WO2023236393A1 - 一种应用于燃气轮机组的故障预警方法、系统和装置 - Google Patents

一种应用于燃气轮机组的故障预警方法、系统和装置 Download PDF

Info

Publication number
WO2023236393A1
WO2023236393A1 PCT/CN2022/121824 CN2022121824W WO2023236393A1 WO 2023236393 A1 WO2023236393 A1 WO 2023236393A1 CN 2022121824 W CN2022121824 W CN 2022121824W WO 2023236393 A1 WO2023236393 A1 WO 2023236393A1
Authority
WO
WIPO (PCT)
Prior art keywords
kriging
basis function
function
basis
gas turbine
Prior art date
Application number
PCT/CN2022/121824
Other languages
English (en)
French (fr)
Inventor
张坤
李红仁
自平洋
李炜
孙亮
Original Assignee
华电电力科学研究院有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 华电电力科学研究院有限公司 filed Critical 华电电力科学研究院有限公司
Priority to US18/313,525 priority Critical patent/US11860619B2/en
Publication of WO2023236393A1 publication Critical patent/WO2023236393A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/17—Mechanical parametric or variational design
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/15—Correlation function computation including computation of convolution operations
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G06F30/25—Design optimisation, verification or simulation using particle-based methods
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2111/00—Details relating to CAD techniques
    • G06F2111/08—Probabilistic or stochastic CAD
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
    • G06F2119/02—Reliability analysis or reliability optimisation; Failure analysis, e.g. worst case scenario performance, failure mode and effects analysis [FMEA]

Definitions

  • the present application relates to the field of computer technology, and in particular to a fault warning method, system and device applied to a gas turbine unit.
  • Gas turbines have the advantages of high thermal efficiency, stable operation, small size, high power, fast start-up and low pollution. In addition, they are relatively safer than traditional power equipment in actual production and operation. They are a new generation core power device after steam turbines and internal combustion engines. . The application scope of gas turbines is getting wider and wider. Since its birth, gas turbines have been widely used in aviation, shipbuilding, power generation and other fields. With the massive development of global natural gas resources, the growing demand for power grid peak shaving, and the rapid development of distributed energy systems, gas turbines are playing an increasingly important role in my country's power generation field. Therefore, gas turbines will serve as the core power equipment for efficient energy conversion and clean utilization for a long time to come.
  • Embodiments of the present application provide a fault warning method, system and device applied to a gas turbine unit, so as to at least solve the problem of a large number of false alarms in gas turbine fault warning in related technologies.
  • inventions of the present application provide a fault early warning method applied to a gas turbine unit.
  • the method includes:
  • the fault early warning of the gas turbine unit is performed through the error-compensated mechanism model.
  • selecting the optimal Kriging basis function from the several Kriging basis functions according to the error matrix includes:
  • the error matrix is respectively fitted by the several Kriging basis functions, and the Kriging basis functions are screened according to several fitting results to obtain a first basis function set;
  • calculating the reliability indicators of several Kriging basis functions in the first basis function set, and selecting the best Kriging basis function according to the reliability indicators includes:
  • a second set of basis functions whose convergence rate is higher than a preset threshold is selected, and then the optimal Kriging basis function is selected from the second set of basis functions.
  • selecting the optimal Kriging basis function from the second set of basis functions includes:
  • the optimal Kriging basis function is selected from the second basis function set according to the curve morphological characteristics.
  • selecting the optimal Kriging basis function from the second basis function set according to the curve morphological characteristics includes:
  • the morphological characteristics of the curve determine whether the corresponding Kriging basis function has similar characteristics to the Schweifer function or the trigonometric function;
  • the Tanimoto similarity between the Kriging basis function and the trigonometric function is calculated through the Tanimoto similarity function
  • the optimal Kriging basis function is selected from the second basis function set according to the Tanimoto similarity.
  • screening out the second set of basis functions whose convergence rate is higher than a preset threshold according to the reliability index includes:
  • the optimal calculation algorithm is selected from the Monte Carlo algorithm, particle swarm algorithm, response surface algorithm and one-dimensional quadratic matrix algorithm;
  • a second set of basis functions whose convergence rate is higher than the preset threshold is screened out.
  • constructing several preset Kriging basis functions includes:
  • the input parameters of the mechanism model are used as input parameters of the kriging basis function, and the prediction parameters of the mechanism model are used as the output parameters of the kriging basis function, and then several preset kriging basis functions are constructed.
  • calculating the prediction data of the prediction parameters in the gas turbine unit through the mechanism model includes:
  • the predicted parameters include first axis displacement, second axis displacement, third axis displacement, high pressure cylinder expansion, medium and medium pressure cylinder differential expansion, low pressure cylinder expansion and high pressure main steam pressure.
  • inventions of the present application provide a fault warning system applied to a gas turbine unit.
  • the system includes a data acquisition module, an error compensation module and a fault warning module;
  • the data acquisition module is used to calculate the prediction data of the prediction parameters in the gas turbine unit through the mechanism model, compare the prediction data with the real data of the prediction parameters, and obtain an error matrix;
  • the error compensation module is used to construct several Kriging basis functions according to the mechanism model; according to the error matrix, select the optimal Kriging basis function from the several Kriging basis functions, using the The optimal Kriging basis function performs error compensation on the mechanism model;
  • the fault early warning module is used to perform a fault early warning of the gas turbine unit through the error compensated mechanism model.
  • embodiments of the present application provide an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor.
  • the processor executes the computer program Implement the fault early warning method applied to the gas turbine unit as described in the first aspect above.
  • the embodiments of the present application provide a fault warning method, system and device applied to a gas turbine unit.
  • the prediction data of the prediction parameters in the gas turbine unit are calculated through a mechanism model, and the prediction data and the real data of the prediction parameters are compared. Compare the data to obtain the error matrix; construct several Kriging basis functions based on the mechanism model; select the optimal Kriging basis function from several Kriging basis functions based on the error matrix, and use the optimal Kriging basis function to pair the mechanism
  • the model performs error compensation; the fault early warning of the gas turbine unit is carried out through the error compensated mechanism model. It solves the problem of a large number of false alarms in existing gas turbine fault warnings. After error compensation is performed on the mechanism model based on the improved Kriging basis function, the fault warning of the gas turbine unit is realized, and the false alarm rate of the fault warning of the mechanism model is reduced. .
  • Figure 1 is a step flow chart of a fault early warning method applied to a gas turbine unit according to an embodiment of the present application
  • Figure 2 is a structural block diagram of a fault early warning system applied to a gas turbine unit according to an embodiment of the present application
  • Figure 3 is a schematic diagram of the internal structure of an electronic device according to an embodiment of the present application.
  • an embodiment means that a particular feature, structure or characteristic described in connection with the embodiment may be included in at least one embodiment of the application.
  • the appearances of this phrase in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments. It is explicitly and implicitly understood by those of ordinary skill in the art that the embodiments described in this application may be combined with other embodiments without conflict.
  • Words such as “connected”, “connected”, “coupled” and the like mentioned in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect.
  • the “plurality” mentioned in this application refers to two or more than two.
  • “And/or” describes the association of related objects, indicating that three relationships can exist. For example, “A and/or B” can mean: A exists alone, A and B exist simultaneously, and B exists alone.
  • the character “/” generally indicates that the related objects are in an “or” relationship.
  • the terms “first”, “second”, “third”, etc. used in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.
  • FIG. 1 is a step flow chart of a fault early warning method applied to a gas turbine unit provided according to an embodiment of the present application. As shown in Figure 1, the method includes the following step:
  • Step S102 calculate the prediction data of the prediction parameters in the gas turbine unit through the mechanism model, compare the prediction data with the real data of the prediction parameters, and obtain the error matrix;
  • the input data of the gas turbine unit is input into the mechanism model, and the prediction data of the prediction parameters in the gas turbine unit are calculated.
  • the input parameters corresponding to the input data include gas turbine speed, generator power, ambient temperature and environmental humidity.
  • the prediction parameters are Including the first axis displacement, the second axis displacement, the third axis displacement, high pressure cylinder expansion, medium and medium pressure cylinder differential expansion, low pressure cylinder expansion and high pressure main steam pressure.
  • the pressure cylinder differential expansion, low-pressure cylinder expansion, and high-pressure main steam pressure 1 are used as output parameters (prediction parameters).
  • the prediction parameters are set to G 1 , G 2 , G 3 , G 4 , G 5 , G 6 and G7 .
  • the gas turbine speed goes through 0 to 3000 revolutions during the starting process.
  • Table 1 is a real data table of predicted parameters in a heavy-duty gas turbine unit.
  • the mechanism model is also called the white box model.
  • An accurate mathematical model established based on the internal mechanism of the object, the production process, or the transfer mechanism of the material flow. It is a mathematical model of an object or process based on the mass balance equation, energy balance equation, momentum balance equation, phase balance equation, and certain physical property equations, chemical reaction laws, basic circuit laws, etc.
  • the advantage of the mechanism model is that the parameters have very clear physical meaning. The model parameters are easy to adjust and the resulting model has strong adaptability.
  • Step S104 construct several Kriging basis functions according to the mechanism model
  • the input parameters of the mechanism model are used as the input parameters of the kriging basis function
  • the prediction parameters of the mechanism model are used as the output parameters of the kriging basis function, and then several preset kriging basis functions are constructed.
  • Common kriging custom functions include the following:
  • X 1 , X 2 ,..., X n are the coordinate positions within the input sample parameters, is the fluctuation center value, ⁇ x 1 , ⁇ x 2 ,..., ⁇ x n are the fluctuation ranges. Simplify the change amount of the input parameters and simplify the basis function to:
  • ⁇ F is the function output value
  • ⁇ 2 ( ⁇ x) is the high-order derivative of the input parameter fluctuation.
  • w is the weight coefficient, which shows the importance of each point in the total amount.
  • x il and x ih respectively represent the upper and lower limits of the input quantity in the i-th component.
  • G(x 1 ,x 2 ...x n ) is the interpolation function about x 1 ,x 2 ...x n .
  • the data change trend of each component of the heavy-duty gas turbine unit studied by this method has a certain relationship with changes in power generation and gas turbine speed. Therefore, an interpolation function is needed to satisfy the boundary constraints and at the same time reflect the functional relationship between the input characteristics and the output characteristics. Substitute G(x 1 ,x 2 ...x n ) into equations (3) and (4) to get
  • ⁇ i is the unit output power, gas turbine speed and other parameters. Determine the definition domain of the gas turbine speed [0,3000]. Taking G m as an example, 20 verification points are selected within the fluctuation range of the gas turbine speed.
  • the weight coefficient model is obtained according to equations (7) and (8).
  • ⁇ i can be calculated by using the particle swarm algorithm hyperparameter optimization method.
  • Step S106 select the optimal Kriging basis function from several Kriging basis functions according to the error matrix
  • the first step is to fit the error matrix separately through several Kriging basis functions, and filter the Kriging basis functions according to several fitting results to obtain the first basis function set.
  • step two is to calculate the reliability index of the Kriging basis function in the first basis function set through the Monte Carlo algorithm, particle swarm algorithm, response surface algorithm and one-dimensional quadratic matrix algorithm respectively; filter out based on the reliability index A second set of basis functions with a convergence rate higher than a preset threshold.
  • step three is to calculate the derivative of the Kriging basis function in the second basis function set, and determine the curve shape characteristics of the Kriging basis function based on the derivative.
  • the curve shape characteristics include single peak maximum characteristics, single peak Minimum features and multi-peak features;
  • the morphological characteristics of the curve determine whether the corresponding Kriging basis function has similar characteristics to the Schweifer function or trigonometric function; if it has similar characteristics to the Schweifer function, calculate the Kriging basis function through the Tanimoto similarity function Tanimoto similarity with the Schweifer function; if it has similar characteristics with the trigonometric function, calculate the Tanimoto similarity between the Kriging basis function and the trigonometric function through the Tanimoto similarity function; according to the Tanimoto similarity, from the second basis function set Screen out the optimal Kriging basis function.
  • ⁇ G m is used as the error term between the real data of the gas turbine unit and the predicted data calculated by the mechanism model.
  • the gas turbine speed variable x, shaft displacement 1, shaft displacement 2, shaft displacement 3 and other parameters are the dependent variables f(x).
  • the basis function is screened, so it can be determined which form of change curve is consistent with ⁇ G m .
  • the exponential change curve does not match ⁇ G m .
  • other Kriging basis functions will be screened in the next step (the first basis function set is obtained).
  • step two it is usually impossible to screen out the optimal Kriging basis function at one time through step one.
  • the function based on the method of selecting sample accumulation, Monte Carlo, particle swarm, response Surface and one-dimensional quadratic matrix methods to calculate the reliability index of the product function.
  • the Monte Carlo method is used to calculate the failure probability, and the reliability index is obtained based on the results. This index is used as a standard to compare the basis function reliability index calculated by other methods.
  • Table 2 shows the calculation results of the reliability index of the basis function model for ⁇ G i .
  • the basis functions are Gaussian model, Fourier model, trigonometric function and polynomial model respectively, comparing the Monte Carlo algorithm, particle swarm algorithm, response surface algorithm and one-dimensional quadratic matrix algorithm, it is concluded that the particle swarm method can achieve the minimum
  • the number of function calculations reaches the reliability index (i.e., the optimal calculation algorithm).
  • the second basis function set whose convergence rate is higher than the preset threshold is screened out from each model.
  • the curve in order to further improve the curve fitness, the curve needs to be discriminated before error correction to select an appropriate Kriging basis function.
  • Each basis function of the Kriging interpolation method has different curve characteristics and its operating efficiency is different, so it can be judged in advance based on the function characteristics.
  • the curve ⁇ G m has the characteristics of multi-peak, single-peak maximum or single-peak minimum.
  • f(x) is the maximum value at point a.
  • f(x) is the minimum value at point a.
  • the function graph can be judged to be a multi-peak shape.
  • the extreme values of scatter data can be identified by derivation of the function, but the specific form of the scatter data cannot be specifically identified.
  • a benchmark function is introduced to perform correlation analysis on the scatter point data. If the correlation between the benchmark function curve and the scatter point data distribution is high, then the scatter point data The Kriging basis function corresponding to this basis function can be used.
  • the Schweifer graph has the trend characteristic of nonlinear multi-wave peaks gradually amplifying.
  • the basic types of Schweifer functions are as follows:
  • equation (24) can be modified:
  • the Tanimoto similarity function is as follows:
  • the optimal Kriging basis function for the corresponding ⁇ G curve is selected from the second basis function set.
  • Step S108 Use the optimal Kriging basis function to perform error compensation on the mechanism model, and use the error-compensated mechanism model to provide a fault warning for the gas turbine unit.
  • k 1, 2, 3,...n.
  • q represents the input sample dimension.
  • ⁇ G p 00 +p 10 r+p 01 f 3 (x 1 ,x 2 ...x q )+p 20 r+p 11 rf (x 1 ,x 2 ...x q ) (33)
  • r is the gas turbine speed
  • f Er (r,x 1 ,x 2 ...x q ) is the error compensation function of the gas turbine speed sampling point change.
  • w i (x) is the error-corrected mechanism model calculation data. Compared with q i (x), w i (x) is closer to the measured data.
  • this method can perform error compensation on the mechanism model, thereby reducing false alarms.
  • FIG. 2 is a structural block diagram of a fault early warning system applied to a gas turbine unit provided according to an embodiment of the present application. As shown in Figure 2, the system includes data collection Module 21, error compensation module 22 and fault warning module 23;
  • the data acquisition module 21 is used to calculate the prediction data of the prediction parameters in the gas turbine unit through the mechanism model, compare the prediction data with the real data of the prediction parameters, and obtain the error matrix;
  • the error compensation module 22 is used to construct several Kriging basis functions according to the mechanism model; select the optimal Kriging basis function from several Kriging basis functions according to the error matrix, and use the optimal Kriging basis function to Mechanism model performs error compensation;
  • the fault early warning module 23 is used to provide a fault early warning for the gas turbine unit through the error compensated mechanism model.
  • the error compensation module 22 and the fault warning module 23 in the embodiment of the present application the problem of a large number of fault false alarms existing in the existing gas turbine fault warning is solved, and the mechanism model based on the improved Kriging basis function is realized. After error compensation is carried out, the fault warning of the gas turbine unit reduces the false alarm rate of the fault warning of the mechanism model.
  • each of the above modules can be a functional module or a program module, and can be implemented by software or hardware.
  • each of the above-mentioned modules can be located in the same processor; or the above-mentioned modules can also be located in different processors in any combination.
  • This embodiment also provides an electronic device, including a memory and a processor.
  • a computer program is stored in the memory, and the processor is configured to run the computer program to perform the steps in any of the above method embodiments.
  • the above-mentioned electronic device may further include a transmission device and an input-output device, wherein the transmission device is connected to the above-mentioned processor, and the input-output device is connected to the above-mentioned processor.
  • the embodiment of the present application can provide a storage medium for implementation.
  • the storage medium stores a computer program; when the computer program is executed by the processor, any one of the fault warning methods in the above embodiments applied to the gas turbine unit is implemented.
  • a computer device which may be a terminal.
  • the computer equipment includes a processor, memory, network interface, display screen and input device connected by a system bus.
  • the processor of the computer device is used to provide computing and control capabilities.
  • the memory of the computer device includes non-volatile storage media and internal memory.
  • the non-volatile storage medium stores operating systems and computer programs. This internal memory provides an environment for the execution of operating systems and computer programs in non-volatile storage media.
  • the network interface of the computer device is used to communicate with external terminals through a network connection.
  • the computer program is executed by a processor to implement a fault early warning method applied to the gas turbine unit.
  • the display screen of the computer device may be a liquid crystal display or an electronic ink display.
  • the input device of the computer device may be a touch layer covered on the display screen, or may be a button, trackball or touch pad provided on the computer device shell. , it can also be an external keyboard, trackpad or mouse, etc.
  • Figure 3 is a schematic diagram of the internal structure of an electronic device according to an embodiment of the present application.
  • an electronic device is provided.
  • the electronic device can be a server, and its internal structure diagram can be as shown in Figure 3 shown.
  • the electronic device includes a processor, a network interface, an internal memory and a non-volatile memory connected through an internal bus, wherein the non-volatile memory stores an operating system, a computer program and a database.
  • the processor is used to provide computing and control capabilities
  • the network interface is used to communicate with external terminals through a network connection
  • the internal memory is used to provide an environment for the operation of the operating system and computer programs.
  • an application A fault early warning method for gas turbine units, and a database is used to store data.
  • FIG. 3 is only a block diagram of a partial structure related to the solution of the present application, and does not constitute a limitation on the electronic equipment to which the solution of the present application is applied.
  • Specific electronic devices may May include more or fewer parts than shown, or combine certain parts, or have a different arrangement of parts.
  • Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
  • Volatile memory may include random access memory (RAM) or external cache memory.
  • RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Synchlink DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
  • SRAM static RAM
  • DRAM dynamic RAM
  • SDRAM synchronous DRAM
  • DDRSDRAM double data rate SDRAM
  • ESDRAM enhanced SDRAM
  • SLDRAM synchronous chain Synchlink DRAM
  • Rambus direct RAM
  • DRAM direct memory bus dynamic RAM
  • RDRAM memory bus dynamic RAM

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Geometry (AREA)
  • General Engineering & Computer Science (AREA)
  • Pure & Applied Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Mathematical Physics (AREA)
  • Evolutionary Computation (AREA)
  • Mathematical Optimization (AREA)
  • Mathematical Analysis (AREA)
  • Computational Mathematics (AREA)
  • Computing Systems (AREA)
  • Software Systems (AREA)
  • Computer Hardware Design (AREA)
  • Artificial Intelligence (AREA)
  • Databases & Information Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Algebra (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本申请涉及一种应用于燃气轮机组的故障预警方法、系统和装置，其中，该方法包括：通过机理模型计算燃气轮机组中预测参数的预测数据，将预测数据和预测参数的真实数据进行数据对比，得到误差矩阵；根据机理模型构建若干克里金基函数；根据误差矩阵，从若干克里金基函数中筛选出最优克里金基函数，采用最优克里金基函数对机理模型进行误差补偿；通过误差补偿后的机理模型进行燃气轮机组的故障预警。通过本申请，解决了现有燃气轮机故障预警存在的大量故障误报的问题，实现了基于改进克里金基函数对机理模型进行误差补偿后，燃气轮机组的故障预警，降低了机理模型对故障预警的误报率。

Description

一种应用于燃气轮机组的故障预警方法、系统和装置 技术领域
本申请涉及计算机技术领域，特别是涉及一种应用于燃气轮机组的故障预警方法、系统和装置。
背景技术
燃气轮机有热效率高、运行平稳、体积小、功率大、启动快和污染小等优点，而且在实际生产运行中相比传统动力设备相对安全，是作为继蒸汽轮机和内燃机之后的新一代核心动力装置。燃气轮机的应用范围越来越广，自诞生以来，燃气轮机已经广泛地应用到了航空、船舶和发电等领域。随着全球天然气资源的大量开发、电网调峰日益增长的需求以及分布式能源系统的飞速发展，燃气轮机在我国发电领域的地位越来越重要。因此，燃气轮机作为今后很长时间内能源高效转化和清洁利用的核心动力装备。
为确保燃气轮机能持续安全稳定可靠的长时间运行，其运行状态越来越成为重要的研究内容。燃气轮机组结构复杂，其部件通常工作在高温、高压和高速旋转的极端条件下，对于调峰燃汽轮机组，还会频繁启停，因此，燃气轮机组容易在运行状态中发生故障损伤，影响电网保供任务。传统的故障预警诊断模型存在大量的故障误报的问题，导致真实的故障信息被掩盖从而丧失预警的作用。
目前针对相关技术中燃气轮机故障预警存在的大量故障误报的问题，尚未提出有效的解决方案。
发明内容
本申请实施例提供了一种应用于燃气轮机组的故障预警方法、系统和装置，以至少解决相关技术中燃气轮机故障预警存在的大量故障误报的问题。
第一方面，本申请实施例提供了一种应用于燃气轮机组的故障预警方法，所述方法包括：
通过机理模型，计算燃气轮机组中预测参数的预测数据，将所述预测数据和所述预测参数的真实数据进行数据对比，得到误差矩阵；
根据所述机理模型，构建若干克里金基函数；
根据所述误差矩阵，从所述若干克里金基函数中筛选出最优克里金基函数， 采用所述最优克里金基函数对所述机理模型进行误差补偿；
通过所述误差补偿后的机理模型进行所述燃气轮机组的故障预警。
在其中一些实施例中，根据所述误差矩阵，从所述若干克里金基函数中筛选出最优克里金基函数包括：
通过所述若干克里金基函数对所述误差矩阵分别进行拟合，根据若干拟合结果对所述克里金基函数进行筛选，得到第一基函数集合；
计算所述第一基函数集合中若干克里金基函数的可靠性指标，根据所述可靠性指标筛选出最佳克里金基函数。
在其中一些实施例中，计算所述第一基函数集合中若干克里金基函数的可靠性指标，根据所述可靠性指标筛选出最佳克里金基函数包括：
通过蒙特卡洛算法、粒子群算法、响应面算法和一元二次矩阵算法，分别计算所述第一基函数集合中克里金基函数的可靠性指标；
根据所述可靠性指标筛选出收敛速率高于预设阈值的第二基函数集合，进而从所述第二基函数集合中筛选出最优克里金基函数。
在其中一些实施例中，进而从所述第二基函数集合中筛选出最优克里金基函数包括：
计算所述第二基函数集合中克里金基函数的导数，根据所述导数判断所述克里金基函数的曲线形态特征，其中，所述曲线形态特征包括单峰极大值特征、单峰极小值特征和多峰特征；
根据所述曲线形态特征从所述第二基函数集合中筛选出最优克里金基函数。
在其中一些实施例中，根据所述曲线形态特征从所述第二基函数集合中筛选出最优克里金基函数包括：
根据所述曲线形态特征，判断对应的克里金基函数是否与施韦费尔函数或三角函数具有相似特征；
若与施韦费尔函数具有相似特征，则通过谷本相似度函数计算所述克里金基函数与施韦费尔函数的谷本相似度；
若与三角函数具有相似特征，则通过谷本相似度函数计算所述克里金基函数与三角函数的谷本相似度；
根据所述谷本相似度从所述第二基函数集合中筛选出最优克里金基函数。
在其中一些实施例中，根据所述可靠性指标筛选出收敛速率高于预设阈值的第二基函数集合包括：
根据克里金基函数的可靠性指标的函数计算次数，从蒙特卡洛算法、粒子 群算法、响应面算法和一元二次矩阵算法中选取出最优计算算法；
根据所述最优计算算法计算得到的可靠性指标，筛选出收敛速率高于预设阈值的第二基函数集合。
在其中一些实施例中，根据所述机理模型，构建若干预设克里金基函数包括：
将所述机理模型的输入参数作为克里金基函数的输入参数，将所述机理模型的预测参数作为克里金基函数的输出参数，进而构建若干预设克里金基函数。
在其中一些实施例中，通过机理模型，计算燃气轮机组中预测参数的预测数据包括：
将燃气轮机组的输入数据输入到机理模型中，计算所述燃气轮机组中预测参数的预测数据，其中，所述输入数据对应的输入参数包括燃机转速、发电机功率、环境温度和环境湿度，所述预测参数包括第一轴位移、第二轴位移、第三轴位移、高压缸膨胀、高中压缸差胀、低压缸膨胀和高压主汽压力。
第二方面，本申请实施例提供了一种应用于燃气轮机组的故障预警系统，所述系统包括数据采集模块、误差补偿模块和故障预警模块；
所述数据采集模块，用于通过机理模型，计算燃气轮机组中预测参数的预测数据，将所述预测数据和所述预测参数的真实数据进行数据对比，得到误差矩阵；
所述误差补偿模块，用于根据所述机理模型，构建若干克里金基函数；根据所述误差矩阵，从所述若干克里金基函数中筛选出最优克里金基函数，采用所述最优克里金基函数对所述机理模型进行误差补偿；
所述故障预警模块，用于通过所述误差补偿后的机理模型进行所述燃气轮机组的故障预警。
第三方面，本申请实施例提供了一种电子装置，包括存储器、处理器以及存储在所述存储器上并可在所述处理器上运行的计算机程序，所述处理器执行所述计算机程序时实现如上述第一方面所述的应用于燃气轮机组的故障预警方法。
相比于相关技术，本申请实施例提供的一种应用于燃气轮机组的故障预警方法、系统和装置，通过机理模型计算燃气轮机组中预测参数的预测数据，将预测数据和预测参数的真实数据进行数据对比，得到误差矩阵；根据机理模型构建若干克里金基函数；根据误差矩阵，从若干克里金基函数中筛选出最优克里金基函数，采用最优克里金基函数对机理模型进行误差补偿；通过误差补偿 后的机理模型进行燃气轮机组的故障预警。解决了现有燃气轮机故障预警存在的大量故障误报的问题，实现了基于改进克里金基函数对机理模型进行误差补偿后，燃气轮机组的故障预警，降低了机理模型对故障预警的误报率。
附图说明
此处所说明的附图用来提供对本申请的进一步理解，构成本申请的一部分，本申请的示意性实施例及其说明用于解释本申请，并不构成对本申请的不当限定。在附图中：
图1是根据本申请实施例提供的应用于燃气轮机组的故障预警方法的步骤流程图；
图2是根据本申请实施例提供的应用于燃气轮机组的故障预警系统的结构框图；
图3是根据本申请实施例的电子设备的内部结构示意图。
附图说明：21、数据采集模块；22、误差补偿模块；23、故障预警模块。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行描述和说明。应当理解，此处所描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。基于本申请提供的实施例，本领域普通技术人员在没有作出创造性劳动的前提下所获得的所有其他实施例，都属于本申请保护的范围。
显而易见地，下面描述中的附图仅仅是本申请的一些示例或实施例，对于本领域的普通技术人员而言，在不付出创造性劳动的前提下，还可以根据这些附图将本申请应用于其他类似情景。此外，还可以理解的是，虽然这种开发过程中所作出的努力可能是复杂并且冗长的，然而对于与本申请公开的内容相关的本领域的普通技术人员而言，在本申请揭露的技术内容的基础上进行的一些设计，制造或者生产等变更只是常规的技术手段，不应当理解为本申请公开的内容不充分。
在本申请中提及“实施例”意味着，结合实施例描述的特定特征、结构或特性可以包含在本申请的至少一个实施例中。在说明书中的各个位置出现该短语并不一定均是指相同的实施例，也不是与其它实施例互斥的独立的或备选的实施例。本领域普通技术人员显式地和隐式地理解的是，本申请所描述的实施 例在不冲突的情况下，可以与其它实施例相结合。
除非另作定义，本申请所涉及的技术术语或者科学术语应当为本申请所属技术领域内具有一般技能的人士所理解的通常意义。本申请所涉及的“一”、“一个”、“一种”、“该”等类似词语并不表示数量限制，可表示单数或复数。本申请所涉及的术语“包括”、“包含”、“具有”以及它们任何变形，意图在于覆盖不排他的包含；例如包含了一系列步骤或模块(单元)的过程、方法、系统、产品或设备没有限定于已列出的步骤或单元，而是可以还包括没有列出的步骤或单元，或可以还包括对于这些过程、方法、产品或设备固有的其它步骤或单元。本申请所涉及的“连接”、“相连”、“耦接”等类似的词语并非限定于物理的或者机械的连接，而是可以包括电气的连接，不管是直接的还是间接的。本申请所涉及的“多个”是指两个或两个以上。“和/或”描述关联对象的关联关系，表示可以存在三种关系，例如，“A和/或B”可以表示：单独存在A，同时存在A和B，单独存在B这三种情况。字符“/”一般表示前后关联对象是一种“或”的关系。本申请所涉及的术语“第一”、“第二”、“第三”等仅仅是区别类似的对象，不代表针对对象的特定排序。
本申请实施例提供了一种应用于燃气轮机组的故障预警方法，图1是根据本申请实施例提供的应用于燃气轮机组的故障预警方法的步骤流程图，如图1所示，该方法包括以下步骤：
步骤S102，通过机理模型，计算燃气轮机组中预测参数的预测数据，将预测数据和预测参数的真实数据进行数据对比，得到误差矩阵；
具体地，将燃气轮机组的输入数据输入到机理模型中，计算燃气轮机组中预测参数的预测数据，其中，输入数据对应的输入参数包括燃机转速、发电机功率、环境温度和环境湿度，预测参数包括第一轴位移、第二轴位移、第三轴位移、高压缸膨胀、高中压缸差胀、低压缸膨胀和高压主汽压力。
优选地，以某型号重型燃汽轮机组为例，以燃机转速、发电机功率、大气温度、大气湿度等作为输入参数，将轴位移1、轴位移2、轴位移3、高压缸膨胀、高中压缸差胀、低压缸膨胀、高压主汽压力1(机侧)作为输出参数(预测参数)，将预测参数依次设为G 1、G 2、G 3、G 4、G 5、G 6和G 7。燃机转速在起机过程中经历0到3000转，表1是某重型燃汽轮机组中预测参数的真实数据表，如表1所示，以燃机转速3000转，20个采样点为例，得到燃气轮机组中预测参数的真实数据，将有真实数据与机理模型计算的预测数据进行对比(即将实测数据与预测数据相减)，所得结果存入误差矩阵Er。
表1
采样点 G 1 G 2 G 3 G 4 G 5 G 6 G 7 燃机转速
1 0.191 0.273 0.198 16.702 7.203 32.247 9.134 2997.8
2 0.185 0.269 0.193 16.704 7.226 32.258 9.081 2997.8
3 0.183 0.269 0.193 16.704 7.234 32.26 9.047 2999.3
4 0.183 0.271 0.194 16.702 7.26 32.27 9.084 3001.9
5 0.209 0.297 0.221 16.704 7.263 32.268 9.202 3001.9
6 0.201 0.289 0.212 16.702 7.257 32.262 9.322 3000.7
7 0.207 0.296 0.219 16.702 7.251 32.248 9.284 3001.9
8 0.219 0.308 0.231 16.7 7.239 32.238 9.316 3001.9
9 0.229 0.319 0.241 16.704 7.234 32.232 9.516 3001.9
10 0.215 0.306 0.227 16.7 7.237 32.227 9.734 2998.9
11 0.226 0.317 0.239 16.7 7.206 32.187 9.503 2997.8
12 0.231 0.323 0.244 16.7 7.207 32.18 9.759 3000.4
13 0.187 0.279 0.2 16.703 7.217 32.186 9.458 2999.6
14 0.239 0.331 0.252 16.704 7.197 32.164 9.575 2999.3
15 0.205 0.297 0.218 16.702 7.202 32.165 9.64 2998.9
16 0.209 0.301 0.223 16.701 7.18 32.138 9.563 2998.9
17 0.234 0.326 0.247 16.7 7.148 32.095 9.525 3000.4
18 0.216 0.309 0.23 16.699 7.149 32.1 9.566 2997.8
19 0.225 0.318 0.238 16.703 7.144 32.089 9.727 2998.1
20 0.223 0.316 0.237 16.704 7.134 32.07 9.719 3000
需要说明的是，机理模型，亦称白箱模型。根据对象、生产过程的内部机制或者物质流的传递机理建立起来的精确数学模型。它是基于质量平衡方程、能量平衡方程、动量平衡方程、相平衡方程以及某些物性方程、化学反应定律、电路基本定律等而获得对象或过程的数学模型。机理模型的优点是参数具有非常明确的物理意义。模型参数易于调整，所得的模型具有很强的适应性。
步骤S104，根据机理模型，构建若干克里金基函数；
具体地，将机理模型的输入参数作为克里金基函数的输入参数，将机理模型的预测参数作为克里金基函数的输出参数，进而构建若干预设克里金基函数。
优选地，为了提高机理模型的数据跟随能力，需要使用合适的基函数来提升克里金插值法的计算能力。常见的克里金自定义函数有以下几种：
傅里叶：
Figure PCTCN2022121824-appb-000001
多项式：
Figure PCTCN2022121824-appb-000002
指数：
Figure PCTCN2022121824-appb-000003
高斯：
Figure PCTCN2022121824-appb-000004
三角函数：
Figure PCTCN2022121824-appb-000005
为了构建自定义基函数，需要将燃气轮机组关键参数作为输入参数及观测参数作为输出值(即机理模型的输入参数和预测参数)，以此确定基函数的输出特性。定义函数k(x 1,x 2…x n)，则输入参数x 1,x 2…x n的定义域为：
Figure PCTCN2022121824-appb-000006
其中，X 1、X 2、…、X n为输入样本参数内的坐标位置，
Figure PCTCN2022121824-appb-000007
为波动中心值，Δx 1、Δx 2、…、Δx n为波动范围。将输入参数的改变量进行一定简化，将基函数简化为：
Figure PCTCN2022121824-appb-000008
其中，ΔF为函数输出值，Δ 2(Δx)为输入参数波动的高阶导数。
如果存在函数k(x 1,x 2…x n)对Δx 1、Δx 2、…、Δx n存在一定的函数关系，那么存在：
Figure PCTCN2022121824-appb-000009
其中，w为权重系数，显示各点在总量中所具有的重要程度。当所取点为节点时权重系数w＝1，当所取点非为节点时权重系数0<w<1。设定输入参数的边界条件，存在：
Figure PCTCN2022121824-appb-000010
其中、x il、x ih分别代表输入量在第i分量的上下限。将式(6)、(7)与(8)联立得到
Figure PCTCN2022121824-appb-000011
式中G(x 1,x 2…x n)为关于x 1,x 2…x n的插值函数。本方法研究的重型燃气轮机组各个部件数据变化趋势与发电功率和燃机转速等改变具有一定的变化关系，因此需要一个插值函数满足边界约束条件，同时能够体现输入特性与输出特性的函数关系。将G(x 1,x 2…x n)代入到式(3)与式(4)中得到
Figure PCTCN2022121824-appb-000012
Figure PCTCN2022121824-appb-000013
其中，d i＝|x-x i|，x为未知值。θ i为机组输出功率、燃机转速等参数。确定燃机转速的定义域[0,3000]。以G m为例，在燃机转速的波动范围区域内选取20个验证点。根据式(7)与(8)得到权重系数模型。通过使用粒子群算法超参数寻优的方法计算即可得到θ i。
步骤S106，根据误差矩阵，从若干克里金基函数中筛选出最优克里金基函数；
具体地，步骤一，通过若干克里金基函数对误差矩阵分别进行拟合，根据若干拟合结果对克里金基函数进行筛选，得到第一基函数集合。
进一步地，步骤二，通过蒙特卡洛算法、粒子群算法、响应面算法和一元二次矩阵算法，分别计算第一基函数集合中克里金基函数的可靠性指标；根据可靠性指标筛选出收敛速率高于预设阈值的第二基函数集合。
更进一步地，步骤三，计算第二基函数集合中克里金基函数的导数，根据导数判断克里金基函数的曲线形态特征，其中，曲线形态特征包括单峰极大值 特征、单峰极小值特征和多峰特征；
根据曲线形态特征，判断对应的克里金基函数是否与施韦费尔函数或三角函数具有相似特征；若与施韦费尔函数具有相似特征，则通过谷本相似度函数计算克里金基函数与施韦费尔函数的谷本相似度；若与三角函数具有相似特征，则通过谷本相似度函数计算克里金基函数与三角函数的谷本相似度；根据谷本相似度从第二基函数集合中筛选出最优克里金基函数。
优选地，在步骤一中，以ΔG m为燃汽轮机组的真实数据与机理模型计算的预测数据之间的误差项。以燃机转速变量x，轴位移1、轴位移2、轴位移3等参数为因变量f(x)。
对ΔG m进行多项式拟合，拟合公式如下：
f(x)＝p 1x 7+p 2x 6+p 3x 5+p 4x 4+p 5x 3+p 6x 2+p 7x+p 8    (13)
对上式进行求导，在定义域[0，3000]区间内df(x)/dx等于零的解有k 1个。
对ΔG m进行高斯拟合，则拟合得到的高斯函数为：
Figure PCTCN2022121824-appb-000014
对上式求导得到可得到df(x)/dx等于零的解有k 2个。
对ΔG m进行傅里叶拟合，则得到的傅里叶函数为：
Figure PCTCN2022121824-appb-000015
对上式求导，在定义域[0，3000]区间内df(x)/dx等于零的解有k 3个。
对ΔG m进行指数拟合，得到的函数为：
f(x)＝ae bx+ce dx         (16)
对上式求导，在定义域[0，3000]区间内df(x)/dx等于零的解有k 4个。
对ΔG m进行三角函数拟合，则得到的函数为：
Figure PCTCN2022121824-appb-000016
对上式求导，在定义域[0，3000]区间内df(x)/dx等于零的解有k 5个。
通过零解的个数k i(i＝1,2,3…n)与ΔG m曲线进行对比，对基函数进行筛选，因此可以确定哪种形式的变化曲线与ΔG m是否相符。例，指数形式的变化曲线与ΔG m并不相符。则其他克里金基函数会进行下一步筛选(得到第一基函数集合)。
优选地，在步骤二中，通过步骤一通常无法一次性筛选出最优克里金基函数，为了进一步进行函数精选，以选择样本累计的方法为基础，以蒙特卡洛、粒子群、响应面以及一元二次矩阵方法计算积函数的可靠性指标。以蒙特卡洛方法计算失效概率，根据结果得到可靠性指标，并以此指标为标准，针对其他方法计算的基函数可靠性指标进行对比。
以ΔG i为例，表2对于ΔG i的基函数模型可靠性指标计算结果表。通过将高斯模型、傅里叶模型与多项式模型进行对比，可得到在以高斯模型为基函数的前提下，使用粒子群方法能够以最少的函数计算次数达到可靠性指标。
表2
Figure PCTCN2022121824-appb-000017
在基函数分别为高斯模型、傅里叶模型、三角函数和多项式模型的情况下，比较蒙特卡洛算法、粒子群算法、响应面算法和一元二次矩阵算法，得出粒子群方法能够以最少的函数计算次数达到可靠性指标(即最优计算算法)，在粒子群方法的前提下，从各模型中筛选出收敛速率高于预设阈值的第二基函数集合。
优选地，在步骤三中，为了进一步提升曲线适应度，需要在误差修正前对曲线进行判别以选择合适的克里金基函数。克里金插值法各个基函数具有不同的曲线特征，同时其运算效率不尽相同，因此可根据函数特征进行提前判别。曲线ΔG m无非具有多峰、单峰极大或者单峰极小值的特征。为了进一步判别曲线的形态，需要对曲线函数进行一阶和二阶求导，依据求导结果对数据进行预 判。
误差数据预判导
为了对函数极值进行识别，设克里金插值基函数为f(x)，其中a＝(x 1,x 2,...x n)∈R n，则f(x)在定义域内连续可微。因此可得到：
Figure PCTCN2022121824-appb-000018
可得f(x)在点a处为极大值。
同理，若存在：
Figure PCTCN2022121824-appb-000019
可得f(x)在点a处为极小值。
根据f(x)为函数在x位置处的梯度，可得：
Figure PCTCN2022121824-appb-000020
若相邻的极值点x 1，x 2，…x n使得f(x)的一阶导为零，因为一阶导数的特性使得相邻两点x i和x i+1为局部最大或者最小值，因此采用相隔两点作为判断函数曲线的标准，如果存在相隔两点的函数值存在单调递增，且二阶导数相乘为正数，则可将该段函数图形判断为具有单峰极大值特征。即：
Figure PCTCN2022121824-appb-000021
若相邻两点的函数值单调递减，且二阶导数相乘为正数，则可将该段函数图形认为具有单峰极小值。即：
Figure PCTCN2022121824-appb-000022
若相邻两点的函数值存在单调递增或者递减，则可将该段函数图形判断为多波峰形状。通过对函数求导可判别散点数据极值，但无法具体识别散点数据 的具体形态。即：
Figure PCTCN2022121824-appb-000023
因函数求导判别的方法无法处理散点数据分布复杂的情况，因此引入基准函数对散点数据进行相关度分析，若基准函数曲线与散点数据分布相关度较高时，则该散点数据可使用该基准函数对应的克里金基函数。
通过引入施韦费尔函数图形和三角函数图形作为散点数据曲线趋势判别方法曲线。施韦费尔图形具有非线性多波峰逐级放大的趋势特征。施韦费尔函数基本类型如下：
Figure PCTCN2022121824-appb-000024
若ΔG 3函数曲线与施韦费尔函数曲线具有类似的特征，可对式(24)修改：
Figure PCTCN2022121824-appb-000025
通过使用谷本系数验证两组曲线的相似度，其谷本相似度函数如下：
Figure PCTCN2022121824-appb-000026
根据式(25)(26)可得到施韦费尔与ΔG 3的相似度为0.77。同理，使用改进的施韦费尔函数与ΔG 4进行对比，谷本相似度为0.8。针对ΔG 4改进的施韦费尔函数公式如下：
Figure PCTCN2022121824-appb-000027
若ΔG 1函数曲线与三角函数曲线具有类似的特征，则针对ΔG 1改进的三角函数f(x)公式如下：
f(x)＝-0.25cos(3.5x-0.5)+0.05               (28)
可得到三角函数与ΔG 1的相似度为0.86。同理，使用改进的三角函数与ΔG 2进行对比，谷本相似度为0.92。则针对ΔG 2改进的三角函数公式如下：
f(x)＝1.5cos(5x-12)-1.2       (29)
在所有ΔG k曲线进行判断比较后，得出ΔG 1、ΔG 2、ΔG 5、ΔG 6和ΔG 7与三角函数的谷本相似度超过0.7，ΔG 3和ΔG 4与施韦费尔函数的谷本相似度超过0.7。根据ΔG曲线与三角函数、施韦费尔函数的谷本相似度，从第二基函数集合中为 对应ΔG曲线筛选出最优克里金基函数。例如，与施韦费尔函数的谷本相似度超过0.7的曲线，其克里金插值基函数采用高斯模型，与三角函数的谷本相似度超过0.7的曲线，其克里金插值基函数采用傅里叶模型。
需要说明的是，误差矩阵的散点数据分布基本符合两种标准函数图形的特征1.单峰函数；2.多峰函数。因此各段漏磁导数据使用施韦费尔与三角函数作为判别标准。谷本相似度适合处理无打分的偏好数据。谷本相似度衡量的是维度间取值方向的一致性，注重维度之间的差异，不注重数值上的差异，谷本相似度更适宜于曲线相似度的计算上。谷本相似度结果范围在0至1之间，当T(x 1,x 2)＝1时代表完全重合，当T(x 1,x 2)＝0时代表无重叠项。
步骤S108，采用最优克里金基函数对机理模型进行误差补偿，通过误差补偿后的机理模型进行燃气轮机组的故障预警。
优选地，设P为真实数据，Q为机理模型计算的预测数据，则存在：
Er＝P-Q              (30)
设每个采样点的误差为ΔG，则存在：
Figure PCTCN2022121824-appb-000028
其中，ΔG ij代表各个采样点的误差数据；n代表大气压力或其他输入参数；m代表燃机转速。
将各个基函数计算所得数据与误差项进行对比，确定最优克里金基函数的基函数，表达式如下：
f(x 1,x 2,...,x q)＝min{f k(x 1,x 2,...,x q)-Er[k；]}       (32)
其中，k＝1，2，3，…n。q代表输入样本维度。
在确定f(x 1,x 2…x q)之后，对转速、采样点的误差ΔG与最优克里金基函数进行拟合，得到函数关系：
ΔG＝p 00+p 10r+p 01f 3(x 1,x 2...x q)+p 20r+p 11rf(x 1,x 2...x q)        (33)
其中，r为燃机转速
将上式进行转化得到：
f Er(r,x 1,x 2,...,x q)＝X[ΔG,r,f(x 1,x 2,...,x q)]        (34)
其中，f Er(r,x 1,x 2…x q)为燃气轮机转速采样点变化的误差补偿函数。
设第i段的输出函数为q i(x)，可通过将样本[x 1，x 2，，，x q]代入机理模型计算得到，x为样本量之外的待测点。可得：
w i(x)＝f eri(x)+q i(x)           (35)
其中，w i(x)为经过误差修正的机理模型计算数据，相较q i(x)，w i(x)更接近实测数据。当燃汽轮机组出现从数据异常时，w i(x)会与实测数据产生差距，从而实现故障预警。同时当燃汽轮机组数据无异常时，该方法能够对机理模型进行误差补偿，从而降低故障误报。
通过本申请实施例中的步骤S102至步骤S108，解决了现有燃气轮机故障预警存在的大量故障误报的问题，实现了基于改进克里金基函数对机理模型进行误差补偿后，燃气轮机组的故障预警，降低了机理模型对故障预警的误报率。
需要说明的是，在上述流程中或者附图的流程图中示出的步骤可以在诸如一组计算机可执行指令的计算机系统中执行，并且，虽然在流程图中示出了逻辑顺序，但是在某些情况下，可以以不同于此处的顺序执行所示出或描述的步骤。
本申请实施例提供了一种应用于燃气轮机组的故障预警系统，图2是根据本申请实施例提供的应用于燃气轮机组的故障预警系统的结构框图，如图2所示，该系统包括数据采集模块21、误差补偿模块22和故障预警模块23；
数据采集模块21，用于通过机理模型，计算燃气轮机组中预测参数的预测数据，将预测数据和预测参数的真实数据进行数据对比，得到误差矩阵；
误差补偿模块22，用于根据机理模型，构建若干克里金基函数；根据误差矩阵，从若干克里金基函数中筛选出最优克里金基函数，采用最优克里金基函数对机理模型进行误差补偿；
故障预警模块23，用于通过误差补偿后的机理模型进行燃气轮机组的故障预警。
通过本申请实施例中的数据采集模块21、误差补偿模块22和故障预警模块23，解决了现有燃气轮机故障预警存在的大量故障误报的问题，实现了基于改进克里金基函数对机理模型进行误差补偿后，燃气轮机组的故障预警，降低了机理模型对故障预警的误报率。
需要说明的是，上述各个模块可以是功能模块也可以是程序模块，既可以通过软件来实现，也可以通过硬件来实现。对于通过硬件来实现的模块而言，上述各个模块可以位于同一处理器中；或者上述各个模块还可以按照任意组合 的形式分别位于不同的处理器中。
本实施例还提供了一种电子装置，包括存储器和处理器，该存储器中存储有计算机程序，该处理器被设置为运行计算机程序以执行上述任一项方法实施例中的步骤。
可选地，上述电子装置还可以包括传输设备以及输入输出设备，其中，该传输设备和上述处理器连接，该输入输出设备和上述处理器连接。
需要说明的是，本实施例中的具体示例可以参考上述实施例及可选实施方式中所描述的示例，本实施例在此不再赘述。
另外，结合上述实施例中的应用于燃气轮机组的故障预警方法，本申请实施例可提供一种存储介质来实现。该存储介质上存储有计算机程序；该计算机程序被处理器执行时实现上述实施例中的任意一种应用于燃气轮机组的故障预警方法。
在一个实施例中，提供了一种计算机设备，该计算机设备可以是终端。该计算机设备包括通过系统总线连接的处理器、存储器、网络接口、显示屏和输入装置。其中，该计算机设备的处理器用于提供计算和控制能力。该计算机设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作系统和计算机程序。该内存储器为非易失性存储介质中的操作系统和计算机程序的运行提供环境。该计算机设备的网络接口用于与外部的终端通过网络连接通信。该计算机程序被处理器执行时以实现一种应用于燃气轮机组的故障预警方法。该计算机设备的显示屏可以是液晶显示屏或者电子墨水显示屏，该计算机设备的输入装置可以是显示屏上覆盖的触摸层，也可以是计算机设备外壳上设置的按键、轨迹球或触控板，还可以是外接的键盘、触控板或鼠标等。
在一个实施例中，图3是根据本申请实施例的电子设备的内部结构示意图，如图3所示，提供了一种电子设备，该电子设备可以是服务器，其内部结构图可以如图3所示。该电子设备包括通过内部总线连接的处理器、网络接口、内存储器和非易失性存储器，其中，该非易失性存储器存储有操作系统、计算机程序和数据库。处理器用于提供计算和控制能力，网络接口用于与外部的终端通过网络连接通信，内存储器用于为操作系统和计算机程序的运行提供环境，计算机程序被处理器执行时以实现一种应用于燃气轮机组的故障预警方法，数据库用于存储数据。
本领域技术人员可以理解，图3中示出的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应用于其上的电子设备的限定， 具体的电子设备可以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机程序来指令相关的硬件来完成，该计算机程序可存储于一非易失性计算机可读取存储介质中，该计算机程序在执行时，可包括如上述各方法的实施例的流程。其中，本申请所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用，均可包括非易失性和/或易失性存储器。非易失性存储器可包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦除可编程ROM(EEPROM)或闪存。易失性存储器可包括随机存取存储器(RAM)或者外部高速缓冲存储器。作为说明而非局限，RAM以多种形式可得，诸如静态RAM(SRAM)、动态RAM(DRAM)、同步DRAM(SDRAM)、双数据率SDRAM(DDRSDRAM)、增强型SDRAM(ESDRAM)、同步链路(Synchlink)DRAM(SLDRAM)、存储器总线(Rambus)直接RAM(RDRAM)、直接存储器总线动态RAM(DRDRAM)、以及存储器总线动态RAM(RDRAM)等。
本领域的技术人员应该明白，以上所述实施例的各技术特征可以进行任意的组合，为使描述简洁，未对上述实施例中的各个技术特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛盾，都应当认为是本说明书记载的范围。
以上所述实施例仅表达了本申请的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对发明专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本申请构思的前提下，还可以做出若干变形和改进，这些都属于本申请的保护范围。因此，本申请专利的保护范围应以所附权利要求为准。

Claims (10)

  1. 一种应用于燃气轮机组的故障预警方法，其特征在于，所述方法包括：
    通过机理模型，计算燃气轮机组中预测参数的预测数据，将所述预测数据和所述预测参数的真实数据进行数据对比，得到误差矩阵；
    根据所述机理模型，构建若干克里金基函数；
    根据所述误差矩阵，从所述若干克里金基函数中筛选出最优克里金基函数，采用所述最优克里金基函数对所述机理模型进行误差补偿；
    通过所述误差补偿后的机理模型进行所述燃气轮机组的故障预警。
  2. 根据权利要求1所述的方法，其特征在于，根据所述误差矩阵，从所述若干克里金基函数中筛选出最优克里金基函数包括：
    通过所述若干克里金基函数对所述误差矩阵分别进行拟合，根据若干拟合结果对所述克里金基函数进行筛选，得到第一基函数集合；
    计算所述第一基函数集合中若干克里金基函数的可靠性指标，根据所述可靠性指标筛选出最佳克里金基函数。
  3. 根据权利要求2所述的方法，其特征在于，计算所述第一基函数集合中若干克里金基函数的可靠性指标，根据所述可靠性指标筛选出最佳克里金基函数包括：
    通过蒙特卡洛算法、粒子群算法、响应面算法和一元二次矩阵算法，分别计算所述第一基函数集合中克里金基函数的可靠性指标；
    根据所述可靠性指标筛选出收敛速率高于预设阈值的第二基函数集合，进而从所述第二基函数集合中筛选出最优克里金基函数。
  4. 根据权利要求3所述的方法，其特征在于，进而从所述第二基函数集合中筛选出最优克里金基函数包括：
    计算所述第二基函数集合中克里金基函数的导数，根据所述导数判断所述克里金基函数的曲线形态特征，其中，所述曲线形态特征包括单峰极大值特征、单峰极小值特征和多峰特征；
    根据所述曲线形态特征从所述第二基函数集合中筛选出最优克里金基函数。
  5. 根据权利要求4所述的方法，其特征在于，根据所述曲线形态特征从所述第二基函数集合中筛选出最优克里金基函数包括：
    根据所述曲线形态特征，判断对应的克里金基函数是否与施韦费尔函数或三角函数具有相似特征；
    若与施韦费尔函数具有相似特征，则通过谷本相似度函数计算所述克里金基函数与施韦费尔函数的谷本相似度；
    若与三角函数具有相似特征，则通过谷本相似度函数计算所述克里金基函数与三角函数的谷本相似度；
    根据所述谷本相似度从所述第二基函数集合中筛选出最优克里金基函数。
  6. 根据权利要求3所述的方法，其特征在于，根据所述可靠性指标筛选出收敛速率高于预设阈值的第二基函数集合包括：
    根据克里金基函数的可靠性指标的函数计算次数，从蒙特卡洛算法、粒子群算法、响应面算法和一元二次矩阵算法中选取出最优计算算法；
    根据所述最优计算算法计算得到的可靠性指标，筛选出收敛速率高于预设阈值的第二基函数集合。
  7. 根据权利要求1所述的方法，其特征在于，根据所述机理模型，构建若干预设克里金基函数包括：
    将所述机理模型的输入参数作为克里金基函数的输入参数，将所述机理模型的预测参数作为克里金基函数的输出参数，进而构建若干预设克里金基函数。
  8. 根据权利要求1所述的方法，其特征在于，通过机理模型，计算燃气轮机组中预测参数的预测数据包括：
    将燃气轮机组的输入数据输入到机理模型中，计算所述燃气轮机组中预测参数的预测数据，其中，所述输入数据对应的输入参数包括燃机转速、发电机功率、环境温度和环境湿度，所述预测参数包括第一轴位移、第二轴位移、第三轴位移、高压缸膨胀、高中压缸差胀、低压缸膨胀和高压主汽压力。
  9. 一种应用于燃气轮机组的故障预警系统，其特征在于，所述系统包括数据采集模块、误差补偿模块和故障预警模块；
    所述数据采集模块，用于通过机理模型，计算燃气轮机组中预测参数的预测数据，将所述预测数据和所述预测参数的真实数据进行数据对比，得到误差矩阵；
    所述误差补偿模块，用于根据所述机理模型，构建若干克里金基函数；根据所述误差矩阵，从所述若干克里金基函数中筛选出最优克里金基函数，采用所述最优克里金基函数对所述机理模型进行误差补偿；
    所述故障预警模块，用于通过所述误差补偿后的机理模型进行所述燃气轮机组的故障预警。
  10. 一种电子装置，包括存储器和处理器，其特征在于，所述存储器中存储有计算机程序，所述处理器被设置为运行所述计算机程序以执行权利要求1至8中任一项所述的应用于燃气轮机组的故障预警方法。
PCT/CN2022/121824 2022-06-06 2022-09-27 一种应用于燃气轮机组的故障预警方法、系统和装置 WO2023236393A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US18/313,525 US11860619B2 (en) 2022-06-06 2023-05-08 Fault early-warning method and system applied to gas turbine unit, and apparatus

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202210633645.3 2022-06-06
CN202210633645.3A CN115062425B (zh) 2022-06-06 2022-06-06 一种应用于燃气轮机组的故障预警方法、系统和装置

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US18/313,525 Continuation US11860619B2 (en) 2022-06-06 2023-05-08 Fault early-warning method and system applied to gas turbine unit, and apparatus

Publications (1)

Publication Number Publication Date
WO2023236393A1 true WO2023236393A1 (zh) 2023-12-14

Family

ID=83199733

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/121824 WO2023236393A1 (zh) 2022-06-06 2022-09-27 一种应用于燃气轮机组的故障预警方法、系统和装置

Country Status (2)

Country Link
CN (1) CN115062425B (zh)
WO (1) WO2023236393A1 (zh)

Families Citing this family (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110208365B (zh) * 2019-05-31 2023-03-31 南京航空航天大学 一种基于脉冲漏磁信号暂态特征的缺陷量化评估方法
US11860619B2 (en) 2022-06-06 2024-01-02 Huadian Electric Power Research Institute Co., Ltd. Fault early-warning method and system applied to gas turbine unit, and apparatus
CN115062425B (zh) * 2022-06-06 2023-08-18 华电电力科学研究院有限公司 一种应用于燃气轮机组的故障预警方法、系统和装置

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190284955A1 (en) * 2018-03-14 2019-09-19 Doosan Heavy Industries & Construction Co., Ltd. Fault detecting apparatus, gas turbine, and method of detecting fault
CN111693084A (zh) * 2020-06-23 2020-09-22 南京航空航天大学 一种基于误差相似性的测量误差补偿方法
CN113703422A (zh) * 2021-08-26 2021-11-26 华北电力大学 一种基于特征分析处理的燃气轮机气动执行机构故障诊断方法
CN113761803A (zh) * 2021-09-13 2021-12-07 中国科学院工程热物理研究所 燃气轮机的补偿模型的训练方法及使用
CN115062425A (zh) * 2022-06-06 2022-09-16 华电电力科学研究院有限公司 一种应用于燃气轮机组的故障预警方法、系统和装置

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111581803B (zh) * 2020-04-30 2022-04-26 北京航空航天大学 一种全球电离层电子含量的克里金代理模型构建方法
CN111829782B (zh) * 2020-07-16 2021-12-07 苏州大学 一种基于自适应流形嵌入动态分布对齐的故障诊断方法
CN113515893B (zh) * 2021-07-01 2024-05-17 中国科学院过程工程研究所 一种稀土萃取过程实时预测模型的建立方法及预测方法和预测装置
CN113705045B (zh) * 2021-08-20 2024-04-12 上海交通大学 一种基于代理模型的转静子系统碰摩可靠性分析方法

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20190284955A1 (en) * 2018-03-14 2019-09-19 Doosan Heavy Industries & Construction Co., Ltd. Fault detecting apparatus, gas turbine, and method of detecting fault
CN111693084A (zh) * 2020-06-23 2020-09-22 南京航空航天大学 一种基于误差相似性的测量误差补偿方法
CN113703422A (zh) * 2021-08-26 2021-11-26 华北电力大学 一种基于特征分析处理的燃气轮机气动执行机构故障诊断方法
CN113761803A (zh) * 2021-09-13 2021-12-07 中国科学院工程热物理研究所 燃气轮机的补偿模型的训练方法及使用
CN115062425A (zh) * 2022-06-06 2022-09-16 华电电力科学研究院有限公司 一种应用于燃气轮机组的故障预警方法、系统和装置

Also Published As

Publication number Publication date
CN115062425A (zh) 2022-09-16
CN115062425B (zh) 2023-08-18

Similar Documents

Publication Publication Date Title
WO2023236393A1 (zh) 一种应用于燃气轮机组的故障预警方法、系统和装置
WO2019144384A1 (zh) 一种航空发动机启动过程排气温度预测方法
Wang et al. A Fault Diagnosis Approach for Gas Turbine Exhaust Gas Temperature Based on Fuzzy C‐Means Clustering and Support Vector Machine
Fei et al. Dynamic probabilistic design approach of high-pressure turbine blade-tip radial running clearance
Chen Fuzzy testing of operating performance index based on confidence intervals
Xue et al. An online updating approach for testing the proportional hazards assumption with streams of survival data
Li et al. Improved method for gas-turbine off-design performance adaptation based on field data
US20120078567A1 (en) Combustion reference temperature estimation
Xia et al. Low-dimensional confounder adjustment and high-dimensional penalized estimation for survival analysis
Wang et al. Novel numerical methods for reliability analysis and optimization in engineering fuzzy heat conduction problem
CN112215398A (zh) 电力用户负荷预测模型建立方法、装置、设备及存储介质
Zhao et al. A GM (1, 1) Markov Chain‐Based Aeroengine Performance Degradation Forecast Approach Using Exhaust Gas Temperature
CN116816509A (zh) 一种燃气轮机变系数特性曲线修正方法和系统
Huang et al. Gas path deterioration observation based on stochastic dynamics for reliability assessment of aeroengines
Huang et al. Testing for the shape parameter of generalized extreme value distribution based on the-likelihood ratio statistic
Knoth et al. EWMA p charts under sampling by variables
US11860619B2 (en) Fault early-warning method and system applied to gas turbine unit, and apparatus
CN109388884A (zh) 一种计算耦合叶盘疲劳寿命的广义回归极值响应面法
Wang et al. Performance dispersion control of a multistage compressor based on precise identification of critical features
Ramerth et al. A probabilistic secondary flow system design process for gas turbine engines
CN112182739A (zh) 一种飞行器结构非概率可信可靠性拓扑优化设计方法
Dou et al. Optimization method of suspected electricity theft topic model based on chi-square test and logistic regression
CN116227367B (zh) 直接空冷系统的背压预测模型构建方法、预测方法及装置
CN111625753A (zh) 用于计算直燃机能效参数的方法、装置、设备及存储介质
Kapusuzoglu et al. Multi-Level Bayesian Calibration of a Multi-Component Dynamic System Model

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22945525

Country of ref document: EP

Kind code of ref document: A1