Intelligent power grid energy storage fault monitoring method and system based on digital twinning

Abstract

The invention discloses an intelligent power grid energy storage fault monitoring method and system based on digital twinning. The method comprises the following steps: step 1, constructing a dynamic model of an energy storage system; 2, constructing a digital twin estimator based on the dynamic model constructed in the step 1; step 3, fault detection is carried out based on the estimator constructed in the step 2; and step 4, performing fault isolation based on a fault detection result in the step 3. And constructing a theoretical relative fault sensitivity matrix by analyzing the relative sensitivity of the residual error to the reconstructed different faults, and accurately identifying the fault type based on the Euclidean distance method. According to the method, the fault detection and isolation precision can be improved, the operation safety and reliability of the intelligent power grid energy storage system are effectively improved, and the maintenance cost is reduced.

Description

Technical Field

[0001] The present invention relates to the field of smart grid technology, and in particular to a smart grid energy storage fault monitoring method based on digital twins. Background Art

[0002] With the development of smart grids, energy storage systems have played an important role in balancing electricity supply and demand and improving grid stability. However, due to their complex structure and operating environment, energy storage systems are susceptible to component aging, sensor failure and other external factors, which can lead to system performance degradation and even safety hazards. Current fault monitoring methods mainly rely on data-driven machine learning algorithms or model-driven diagnostic techniques, but these methods have the following limitations: 1. Data-driven methods require a large amount of labeled data and are difficult to adapt to distribution changes or insufficient data. 2. Traditional model-driven methods have high requirements for system modeling accuracy and are not adaptable to complex systems with nonlinear and dynamic changes.

[0003] The prior art discloses "Research on Digital Twin Fault Diagnosis Model of Proton Exchange Membrane Fuel Cell System", the first author is Zhu Jing, "Control Theory and Applications" in Volume 39, Issue 3, March 2022. The parallel fault diagnosis method based on digital twins performs fault diagnosis by detecting and evaluating the residuals between the real system and the digital twin system. The relevant digital twin is established based on the proton exchange membrane fuel cell kinetic model and data set. If the residual vector exceeds the fault detection threshold, the relative sensitivity of the fault residual is used to perform fault isolation. However, the shortcomings of the prior art are:

[0004] 1. Insufficient robustness. The digital twin method relies on the collection and transmission of real-time sensor data. If the sensor data is noisy, inaccurate, or has packet loss, it may affect the diagnostic accuracy. However, the prior art does not discuss in detail how to maintain diagnostic accuracy in the case of strong disturbances or significant nonlinear uncertainties.

[0005] 2. Dependence on model accuracy. Digital twins require highly accurate models to perform effective fault diagnosis. However, improving model accuracy usually relies on a large amount of experimental data and a complex modeling process.

[0006] Digital twin technology provides a new idea for fault monitoring of energy storage systems. By building a digital virtual model of the physical system and combining real-time data updates with dynamic analysis, more accurate fault detection and isolation can be achieved. Summary of the invention

[0007] In order to solve the deficiencies in the prior art, the present invention provides a smart grid energy storage fault monitoring method based on digital twins to solve the problems existing in the background technology.

[0008] The present invention adopts the following technical solution.

[0009] The first aspect of the present invention provides a smart grid energy storage fault monitoring method based on digital twin, comprising:

[0010] Step 1, constructing a dynamic model of the energy storage system;

[0011] Step 2, building a digital twin estimator based on the dynamic model built in step 1;

[0012] Step 3, perform fault detection based on the estimator constructed in step 2;

[0013] Step 4: perform fault isolation based on the fault detection result of step 3.

[0014] Preferably, in step 1, the dynamic model is in the form of a differential equation:

[0015]

[0016] Among them, F represents the state matrix, which describes the linear relationship within the system; G represents the nonlinear function of input and state; Eξ represents the external interference term; u is the system voltage; is the state variable; p cp , ω cp ,p sm represent the cathode pressure, nitrogen pressure, compressor rotation speed and supply pipe pressure at atmospheric temperature respectively.

[0017] Preferably, in step 1, the state matrix F, the nonlinear function G of the input and state, and the external interference term Eξ are:

[0018]

[0019]

[0020] E = [-b7 000] T

[0021] X=x1+x2+b2

[0022]

[0023] f2(x3,x4)=W cp

[0024]

[0025] Among them, W ca,out is cathode gas flow; b i, i∈{1,...,17} is a constant obtained from the physical constants of the system, expressed as:

[0026]

[0027] Among them, W cp is the compressor flow rate; R is the universal gas constant; T fc is the temperature of the fuel cell; k ca,in is the cathode inlet constant; is the oxygen mass fraction at the air inlet; is the molar mass of oxygen; V ca is the cathode volume; ω atm is the relative humidity of the ambient air; p sat is the supply pipe saturation pressure at atmospheric temperature; is the molar mass of nitrogen; M v,ca is the steam molar mass; n is the number of cells in the fuel cell stack; F is the Faraday constant; k v ,k t ,R cm are all motor constants; η cm is the motor mechanical efficiency; J cp is the compressor inertia; C p is the constant pressure specific heat of air; T atm is the ambient temperature; η cp is the compressor efficiency; p atm is the atmospheric pressure; γ is the specific heat ratio of air; M a is the molar mass of air; V sm is the supply manifold volume;

[0028] The output y(t) of the energy storage system is expressed as:

[0029]

[0030] Among them, the oxygen excess ratio Indicates the oxygen consumption in the air supply system; V st(t) is the stack voltage; I cm (t) and ω cp(t) Respectively represent the current and speed of the compressor;

[0031] The performance variables of the energy storage system are expressed as:

[0032]

[0033] Among them, P net (t) represents the power of the battery.

[0034] Preferably, in step 2, a sliding mode observer is established in the digital twin estimator model based on a super-helical sliding mode control algorithm; the dynamic model of the sliding mode observer is:

[0035]

[0036] Where f(t) represents the error signal, D(y(t),u) is a smooth and bounded function that depends on the input and output of the system, and g(z(t),u) is Lipschitz continuous;

[0037] The output of the sliding mode observer is:

[0038]

[0039] The performance variables of the sliding mode observer are expressed as:

[0040]

[0041] Preferably, the dynamic model of the sliding mode observer is:

[0042]

[0043]

[0044] Among them, v is the sliding mode error injection term, k1, k2 are gain functions.

[0045] Preferably, in step 3, fault detection adopts a residual detection method, and the residual calculation formula is:

[0046]

[0047] Under sliding mode motion, the dynamic equation of the residual is:

[0048]

[0049] in,

[0050] Preferably, when γ(t) = 0 and When , the energy storage system is in sliding mode motion;

[0051] Under sliding mode motion, if e2+Δg=0, no fault occurs, and γ(t) and remains stable; otherwise, it means that the energy storage system fails, and γ(t) and fluctuation.

[0052] Preferably, in step 4, fault isolation is performed by analyzing the fault detection residual γ(t) and the fault The relationship between the residual and the fault is defined in the Cartesian product, where FSM is the theoretical fault feature matrix. Fault isolation is achieved by comparing the residuals corresponding to different faults in FSM.

[0053] Fault The calculation formula is:

[0054]

[0055] Preferably, in step 4, a residual error sensitivity analysis method to faults is adopted;

[0056] The sensitivity is:

[0057]

[0058] Based on the sensitivity, the relative sensitivity function is established, and the calculation formula is:

[0059]

[0060] Through the relative sensitivity function, a theoretical relative fault sensitivity matrix is ​​established. The values ​​in this matrix are the fault f obtained by statistical methods. j Theoretical value when it occurs Faults are identified by Euclidean distance, and the calculation formula is:

[0061]

[0062] in, For identified faults.

[0063] The second aspect of the present invention discloses an energy storage fault monitoring system, which adopts the above-mentioned smart grid energy storage fault monitoring method based on digital twin, including:

[0064] Dynamic model building module, digital twin estimator building module, fault detection module and fault isolation module;

[0065] Among them, the dynamic model building module is used to build a system dynamic model to describe the state changes of the system;

[0066] The digital twin estimator building module is used to build a digital twin estimator;

[0067] The fault detection module is used to determine whether a fault occurs in the system;

[0068] The fault isolation module is used to achieve fault isolation by comparing the residuals corresponding to different faults.

[0069] The beneficial effect of the present invention is that, compared with the prior art,

[0070] The shortcomings of the current related patents are mainly concentrated in real-time, accuracy, energy efficiency and system stability. The lack of application of digital twin technology may limit the accuracy and response speed of fault detection and diagnosis. At the same time, the energy loss of the system needs to be further optimized to improve the efficiency of power utilization. In addition, for fluctuations and interference in complex power grid environments, system stability and robustness also need to be better considered and strengthened. Compared with some existing fault monitoring methods based on digital twins, the present invention provides a system dynamic equation with more real-time and dynamic characteristics, and adopts residual analysis and sensitivity matrix methods, which can accurately isolate the fault location and improve the accuracy of fault classification.

[0071] The present invention enhances the robustness to external disturbances and uncertainties through the super-helical sliding mode control technology, and at the same time, makes the fault diagnosis more accurate and efficient through the real-time updating of the residual dynamic equation.

[0072] The present invention establishes a sliding mode observer based on digital twins, and monitors faults through the dynamic equation of residuals, which can maintain a high diagnostic capability even if there is a certain uncertainty in the model.

[0073] The significant advantages of the present invention are as follows. First, it has strong real-time and dynamic performance. Through the real-time synchronization of the digital twin model and the physical energy storage system, millisecond-level fault detection and diagnostic response can be achieved, effectively avoiding system losses caused by delays. Secondly, the present invention adopts residual analysis and sensitivity matrix methods to accurately isolate the fault location, and improve the accuracy of fault classification through the theoretical relative fault sensitivity matrix to avoid misjudgment and missed judgment. At the same time, the present invention supports dynamic parameter optimization and real-time correction, maximizes the efficiency of power utilization, and significantly reduces the energy loss of the system. In response to fluctuations and interference in complex power grid environments, the digital twin estimator can be dynamically adjusted to enhance the robustness of the system and ensure the stable operation of the energy storage system under complex conditions. In addition, the present invention supports predictive maintenance, which can extend the service life of equipment and reduce operating and maintenance costs by identifying potential problems in advance. It has strong versatility and scalability, and is suitable for a variety of energy storage technologies and energy storage systems of different scales, especially for the integration of smart grids and renewable energy. Overall, the present invention significantly improves the reliability, operating efficiency and adaptability of the energy storage system, and provides an efficient and economical solution for smart grid energy storage. BRIEF DESCRIPTION OF THE DRAWINGS

[0074] Attached Figure 1 is a flow chart of the steps of the present invention;

[0075] Attached Figure 2 is a detailed process flow chart of the present invention;

[0076] Attached Figure 3 A detailed flowchart of the fault isolation steps. DETAILED DESCRIPTION

[0077] In order to make the purpose, technical scheme and advantages of the present invention clearer, the technical scheme of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. The embodiments described in this application are only embodiments of a part of the present invention, rather than all embodiments. Based on the spirit of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the protection scope of the present invention.

[0078] The technical solution of the present invention is further described below in conjunction with the accompanying drawings.

[0079] like Figure 1-Figure 3 As shown, an embodiment of the present invention provides a smart grid energy storage fault monitoring method based on digital twin, comprising the following steps:

[0080] 1. Construction of dynamic model of energy storage system;

[0081] Specifically as follows: In a preferred but non-limiting embodiment of the present invention, the dynamic behavior of the gas in the cathode of the energy storage battery is described as follows:

[0082]

[0083] in, and are the partial pressures of oxygen and nitrogen respectively; T fc is the temperature of the fuel cell; R is the universal gas constant, which is 8.314 J / (mol·k); and is the molar mass of oxygen and nitrogen; V ca is the cathode volume; and are the inlet and outlet flow rates of oxygen respectively; and are the inlet and outlet flow rates of nitrogen respectively; is the oxygen consumption in the reaction, expressed as:

[0084]

[0085] Where n is the number of proton exchange membrane fuel cells; I st is the current generated by the reaction; F is the Faraday constant.

[0086] The pressure dynamics of the supply manifold is described by the following formula:

[0087]

[0088] Among them, p smis the supply manifold pressure; R is the universal gas constant; T cp,out is the compressor outlet temperature; W cp is the compressor flow rate; V sm is the supply manifold volume; W sm,out is the supply manifold outlet flow rate, given by the nozzle flow equation:

[0089] W sm,out =k sm,out (p sm -p ca )

[0090] Among them, k sm,out is the supply manifold outlet constant; p sm is the supply pipe pressure at atmospheric temperature; p ca is the cathode pressure.

[0091] Dynamic passing speed of the compressor ω cp Description of changes:

[0092]

[0093] Among them, J cp is the compressor inertia; τ sm is the motor driving torque; τ cp is the compressor load torque, expressed by the following formula:

[0094]

[0095] Among them, C p is the constant pressure specific heat of air; T atm is the ambient temperature; η cp is the compressor efficiency; p atm is the atmospheric pressure; γ is the specific heat ratio of air.

[0096] The dynamic model of the system adopts the form of differential equations to describe the state changes of the system. The above equations are unified in the form of state space as follows:

[0097]

[0098] Among them, F is the state matrix, which describes the linear relationship within the system; G is the nonlinear function of input and state, which represents the electrochemical dynamic characteristics of the energy storage device; Eξ is the external interference term, such as the influence of ambient temperature or grid load; u is the system voltage. is a state variable. cp , ω cp ,p smThey represent the cathode pressure, nitrogen pressure, compressor rotation speed and supply pipe pressure at atmospheric temperature respectively. The relevant variables can be expressed as

[0099]

[0100]

[0101] E = [-b7 000] T (9)

[0102] in W ca,out is the cathode air flow. u is the system voltage. b i , i∈{1,...,17} is a constant obtained from the physical constants of the system, which can be expressed as

[0103]

[0104] Among them, W cp represents the compressor flow rate. R represents the universal gas constant. T fc represents the temperature of the fuel cell. k ca,in Represents the cathode inlet constant. Indicates the oxygen mass fraction at the air inlet. Indicates the molar mass of oxygen. ca Represents the cathode volume. ω atm Indicates the relative humidity of the ambient air. sat Indicates the supply line saturation pressure at atmospheric temperature. Indicates the molar mass of nitrogen. M v,ca represents the molar mass of steam. n represents the number of cells in the fuel cell stack. F represents the Faraday constant. k v ,k t ,R cm Both represent motor constants. η cm Indicates the mechanical efficiency of the motor. J cp Indicates the compressor inertia. C p represents the constant pressure specific heat of air. atm Indicates the ambient temperature. cp Indicates the compressor efficiency. atm represents the atmospheric pressure. γ represents the specific heat ratio of air. M a Indicates the molar mass of air. V sm Indicates the supply manifold volume.

[0105] The energy storage system output y(t) can be expressed as:

[0106]

[0107] Among them, the oxygen excess ratio Indicates the oxygen consumption in the air supply system; V st(t) is the stack voltage; I cm (t) and ω cp(t) Represent the current and speed of the compressor respectively. The performance variables of the energy storage system are expressed as:

[0108] z(t)=[λ O2 (t),P net (t)], (12)

[0109] Among them, P net (t) represents the power of the battery.

[0110] This section introduces the dynamic model of the energy storage system. Compared with the prior art, the beneficial effects on the monitoring method are mainly reflected in the design and application of dynamic equations. Although the modeling method adopted is similar to the prior art, the introduction of dynamic equations is significantly different: the present invention adds more state variables related to the physical process of the energy storage system to the dynamic equation, strengthens the description of nonlinear dynamic characteristics, and considers the impact of environmental disturbances on the system. This improvement not only improves the accuracy and adaptability of the model, but also provides key support for the efficient implementation of digital twin technology. Ultimately, these improvements enable the present invention to achieve more accurate fault detection and isolation in complex power grid environments, and improve the robustness and energy efficiency of energy storage systems.

[0111] The specific beneficial effects are as follows: (1) New value brought by the differences in dynamic equations. For example, specific state variables, nonlinear descriptions, or considerations of external disturbances are introduced into the dynamic equations. These designs improve the practical adaptability and performance of the model. Innovation: Variable design: The dynamic equation introduces more physically meaningful state variables (such as oxygen and nitrogen partial pressures, compressor speed, etc.), which are more refined than existing technologies. Nonlinear processing: The dynamic equation expresses the nonlinear relationships within the system (such as gas flow and reaction dynamics) in a specific form, which can more accurately reflect the actual behavior of the energy storage system. External disturbance modeling: Special consideration is given to the influence of external disturbances such as ambient temperature and grid load, laying the foundation for robust design under complex operating conditions.

[0112] (2) The key role of dynamic equations in the overall system. Dynamic equations are the core of the combination of energy storage systems and digital twins, and directly affect the subsequent estimator design and fault diagnosis performance. Although the methods are similar, the differences in dynamic equations provide key support for the implementation of digital twins. The newly established dynamic equations optimize the existing technology in the following aspects: Improved model accuracy: Through dynamic equations that are closer to physical reality, modeling errors are reduced and the reliability of system description is enhanced. Real-time and dynamic adjustment capabilities: Combined with dynamic equations, it is possible to capture system state changes in real time and provide an efficient parameter update mechanism for digital twins.

[0113] (3) The unique value of dynamic equations in the overall framework of the present invention. Although the way of establishing dynamic equations is inherited, it is an indispensable part of the overall architecture of the present invention. Combined with dynamic equations, the present invention can: achieve more accurate fault detection and isolation (through residual analysis). Improve the adaptability of energy storage systems in dynamic environments (real-time updates through digital twin estimators). Support a wider range of energy storage technologies and complex power grid environments, and enhance the versatility of the system.

[0114] (4) The dynamic equations in the prior art may be oversimplified, ignoring some key variables or dynamic characteristics, resulting in insufficient applicability and accuracy of the model. The present invention explicitly considers which variables or nonlinear relationships (such as oxygen consumption rate, stack output power, etc.) in the dynamic equations. The dynamic equations more comprehensively describe the complex behavior within the energy storage system, providing new possibilities for optimization and control.

[0115] 2. Digital Twin Estimator Establishment

[0116] The digital twin estimator can be model-driven, data-driven, or a combination of the two. In a preferred but non-limiting embodiment of the present invention, a model-driven digital twin estimator is constructed. A sliding mode observer is established in the digital twin model based on a super-twisting sliding mode control algorithm (ST).

[0117] The use of this algorithm based on the digital twin-based fault monitoring method requires overcoming the following difficulties: (1) Nonlinearity and uncertainty processing. The model in the prior art still relies on linearization methods to describe the dynamic behavior of the system, which is difficult to accurately deal with complex nonlinearities and dynamic changes. The sliding mode algorithm needs to adapt to the high nonlinear characteristics and parameter uncertainties in the energy storage system, such as ambient temperature fluctuations, grid load changes, etc. (2) Model dependence and error compensation. The digital twin estimator in the prior art relies heavily on the accuracy of the model. If the modeling error or sensor noise is large, it will directly affect the performance of fault detection and isolation. The sliding mode algorithm needs to overcome the problem of high dependence on the model of traditional methods, while avoiding the cumulative effect of model errors. (3) Real-time and computational complexity The sliding mode algorithm involves high-order derivative calculations and real-time residual updates, which may bring high computational complexity, especially in millisecond response scenarios. The application of the sliding mode algorithm needs to take into account the constraints of computing resources and the requirements of real-time performance. (4) Robustness and sensitivity. Optimizing the sliding mode control itself is prone to produce a "jitter effect". Under the fluctuation and disturbance conditions in a complex power grid environment, it is necessary to optimize robustness while maintaining high sensitivity to faults.

[0118] The present invention uses a sliding mode algorithm to solve the following problems: (1) Improved adaptability to dynamic environments. The sliding mode algorithm solves the problem of poor performance of digital twin models in dynamic environments (such as load changes and sudden fault behavior) by compensating for the strong robustness of nonlinear dynamic systems. (2) Improved fault signal reconstruction and isolation accuracy. The use of a super-spiral sliding mode algorithm for real-time reconstruction of fault signals greatly improves the isolation accuracy in complex multi-fault scenarios, especially the ability to distinguish similar residual faults. (3) Reducing the impact of modeling errors. The sliding mode algorithm uses its anti-interference ability to effectively compensate for model errors and environmental interference, avoiding the problem of significant performance degradation of traditional digital twin estimators when modeling is inaccurate or sensor noise exists. (4) Improved real-time performance and response speed. The super-spiral sliding mode algorithm solves the delay and error accumulation problems of traditional sliding mode control algorithms through an improved sliding mode observer design, achieving millisecond-level response.

[0119] The functions of the sliding mode algorithm in the present invention are: (1) enhancing the robustness of the digital twin estimator. The sliding mode algorithm is combined with the digital twin model to compensate for model errors and external disturbances through anti-interference ability, thereby improving the robustness and stability of the fault monitoring system. (2) Real-time dynamic adjustment. Through the prediction-correction framework and the super-helical sliding mode control algorithm, the state and output of the system can be updated in real time to adapt to dynamic changes in complex environments. (3) Residual generation and fault signal reconstruction. The sliding mode observer realizes the reconstruction and high-precision isolation of fault signals through residual generation and dynamic equation processing. (4) Improving fault isolation performance. Using the theoretical relative fault sensitivity matrix and the Euclidean distance method, combined with the wide range stability of the sliding mode algorithm, accurate classification of various complex faults is achieved.

[0120] Based on the above description, the beneficial effects produced by the present invention are: (1) Improving the accuracy of fault detection and isolation. The adaptability of the sliding mode algorithm to nonlinearity, uncertainty and dynamic disturbances significantly improves the accuracy of fault detection and the precision of fault isolation. (2) Improving system robustness. In complex environments (such as fluctuating loads or environmental interference), the sliding mode algorithm maintains the stability and reliability of the digital twin system. (3) Enhancing real-time performance and response speed. Realize rapid fault detection and response, and effectively avoid system performance degradation and safety hazards caused by delays. (4) Expand the scope of application. The sliding mode algorithm combined with digital twin technology makes the system more versatile and suitable for a wider range of energy storage systems and complex power grid environments. (5) Reduce energy loss and maintenance costs. Through real-time fault detection and early fault isolation, predictive maintenance can be achieved, equipment life can be extended, and operation and maintenance costs can be reduced.

[0121] The dynamic model of the sliding mode observer can be expressed as: Among them, f(t) represents the error signal, D(y(t),u) is a smooth and bounded function that depends on the input and output of the system, that is, it represents the mapping relationship between output and input, and g(z(t),u) is Lipschitz continuous. Similarly, the output of the sliding mode observer is: The performance variables of the sliding mode observer are expressed as:

[0122] In a preferred but non-limiting embodiment of the present invention, the matrix D represents the relationship between the output y and the input u under the influence of the error signal f(t), the D function is expressed as [0.02, 0, 01; 0, 05, 0, 03], and y can be expressed as:

[0123] y1=0.02f1+0.01f2

[0124] y2=0.05f1+0.03f2

[0125] Furthermore, the error signal can be expressed as The difference between y(t) and y(t). Therefore, the dynamic model of the sliding mode observer can be further expressed as Where v is the sliding mode error injection term obtained by the ST algorithm, which can be expressed as:

[0126]

[0127] Among them, k1 and k2 are gain functions, which are used to adjust the system response speed and robustness.

[0128] 3. Fault detection based on digital twins

[0129] The following introduces the fault detection (FD) method based on digital twins. As the first step of the fault diagnosis method, FD is detected based on the residual γ(t), which is obtained by subtracting the estimated output y(t) from the measured system output y(t). The residual γ(t) is expressed as

[0130]

[0131] Under sliding mode motion (system error tends to zero), the dynamic equation of the residual can be expressed as:

[0132]

[0133] in

[0134] When the system is in sliding mode motion (i.e., γ(t) = 0 and ), the dynamic equation of the residual can be simplified to:

[0135] 0=e2+Δg+D(y(t),u)f(t). (16)

[0136] At this time, if no fault occurs, e2+Δg=0, and the residual signal remains stable; if a fault occurs, the residual signal will change, expressed as γ(t) and fluctuations.

[0137] Therefore, when γ(t) is close to zero for a long time and When γ(t) deviates or When fluctuations occur, it indicates that there is a problem with the system.

[0138] 4. Fault isolation based on digital twins

[0139] The following introduces the fault isolation (FI) method based on digital twins. Fault isolation is the second step of fault diagnosis, which is to identify the faults that affect the system.

[0140] First, the fault signal is reconstructed. When the system is in sliding mode motion, there is Therefore, we can get

[0141] v(γ(t))=e2+Δg+D(y(t),u)f(t). (17)

[0142] Since D is a reversible matrix, the fault signal can be estimated by the sliding mode error injection term v:

[0143]

[0144] Next, based on the fault reconstruction Analysis of the residual γ(t) of the fault detection FD and the fault The relationship between the residual γ(t) and the fault The relationship between is defined in the Cartesian product Where FSM is the theoretical fault feature matrix. Fault isolation is achieved by comparing the residuals corresponding to different faults in FSM.

[0145] In order to improve the fault isolation performance, the residual error sensitivity analysis method is adopted. The sensitivity function is defined as:

[0146] This formula provides the breakdown by value and sign The quantitative and qualitative information on the impact of the residual γ(t) can be used to solve the problems existing in the method of fault isolation using FSM.

[0147] In online fault diagnosis, It is obtained from the residual error when the fault occurs. It should be noted that the calculation of this sensitivity requires the qualitative and quantitative information of the fault to be known or approximately known, which needs to be passed through a complex fault reconstruction algorithm. For this purpose, the relative sensitivity function based on the sensitivity function formula is given as follows Among them, the remaining γ k (t) The validity of the relative sensitivity function should be ensured, that is, the γ k (t)≠0. In order to facilitate the following analysis, the residual γ k (t) is regarded as the first one in the residual γ(t), that is, k=1.

[0148] As shown in Table 1, using the relative sensitivity function, the theoretical relative fault sensitivity matrix (TRFSM) is established. j After the occurrence, the value was obtained by statistical method.

[0149] Table 1 Theoretical relative fault sensitivity matrix

[0150]

[0151] Then, the real-time ratio of the residuals obtained by the relative sensitivity function is Stored as a vector, and compared with the vector in TRFSM to obtain the corresponding distance information. Calculated by Euclidean distance:

[0152]

[0153] Minimum distance corresponding to fault type This is the identified fault.

[0154] Based on the above statements, the overall algorithm of the smart grid energy storage fault monitoring method based on digital twin is given below.

[0155]

[0156] The embodiment of the present invention further provides a fault monitoring system based on the above fault monitoring method, comprising:

[0157] Dynamic model building module, digital twin estimator building module, fault detection module and fault isolation module;

[0158] Among them, the dynamic model building module is used to build a system dynamic model to describe the state changes of the system;

[0159] The digital twin estimator building module is used to build a digital twin estimator;

[0160] The fault detection module is used to determine whether a fault occurs in the system;

[0161] The fault isolation module is used to achieve fault isolation by comparing the residuals corresponding to different faults.

[0162] The present disclosure may be a system, a method and / or a computer program product. The computer program product may include a computer-readable storage medium carrying computer-readable program instructions for causing a processor to implement various aspects of the present disclosure.

[0163] A computer-readable storage medium may be a tangible device that can hold and store instructions used by an instruction execution device. A computer-readable storage medium may be, for example, but not limited to, an electrical storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples of computer-readable storage media (a non-exhaustive list) include: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a static random access memory (SRAM), a portable compact disk read-only memory (CD-ROM), a digital versatile disk (DVD), a memory stick, a floppy disk, a mechanical encoding device, such as a punch card or a raised structure in a groove on which instructions are stored, and any suitable combination of the foregoing. As used herein, a computer-readable storage medium is not to be interpreted as a transient signal per se, such as a radio wave or other freely propagating electromagnetic wave, an electromagnetic wave propagating through a waveguide or other transmission medium (e.g., a light pulse through a fiber optic cable), or an electrical signal transmitted through a wire.

[0164] The computer-readable program instructions described herein can be downloaded from a computer-readable storage medium to each computing / processing device, or downloaded to an external computer or external storage device via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network. The network can include copper transmission cables, optical fiber transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. The network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in the computer-readable storage medium in each computing / processing device.

[0165] The computer program instructions for performing the operation of the present disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages, such as Smalltalk, C++, etc., and conventional procedural programming languages, such as "C" language or similar programming languages. Computer-readable program instructions may be executed completely on a user's computer, partially on a user's computer, as an independent software package, partially on a user's computer, partially on a remote computer, or completely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer via any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., using an Internet service provider to connect via the Internet). In some embodiments, an electronic circuit, such as a programmable logic circuit, a field programmable gate array (FPGA), or a programmable logic array (PLA), may be customized by utilizing the state information of the computer-readable program instructions, and the electronic circuit may execute the computer-readable program instructions, thereby realizing various aspects of the present disclosure.

[0166] Finally, 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 above embodiments, ordinary technicians in the relevant field should understand that the specific implementation methods of the present invention can still be modified or replaced by equivalents, and any modifications or equivalent replacements that do not depart from the spirit and scope of the present invention should be covered within the scope of protection of the claims of the present invention.

Claims

1. A smart grid energy storage fault monitoring method based on digital twin, characterized in that: include: Step 1, constructing a dynamic model of the energy storage system; Step 2, building a digital twin estimator based on the dynamic model built in step 1; Step 3, perform fault detection based on the estimator constructed in step 2; Step 4: perform fault isolation based on the fault detection result of step 3.

2. The method for monitoring smart grid energy storage faults based on digital twins according to claim 1, characterized in that: In step 1, the dynamic model is in the form of differential equations: Among them, F(x) represents the state matrix, which describes the linear relationship within the system; G(u,x) represents the nonlinear function of input and state; Eξ represents the external interference term; u is the system voltage; is the state variable; p cp , ω cp ,p sm represent the cathode pressure, nitrogen pressure, compressor rotation speed and supply pipe pressure at atmospheric temperature respectively.

3. The method for monitoring smart grid energy storage faults based on digital twins according to claim 2, characterized in that: In step 1, the state matrix F, the nonlinear function of input and state G(u,x), and the external interference term Eξ are: E=[-b7 000] T X=x1+x2+b2 f2(x3,x4)=W cp Among them, W ca,out is cathode gas flow; b i , i∈{1,...,17} is a constant obtained from the physical constants of the system, expressed as: Among them, W cp is the compressor flow rate; R is the universal gas constant; T fc is the temperature of the fuel cell; k ca,in is the cathode inlet constant; is the oxygen mass fraction at the air inlet; is the molar mass of oxygen; V ca is the cathode volume; ω atm is the relative humidity of the ambient air; p sat is the supply pipe saturation pressure at atmospheric temperature; is the molar mass of nitrogen; M v,ca is the steam molar mass; n is the number of cells in the fuel cell stack; F is the Faraday constant; k v ,k t ,R cm are all motor constants; η cm is the motor mechanical efficiency; J cp is the compressor inertia; C p is the constant pressure specific heat of air; T atm is the ambient temperature; η cp is the compressor efficiency; p atm is the atmospheric pressure; γ is the specific heat ratio of air; M a is the molar mass of air; V sm is the supply manifold volume; The output y(t) of the energy storage system is expressed as: Among them, the oxygen excess ratio Indicates the oxygen consumption in the air supply system; V st(t) is the stack voltage; I cm (t) and ω cp(t) Respectively represent the current and speed of the compressor; The performance variables of the energy storage system are expressed as: Among them, P net (t) represents the power of the battery.

4. The method for monitoring smart grid energy storage faults based on digital twins according to claim 3 is characterized in that: In step 2, a sliding mode observer is established in the digital twin estimator model based on the super-helical sliding mode control algorithm; the dynamic model of the sliding mode observer is: Where f(t) represents the error signal, D(y(t),u) is a smooth and bounded function that depends on the input and output of the system, and g(z(t),u) is Lipschitz continuous; The output of the sliding mode observer is: The performance variables of the sliding mode observer are expressed as:

5. The method for monitoring smart grid energy storage faults based on digital twins according to claim 4 is characterized in that: The dynamic model of the sliding mode observer is: Among them, v is the sliding mode error injection term, k1, k2 are gain functions.

6. The method for monitoring smart grid energy storage faults based on digital twins according to claim 5 is characterized in that: In step 3, fault detection uses residual detection, and the residual calculation formula is: Under sliding mode motion, the dynamic equation of the residual is: in, 7. The method for monitoring smart grid energy storage faults based on digital twins according to claim 6 is characterized in that: When γ(t)=0 and When , the energy storage system is in sliding mode motion; Under sliding mode motion, if e2+Δg=0, no fault occurs, and γ(t) and remains stable; otherwise, it means that the energy storage system fails, and γ(t) and fluctuation.

8. The method for monitoring smart grid energy storage faults based on digital twins according to claim 7 is characterized in that: In step 4, fault isolation is performed by analyzing the fault detection residual γ(t) and the fault The relationship between the residual and the fault is defined in the Cartesian product, where FSM is the theoretical fault feature matrix. Fault isolation is achieved by comparing the residuals corresponding to different faults in FSM. Fault The calculation formula is:

9. The method for monitoring smart grid energy storage faults based on digital twins according to claim 7, characterized in that: In step 4, the sensitivity analysis method of residual error to fault is adopted; The sensitivity is: Based on the sensitivity, the relative sensitivity function is established, and the calculation formula is: Through the relative sensitivity function, a theoretical relative fault sensitivity matrix is ​​established. The values ​​in this matrix are the fault f obtained by statistical methods. j Theoretical value when it occurs Faults are identified by Euclidean distance, and the calculation formula is: in, For identified faults.

10. An energy storage fault monitoring system, using the smart grid energy storage fault monitoring method based on digital twins according to any one of claims 1 to 9, characterized in that: include: Dynamic model building module, digital twin estimator building module, fault detection module and fault isolation module; Among them, the dynamic model building module is used to build a system dynamic model to describe the state changes of the system; The digital twin estimator building module is used to build a digital twin estimator; The fault detection module is used to determine whether a fault occurs in the system; The fault isolation module is used to achieve fault isolation by comparing the residuals corresponding to different faults.

Images