CN-122022520-A - Method, system and storage medium for evaluating reliability of park comprehensive energy system based on physical data fusion

CN122022520ACN 122022520 ACN122022520 ACN 122022520ACN-122022520-A

Abstract

The invention discloses a reliability evaluation method, a system and a storage medium of a park comprehensive energy system based on physical data fusion, and relates to the technical field of comprehensive energy systems, wherein the method comprises the steps of forming a natural gas network pipeline average flow velocity data set in a system running state and corresponding scheduling process based on park comprehensive energy system running history data and system parameters; the method comprises the steps of utilizing a dual-path neural network to mine a mapping relation in a data set, constructing a natural gas pipeline average flow velocity prediction model based on a comprehensive energy running state, embedding the average air flow velocity obtained through prediction into a comprehensive energy system linearization dynamic optimal energy flow model, constructing a linearization dynamic optimal energy flow model of physical data fusion, utilizing a Monte Carlo simulation method to form a system running state, and analyzing a load shedding condition of the system state by adopting the optimal energy flow model of the physical data fusion, so as to calculate a reliability index. The method solves the problems of low interpretability and insufficient precision of the traditional data driving method.

Inventors

NI YUWEN
BAO MINGLEI
GUO CHAO
DING YI

Assignees

浙江大学

Dates

Publication Date: 20260512
Application Date: 20260130

Claims (10)

1. A reliability evaluation method of a park comprehensive energy system based on physical data fusion is characterized by comprising the following steps: forming an average flow velocity data set of the natural gas network pipeline in the corresponding scheduling process and containing the system running state based on the park comprehensive energy system running history data and system parameters; utilizing a dual-path neural network (100) to mine a mapping relation in the data set, and constructing a natural gas pipeline average flow velocity prediction model based on the comprehensive energy running state; Embedding the average air flow velocity obtained by prediction into a linearization dynamic optimal energy flow model of the comprehensive energy system, and constructing a linearization dynamic optimal energy flow model of physical data fusion; And forming a system running state by using a Monte Carlo simulation method, and analyzing the load shedding condition of the system state by adopting an optimal energy flow model of physical data fusion so as to calculate the reliability index.
2. The method for evaluating reliability of a campus integrated energy system based on physical data fusion of claim 1, wherein the natural gas network pipeline average flow rate data set comprises, The method comprises the steps of (1) operating state data of a park comprehensive energy system and average flow speed data of a natural gas network pipeline in a corresponding scheduling process, wherein the operating state data comprise an electric power system and a natural gas system; load, power generation capacity, voltage phase angle, branch power and admittance of the power system; load of natural gas system, gas source output, compressor flow, node pressure and pipeline capacity.
3. The method for evaluating reliability of a campus integrated energy system based on physical data fusion according to claim 1 or 2, wherein the natural gas network pipeline average flow rate data set comprises, When the park comprehensive energy system does not actually run and the historical running data is missing, on the basis of the linear dynamic optimal energy flow model, an iteration method is adopted to calculate and collect the gas flow rate, the cut load quantity and the corresponding system state, and training is carried out to generate a data set.
4. The method for evaluating reliability of a campus integrated energy system based on physical data fusion according to claim 3, wherein said linearized dynamic optimal energy flow model comprises, The objective function is expressed as: ; Wherein, the For the integrated energy system reliability evaluation period length, For the time index of the time index, Is the first The unit cost required to cut off the power load per unit time, Indexing the nodes of the power system, For a set of nodes of a power system, Is the first Each unit time is at node The load of the power system is cut off, Is the first The unit cost required to cut off the natural gas load per unit time, Is an index of the nodes of the natural gas system, Is a set of nodes of a natural gas system, Is the first Each unit time is at node Load of the natural gas system cut off.
5. The method for evaluating reliability of a campus integrated energy system based on physical data fusion according to claim 3, wherein the iterative method comprises, And substituting the average air flow velocity obtained by the previous iteration into the linearization dynamic optimal energy flow model as a parameter to solve again, and continuously iteratively updating the air flow velocity value until the difference value of the results of two adjacent iterations is smaller than a set threshold value.
6. The method for evaluating reliability of a campus integrated energy system based on physical data fusion according to any one of claims 1, 2, 4 to 5, wherein the dual-path neural network (100) comprises, The system comprises an input part (101), a plurality of dual-path modules (102) and an output part (103), wherein the input part (101) transmits processed data to the dual-path modules (102), the dual-path modules (102) combine a residual path (1021) and a dense connection path (1022), and output characteristics output a prediction result through a pooling layer and a full connection layer of the output part (103); The residual path (1021) performs addition operation on part of the characteristics after the characteristic extraction and the module input; The dense connection path (1022) concatenates another portion of the feature extracted feature with the module input.
7. The method for evaluating reliability of a campus integrated energy system based on physical data fusion according to claim 5, wherein the linearized dynamic optimal energy flow model of physical data fusion comprises, And the linear dynamic optimal energy flow model is a linearization result of the nonlinear dynamic optimal energy flow model under an operating point based on an embedded model of the average airflow speed obtained by combining the linearization dynamic optimal energy flow model with the prediction.
8. The method for evaluating reliability of a campus integrated energy system based on physical data fusion according to any one of claims 1, 2, 4 to 5 and 7, wherein the reliability index comprises, The power shortage probability, the electricity shortage desire, the gas shortage probability, and the gas shortage desire are expressed as: ; ; ; ; Wherein, the Is the first The amount of electricity per unit time is less than desired, Is the first Probability of power shortage per unit time, Is the first A gas shortage per unit time is expected, Is the first Probability of a gas shortage per unit time, For the time index of the reliability evaluation, For the number of samples of the monte carlo method, For the integrated energy system reliability evaluation period length, Is the first Status of each unit time Lower node Is set to be a power load cut-off amount of the (c), For the system state index(s), Is the first Status of each unit time Lower node Natural gas load cut-off amount of (c).
9. The reliability evaluation system for the park comprehensive energy system based on the physical data fusion is characterized by comprising a data set construction module, a dual-path neural network prediction module, a physical data fusion model construction module and a reliability evaluation module, wherein the reliability evaluation method is based on the physical data fusion; The data set construction module is used for forming a data set containing the system running state and the average flow speed of the natural gas network pipeline in the corresponding scheduling process based on the running history data and the system parameters of the park comprehensive energy system; the dual-path neural network prediction module is used for establishing a natural gas pipeline average flow velocity prediction model based on the comprehensive energy running state by utilizing the constructed data set and through the mapping relation between the running state of the dual-path neural network (100) excavation system and the natural gas pipeline average flow velocity; the physical data fusion model construction module is used for embedding the average air flow speed obtained through prediction into the linear dynamic optimal energy flow model of the comprehensive energy system to construct a linear dynamic optimal energy flow model of physical data fusion; The reliability evaluation module is used for sampling the operation state of the park comprehensive energy system by adopting a Monte Carlo method, analyzing the load shedding condition of each system state by utilizing an optimal energy flow model fused by physical data, and calculating each reliability index.
10. A computer readable storage medium having stored thereon a computer program, wherein the computer program when executed by a processor implements the steps of the method for reliability assessment of a campus integrated energy system based on physical data fusion of any one of claims 1 to 8.

Description

Method, system and storage medium for evaluating reliability of park comprehensive energy system based on physical data fusion Technical Field The invention relates to the technical field of comprehensive energy systems, in particular to a method, a system and a storage medium for evaluating reliability of a park comprehensive energy system based on physical data fusion. Background Under the dual-carbon background, the construction of a novel energy system is important for realizing comprehensive green transformation. In particular, the integrated energy system is an important component of a novel energy system, and can realize complementary and mutual-aid of multiple types of energy sources, so that the safety level of the energy system is improved. The park comprehensive energy system is an important branch form of the floor application of the comprehensive energy system. Due to the rapid development of gas generator sets, the coupling degree of an electric power system and a natural gas system is gradually improved, and an electric-gas coupling system is a typical form of a comprehensive energy system. In an electro-pneumatic coupling system, a failure of a subsystem may propagate through the coupling element to other parts, thereby severely threatening system reliability. Therefore, it is necessary to develop reliability evaluation research of the park comprehensive energy system. For effectively evaluating the reliability of the park comprehensive energy system, the key is to analyze the running state change of the natural gas and the power flow under each state event. But unlike the instantaneous balancing of the power flow, the transient course of the natural gas flow is longer in duration. Therefore, it is necessary for the natural gas tide transient to incorporate a reliability assessment. Due to the high computational complexity of natural gas flow transients, conventional methods typically set the natural gas flow rate to a constant to simplify the original problem, even ignoring transients (using steady state airflow models such as the wechat equation). In practice, however, there is a significant fluctuation in natural gas flow rate due to differences in status events, possibly deviating from a preset reference, a phenomenon known as event-triggered unsteady transient gas flow rate. The natural gas flow rate error can influence the natural gas network pipeline characteristics, and then the reliability calculation accuracy of the comprehensive energy system of the whole park is greatly influenced. Therefore, event triggered unsteady transient airflow rates must be incorporated into the reliability assessment process for the campus integrated energy system. To improve the accuracy of reliability assessment, scholars propose various methods to achieve integrated energy system reliability assessment based on event-triggered unsteady transient airflow velocity, such as space-time discrete methods, iterative methods, etc. However, the space-time discrete method can lead to the rapid increase of the complexity of the model due to the introduction of a large number of differential equations, and the iterative method needs repeated calculation for many times on the linear original model to improve the calculation accuracy, which takes a long time. Therefore, the existing method is difficult to meet the rapidity requirement of reliability evaluation of the park comprehensive energy system. To improve the evaluation efficiency, scholars have proposed various reliability rapid evaluation methods based on data driving models, such as least squares support vector machine, random forest model, etc. However, these methods have inadequate consideration of inherent physical characteristics of the integrated energy system, and relatively low interpretability and accuracy. To achieve interpretable and efficient evaluations, physical information notification data-driven methods are receiving increasing attention. The method is characterized in that the data driving model is integrated into the physical model, so that the problem of black box attribute is effectively relieved. In order to solve the problems, a reliability evaluation method of a park comprehensive energy system based on physical data fusion is provided, which considers event-triggered unstable transient airflow speed. First, in the absence of historical data, an iterative-based optimal energy flow model is used to generate gas flow rates and load shedding amounts at different state events. A dual path neural network approach is then used to construct a mapping of gas flow rates to state events. Based on the predicted airflow velocity, a nonlinear momentum equation describing the airflow is effectively simplified. Finally, a physical data fusion model is proposed to implement the state analysis in reliability assessment. The method can effectively improve the rapidity and the interpretability of reliability evaluation of the park comprehensive energy