potential uncertainty of parameter changes, thereby accurately positioning the state fluctuation subject; the prediction deviation caused by the interference factors that are not covered by a single model is avoided, and the missed judgment and the misjudgment caused by the traditional fuzzy judgment of the whole system are avoided, thereby reducing the accuracy of risk assessment; on this basis, the application qualitatively describes the fluctuation change of the state fluctuation subject as a quantitative range through the parameter fluctuation range; subsequently, the application can generate a plurality of state feasible sets based on the parameter fluctuation ranges of different state fluctuation subjects, each state feasible set containing feasible state values of each state fluctuation subject; each state feasible set is sequentially introduced into the simulation model to obtain the simulation results of each state feasible set; the risk value of the power distribution network is determined based on the simulation results of each state feasible set; based on this, the application performs multi-scenario simulation coverage on the future operation of the power distribution network based on the quantitative range of the state fluctuation subject, ensures that the subsequent simulation can comprehensively verify the power grid risk under different fluctuation combinations, and avoids the risk omission caused by the traditional single-scenario verification. It can be seen that the application can effectively avoid the fluctuation source fuzzy problem of the traditional shallow data through the cooperation of multiple prediction models, reduce the missed judgment rate and the misjudgment rate, cover multiple fluctuation subject combination scenarios, avoid the limitations of the traditional single-scenario verification, ensure the determination of the risk value without risk dead angles, and further improve the accuracy of risk assessment.

[0092] In some embodiments of the application, the process of evaluating whether the subject belongs to a state fluctuation subject based on the predicted state parameters and the predicted time sequence parameters of the subject in step S3 is described in detail as follows:

[0093] S30, obtain a state parameter comparison model.

[0094] Specifically, a state parameter comparison model can be constructed in advance.

[0095] S31, using the state parameter comparison model, based on the predicted state parameter of the subject and the predicted timing parameter, evaluating whether the subject belongs to a state fluctuation subject.

[0096] Specifically, the predicted state parameter and the predicted timing parameter of the same subject can be input into the state parameter comparison model to obtain a fluctuation value.

[0097] When the fluctuation value exceeds the corresponding fluctuation threshold, it is determined that the corresponding subject belongs to a state fluctuation subject.

[0098] Wherein, the historical fault cases of the power distribution network can be obtained, the operating fluctuation value distribution of different types of subjects before the fault occurs is extracted, and the fluctuation threshold is determined according to the operating fluctuation value distribution.

[0099] For example, if the operating fluctuation value of 80% of the line before the fault occurs is greater than or equal to 30, the fluctuation threshold of the line can be set to 30.

[0100] As can be seen from the above technical solutions, the present embodiment provides an optional way to evaluate whether the subject belongs to a state fluctuation subject based on the predicted state parameter of the subject and the predicted timing parameter. Through the above method, the state parameter comparison model can be introduced for state fluctuation subject evaluation, the evaluation standard can be quantified, and the fluctuation source determination reliability can be improved.

[0101] In some embodiments of the present application, the process of step S30, obtaining the state parameter comparison model, is described in detail, and the steps are as follows:

[0102] S300, constructing a deep learning model.

[0103] Specifically, the deep learning model can be constructed according to an open source algorithm.

[0104] S301, obtaining a plurality of training samples of different training subjects, each training sample containing training state parameters and training timing parameters, the training state parameters and the training timing parameters containing parameter values of a plurality of training parameter items.

[0105] Specifically, training samples of a plurality of different types of training subjects can be obtained.

[0106] Each training sample can contain training state parameters and training timing parameters of the corresponding training subject.

[0107] The training state parameters and the training timing parameters both contain parameter values of a plurality of training parameter items.

[0108] The training state parameters and the training timing parameters of the same training subject have the same training parameter items.

[0109] S302, based on the training state parameter and the training timing parameter of each training sample, the absolute difference of different training parameter items in the same training sample is calculated, and the training fluctuation value of the corresponding training sample is calculated by synthesizing each absolute difference of the same training sample.

[0110] Specifically, the parameter value of the same training parameter item in the training state parameter and the training timing parameter of each training sample can be substituted into the difference calculation function to calculate the absolute difference of the same training parameter item in the same training sample.

[0111] The difference calculation function can be as follows:

[0112]

[0113] In the formula, G i is the absolute difference of the training parameter item i; P1 i is the training parameter value of the training parameter item; P2 i is the training parameter timing value of the training parameter item.

[0114] The weight of each training parameter item corresponding to the same training sample can be determined, wherein the sum of the weights of each training parameter item corresponding to the same training sample is 1.

[0115] Based on the weight of each training parameter item of the same training sample, the absolute difference of each training parameter item of the same training sample is weighted and summed to obtain the sum value of the corresponding training sample.

[0116] The sum value of the corresponding training sample is normalized to [1, 50] by using a normalization function to obtain the training fluctuation value of the corresponding training sample.

[0117] The normalization function can be as follows:

[0118] G b =1+49×

[0119] In the formula, G b is the training fluctuation value of the training sample b; B b is the sum value of the training sample b; N is the total number of training samples.

[0120] S303, based on each training sample and its corresponding training fluctuation value, the deep learning model is trained until the deep learning model meets the preset stopping condition, and the finally obtained deep learning model is the state parameter comparison model.

[0121] Specifically, each training sample can be divided into a training set, a test set and a validation set.

[0122] The deep learning model is trained, tested and verified by using the training set, the test set and the verification set until the deep learning model converges, and finally the obtained deep learning model is the state parameter comparison model.

[0123] From the above technical solution, it can be seen that the embodiment provides an optional way of obtaining a state parameter comparison model. Through the above method, the state parameter comparison model can be trained by quantifying the fluctuation values of different training samples, the model result is optimized, and the model is more accurate and stable.

[0124] In some embodiments of the present application, the process of determining the parameter fluctuation range based on the predicted state parameters and the predicted timing parameters in step S3 is described in detail as follows:

[0125] S30, compare the parameter values of the predicted state parameters and the parameter values of the predicted timing parameters corresponding to the same state parameter item, and determine the parameter value range of the corresponding state parameter item.

[0126] Specifically, the lowest parameter value and the highest parameter value can be selected from the parameter values of the same state parameter item of the same state fluctuation subject and each parameter value of the predicted timing parameter, and the lowest parameter value and the highest parameter value of the same state parameter item are taken as the parameter value range of the corresponding state parameter item of the corresponding state fluctuation subject.

[0127] S31, integrate the parameter value ranges of each state parameter item to obtain the parameter fluctuation range of the state fluctuation subject.

[0128] Specifically, the parameter value ranges of each state parameter item of the same state fluctuation subject are integrated to obtain the parameter fluctuation range of the corresponding state fluctuation subject.

[0129] From the above technical solution, it can be seen that the embodiment provides an optional way of determining the parameter fluctuation range. Through the above method, the parameter fluctuation range can be determined by integrating multiple state parameter items, which improves the coverage scenario of the present application and improves the accuracy of risk assessment.

[0130] In some embodiments of the present application, the process of generating a plurality of state feasible sets based on the parameter fluctuation ranges of different state fluctuation subjects in step S4, each state feasible set containing a feasible state value of each state fluctuation subject, is described in detail as follows:

[0131] S40, gradient value is taken for the parameter fluctuation range of different state fluctuation subjects to obtain a plurality of feasible state values of different state fluctuation subjects.

[0132] Specifically, the parameter value range of each state parameter item of the same state fluctuation subject can be fixed gradient value, and one or more feasible state values of different state parameter items of the same state fluctuation subject are obtained.

[0133] S41, each feasible state value of different subjects is randomly combined to obtain a plurality of state feasible sets.

[0134] Specifically, the feasible state values of each state parameter item of each state fluctuation subject can be randomly combined to obtain a plurality of state feasible sets.

[0135] From the above technical solution, it can be seen that the embodiment provides an optional way to generate a plurality of state feasible sets. The above-mentioned way can further improve the generalization ability of the application.

[0136] In some embodiments of the application, the process of determining the risk value of the power distribution network based on the simulation results of each state feasible set is described in detail, and the steps are as follows:

[0137] S60, based on the simulation results of each state feasible set, the resilience risk index of the corresponding state feasible set is calculated.

[0138] Specifically, each resilience risk index of each state feasible set can be calculated according to the simulation results of each state feasible set.

[0139] S61, based on the resilience risk index of each state feasible set, the risk value of the power distribution network is determined.

[0140] Specifically, each resilience risk index of each state feasible set can be input into a risk analysis model to obtain a feasible risk value of the corresponding state feasible set.

[0141] The average risk value is calculated by combining the feasible risk values of each state feasible set.

[0142] The absolute difference value of any two feasible risk values is calculated to calculate the risk swing value of the corresponding two state feasible sets.

[0143] The average value of each risk swing value is calculated.

[0144] The ratio of the average risk value to the average value of each risk swing value is taken as the risk value of the power distribution network.

[0145] When the risk value is less than the risk threshold, the power distribution network is not processed.

[0146] When the risk value is not less than the risk threshold, a warning is given to maintain the power distribution network in time.

[0147] The obtaining process of the risk analysis model can be:

[0148] building a risk deep learning model;

[0149] obtaining a plurality of elasticity index training sets, and each elasticity index training set is labeled with a corresponding training risk value; the training risk value can be a non-negative number within 20.

[0150] training the risk deep learning model based on each elasticity index training set until the risk deep learning model converges, and finally obtaining the risk deep learning model as the risk analysis model.

[0151] From the above technical solution, it can be seen that the embodiment provides an optional way of determining the risk value of the power distribution network based on the simulation results of each state feasible set. Through the above-mentioned manner, the risk value of the power distribution network can be calculated by comprehensively considering multiple possible scenarios, and the reliability of the risk value is further improved.

[0152] In some embodiments of the present application, the process of calculating the elasticity risk index of the corresponding state feasible set based on the simulation results of each state feasible set in step S60 is described in detail, and the steps are as follows:

[0153] S600, based on the simulation results of each state feasible set, calculating the N-1 / N-2 passing rate, transient stability margin, frequency regulation margin and reactive power compensation margin of the corresponding state feasible set.

[0154] Specifically, the N-1 / N-2 passing rate directly reflects the redundancy design level and elasticity reserve of the power distribution network, determines the proportion of normal power supply scenarios in the simulation results, and can calculate the N-1 / N-2 passing rate.

[0155] The transient stability margin can be calculated by transient simulation combined with stability criteria such as rotor relative angle, etc.

[0156] The frequency regulation margin can be calculated using unit regulation parameters and load characteristics.

[0157] The reactive power compensation margin can be calculated according to the parameters of reactive power equipment and the supply and demand situation of system reactive power.

[0158] From the above technical solution, it can be seen that the embodiment provides an optional way of calculating the elasticity risk index of the corresponding state feasible set based on the simulation results of each state feasible set. Through the above-mentioned manner, the risk situation of the corresponding state feasible set can be further measured by comprehensively considering the N-1 / N-2 passing rate, transient stability margin, frequency regulation margin and reactive power compensation margin.

[0159] Next, the technical solutions of the present application will be described in detail with reference to the accompanying drawings. Figure 2The power distribution network risk diagnosis device provided in the present application is introduced in detail, and the power distribution network risk diagnosis device provided in the following can be mutually compared with the power distribution network risk diagnosis method provided in the foregoing.

[0160] Referring to Figure 2 It can be found that the power distribution network risk diagnosis device can include:

[0161] The acquisition module 10 is configured to acquire a simulation model of a power distribution network, parameter prediction models of different subjects in the power distribution network, and time sequence state parameters of the different subjects in the power distribution network in a current period.

[0162] The prediction module 20 is configured to predict, by using the simulation model, prediction state parameters of the different subjects at future time points after a target period based on state parameters at a current time point in the current period.

[0163] The evaluation module 30 is configured to, for each subject, generate, by using the parameter prediction model of the subject, prediction time sequence parameters of the subject in a next period based on the time sequence state parameters of the subject, evaluate whether the subject belongs to a state fluctuation subject based on the prediction state parameters of the subject and the prediction time sequence parameters, and determine a parameter fluctuation range based on the prediction state parameters and the prediction time sequence parameters when the subject is a state fluctuation subject.

[0164] The generation module 40 is configured to generate a plurality of state feasible sets based on the parameter fluctuation ranges of the different state fluctuation subjects, each state feasible set containing feasible state values of each state fluctuation subject.

[0165] The simulation module 50 is configured to sequentially import each state feasible set into the simulation model to obtain simulation results of each state feasible set.

[0166] The determination module 60 is configured to determine a risk value of the power distribution network based on the simulation results of each state feasible set.

[0167] Further, the evaluation module 30 can include:

[0168] The state parameter comparison model acquisition unit is configured to acquire a state parameter comparison model.

[0169] The subject evaluation unit is configured to evaluate, by using the state parameter comparison model, whether the subject belongs to a state fluctuation subject based on the prediction state parameters of the subject and the prediction time sequence parameters.

[0170] Further, the state parameter comparison model acquisition unit can include:

[0171] The first state parameter comparison model acquisition subunit is configured to construct a deep learning model.

[0172] The second state parameter comparison model obtaining subunit is configured to obtain training samples of multiple different training subjects, each training sample containing training state parameters and training time sequence parameters, the training state parameters and the training time sequence parameters containing parameter values of multiple training parameter items;

[0173] The third state parameter comparison model obtaining subunit is configured to calculate absolute differences of different training parameter items in the same training sample based on the training state parameters and the training time sequence parameters of each training sample, and calculate a training fluctuation value of the corresponding training sample by synthesizing the absolute differences of the same training sample.

[0174] The fourth state parameter comparison model obtaining subunit is configured to train the deep learning model based on each training sample and the corresponding training fluctuation value until the deep learning model meets a preset stopping condition, and finally obtain a deep learning model as the state parameter comparison model.

[0175] Further, the evaluation module 30 can further include:

[0176] The parameter value range determining unit is configured to compare the parameter values of the predicted state parameters and the predicted time sequence parameters corresponding to the same state parameter item, and determine a parameter value range of the corresponding state parameter item.

[0177] The parameter fluctuation range determining unit is configured to synthesize the parameter value ranges of the state parameter items to obtain a parameter fluctuation range of the state fluctuation subject.

[0178] Further, the generation module 40 can include:

[0179] The first generation unit is configured to perform gradient value taking on the parameter fluctuation ranges of different state fluctuation subjects to obtain multiple feasible state values of different state fluctuation subjects.

[0180] The second generation unit is configured to perform random combination on the feasible state values of different state fluctuation subjects to obtain multiple state feasible sets.

[0181] Further, the determination module 60 can include:

[0182] The elasticity risk index determining unit is configured to calculate an elasticity risk index of the corresponding state feasible set based on the simulation result of each state feasible set.

[0183] The risk value determining unit is configured to determine a risk value of the power distribution network based on the elasticity risk indexes of the state feasible sets.

[0184] Further, the elasticity risk index determining unit can include:

[0185] The transient stability margin calculation subunit is configured to calculate, based on the simulation result of each state feasible set, the N-1 / N-2 passing rate, the transient stability margin, the frequency regulation margin and the reactive power compensation margin of the corresponding state feasible set.

[0186] The power distribution network risk diagnosis device provided by the embodiments of the present application can be applied to a power distribution network risk diagnosis device, such as a PC terminal, a cloud platform, a server, a server cluster, and the like. Optionally, Figure 3 A hardware structure block diagram of the power distribution network risk diagnosis device is shown, and the hardware structure of the power distribution network risk diagnosis device can include at least one processor 1, at least one communication interface 2, at least one memory 3 and at least one communication bus 4, as shown in Figure 3

[0187] In the embodiments of the present application, the number of the processor 1, the communication interface 2, the memory 3 and the communication bus 4 is at least one, and the processor 1, the communication interface 2 and the memory 3 complete the communication with each other through the communication bus 4;

[0188] The processor 1 can be a central processing unit CPU, or a specific integrated circuit ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present application, etc.

[0189] The memory 3 can include a high-speed RAM memory, and can also include a non-volatile memory, such as at least one disk memory, etc.

[0190] The memory stores a program, and the processor can call the program stored in the memory, and the program is used to:

[0191] Obtain a simulation model of a power distribution network, parameter prediction models of different subjects in the power distribution network, and time sequence state parameters of the different subjects in the power distribution network in a current period;

[0192] Using the simulation model, based on the state parameters at the current time in the current period, predict the predicted state parameters of the different subjects at future time after a target period;

[0193] For each subject, using the parameter prediction model of the subject, based on the time sequence state parameters of the subject, generate the predicted time sequence parameters of the subject in the next period; based on the predicted state parameters of the subject and the predicted time sequence parameters, evaluate whether the subject belongs to a state fluctuation subject; when the subject is a state fluctuation subject, based on the predicted state parameters and the predicted time sequence parameters, determine a parameter fluctuation range;

[0194] ​generate a plurality of state feasible sets based on the parameter fluctuation ranges of the different state fluctuation subjects, each state feasible set containing feasible state values of the state fluctuation subjects;

[0195] import each state feasible set into the simulation model in sequence to obtain simulation results of each state feasible set;

[0196] determine the risk value of the power distribution network based on the simulation results of the state feasible sets.

[0197] Optionally, the refinement function and the extension function of the program can refer to the description above.

[0198] The embodiment of the application further provides a readable storage medium which can store a program suitable for processor execution, and the program is used for:

[0199] obtaining a simulation model of a power distribution network, parameter prediction models of different subjects in the power distribution network, and time sequence state parameters of the different subjects in the power distribution network in a current period;

[0200] using the simulation model to predict predicted state parameters of the different subjects at future time points after a target period based on state parameters at a current time point in the current period;

[0201] for each subject, using the parameter prediction model of the subject to generate predicted time sequence parameters of the subject in a next period based on the time sequence state parameters of the subject, and evaluating whether the subject belongs to a state fluctuation subject based on the predicted state parameters of the subject and the predicted time sequence parameters; when the subject is a state fluctuation subject, determining a parameter fluctuation range based on the predicted state parameters and the predicted time sequence parameters;

[0202] generate a plurality of state feasible sets based on the parameter fluctuation ranges of the different state fluctuation subjects, each state feasible set containing feasible state values of the state fluctuation subjects;

[0203] import each state feasible set into the simulation model in sequence to obtain simulation results of each state feasible set;

[0204] determine the risk value of the power distribution network based on the simulation results of the state feasible sets.

[0205] Optionally, the refinement function and the extension function of the program can refer to the description above.

[0206] Finally, it should be noted that the terms "first", "second", and the like, herein do not denote any order, quantity, combination, or otherwise, but are used to identify one entity from another, and do not imply any actual relationship or sequence among entities. Also, the use of the term "including", "containing" or any other variant to refer to a list of elements to be combined is intended to denote that not only the listed elements can be present, but also other elements not expressly listed. The conjunction "comprising" does not exclude other elements being present in addition to those identified, and does not exclude further elements that can be inherent to the process, method, article or apparatus.

[0207] The various embodiments in the specification are described with progression in this order of description. Embodiments of each order of description can be combined with embodiments of the other orders of description.

[0208] The above description of disclosed embodiments enables one of ordinary skill in the art to make and use the application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein can be applied to other embodiments without departing from the spirit or scope of the application. Various embodiments of the application can be combined with each other. Accordingly, the present application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for risk diagnosis of a power distribution network, characterized in that, include: The simulation model of the distribution network, the parameter prediction model of different entities in the distribution network, and the time-series state parameters of different entities in the distribution network in the current period are obtained. Using the simulation model, based on the state parameters at the current moment in the current period, the predicted state parameters of different subjects at future moments after the target period are predicted; For each subject, using the subject's parameter prediction model, based on the subject's time-series state parameters, predictive time-series parameters for the subject's next period are generated; based on the subject's predicted state parameters and the predicted time-series parameters, it is assessed whether the subject belongs to a state-fluctuating subject; when the subject is a state-fluctuating subject, based on the predicted state parameters and the predicted time-series parameters, the parameter fluctuation range is determined. Based on the parameter fluctuation range of different state fluctuation subjects, multiple state feasible sets are generated, and each state feasible set contains the feasible state values ​​of each state fluctuation subject. Each feasible set of states is sequentially imported into the simulation model to obtain the simulation results for each feasible set of states; Based on the simulation results of each feasible set of states, the risk value of the distribution network is determined.

2. The distribution network risk diagnosis method according to claim 1, characterized in that, The assessment of whether the subject belongs to a state-fluctuating subject based on the predicted state parameters and the predicted time-series parameters of the subject includes: Obtain state parameters and compare them with the model; Using the state parameter comparison model, based on the predicted state parameters and the predicted time series parameters of the subject, it is assessed whether the subject belongs to a state fluctuation subject.

3. The distribution network risk diagnosis method according to claim 2, characterized in that, The acquisition state parameter comparison model includes: Build deep learning models; Acquire training samples from multiple different training subjects. Each training sample contains training state parameters and training time sequence parameters, which contain parameter values ​​for various training parameter items. Based on the training state parameters and training time parameters of each training sample, the absolute difference of different training parameter items in the same training sample is calculated, and the training fluctuation value of the corresponding training sample is calculated by combining the absolute differences of the same training sample. The deep learning model is trained based on each training sample and its corresponding training fluctuation value until the deep learning model meets the preset stopping condition. The final deep learning model is the state parameter comparison model.

4. The distribution network risk diagnosis method according to claim 1, characterized in that, The predicted state parameters and the predicted time series parameters include parameter values ​​for multiple state parameter items; The step of determining the parameter fluctuation range based on the predicted state parameters and the predicted time series parameters includes: The parameter values ​​of the predicted state parameter and the predicted time series parameter corresponding to the same state parameter item are compared to determine the parameter value range of the corresponding state parameter item. By combining the parameter value ranges of each state parameter item, the parameter fluctuation range of the main state fluctuation entity is obtained.

5. The distribution network risk diagnosis method according to claim 1, characterized in that, The parameter fluctuation range based on different state fluctuation subjects generates multiple feasible sets of states, including: Gradient values ​​are obtained for the parameter fluctuation range of different state fluctuating entities to obtain multiple feasible state values ​​for different state fluctuating entities; By randomly combining the feasible state values ​​of different state fluctuation subjects, multiple feasible state sets are obtained.

6. The distribution network risk diagnosis method according to claim 1, characterized in that, The simulation results, based on each feasible set of states, determine the risk value of the distribution network, including: Based on the simulation results for each feasible state set, the elasticity risk index for the corresponding feasible state set is calculated. The risk value of the distribution network is determined based on the elastic risk index of each feasible set of states.

7. The distribution network risk diagnosis method according to claim 6, characterized in that, The calculation of the elasticity risk index for each feasible state set based on the simulation results includes: Based on the simulation results for each feasible state set, calculate the N-1 / N-2 pass rate, transient stability margin, frequency regulation margin, and reactive power compensation margin for the corresponding feasible state set.

8. A power distribution network risk diagnosis device, characterized in that, include: The acquisition module is used to acquire the simulation model of the distribution network, the parameter prediction model of different entities in the distribution network, and the time-series state parameters of different entities in the distribution network in the current period. The prediction module is used to use the simulation model to predict the predicted state parameters of different subjects at future times after the target period, based on the state parameters at the current time in the current cycle. The evaluation module is used to, for each subject, utilize the subject's parameter prediction model to generate predicted time-series parameters for the next period based on the subject's time-series state parameters; evaluate whether the subject belongs to a state-fluctuating subject based on the predicted state parameters and the predicted time-series parameters; and determine the parameter fluctuation range based on the predicted state parameters and the predicted time-series parameters when the subject is a state-fluctuating subject. The generation module is used to generate multiple feasible state sets based on the parameter fluctuation range of different state fluctuation subjects. Each feasible state set contains the feasible state values ​​of each state fluctuation subject. The simulation module is used to sequentially import each feasible set of states into the simulation model to obtain the simulation results for each feasible set of states. A determination module is used to determine the risk value of the distribution network based on the simulation results of each feasible set of states.

9. A power distribution network risk diagnosis device, characterized in that, Including memory and processor; The memory is used to store programs; The processor is used to execute the program to implement each step of the power distribution network risk diagnosis method as described in any one of claims 1-7.

10. A readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements each step of the power distribution network risk diagnosis method as described in any one of claims 1-7.

Images