A method and device for rapid detection of combustion oscillation faults

Abstract

The present invention relates to a method and device for quickly detecting combustion oscillation faults, wherein the method comprises obtaining key signals of a combustion chamber during a combustion process and determining multiple combustion states; the key signals include time series data; symbol segmentation is performed on the time series data based on a maximum entropy segmentation method to obtain a symbol time series; a reference state is obtained based on the symbol time series and multiple combustion states; a current combustion state among multiple combustion states is determined based on the reference state; when the current combustion state is a combustion oscillation mode, an early warning mechanism is activated; the present invention can process the time series data and the combustion state of the combustion chamber during the combustion process using the maximum entropy segmentation method, and determine the current combustion state through the combustion state and the reference state, thereby performing real-time monitoring of the combustion oscillation through the current combustion state, without the need for complex sensor arrangement and algorithm processing, thereby reducing costs, reducing the amount of calculation, and improving early warning efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of combustion fault diagnosis, and in particular to a method and device for quickly detecting combustion oscillation faults. Background Art

[0002] With increasingly stringent environmental regulations, the requirements for reducing NOx emissions are gradually increasing. Lean premixed combustion technology is one of the most effective emission reduction technologies in modern gas turbine combustion. However, when using lean premixed combustion, combustion oscillation is prone to occur. Combustion oscillation often causes large fluctuations in the combustion chamber pressure. It is mainly caused by the coupling of unsteady heat release and combustion chamber acoustics. It is widely present in the combustion chambers of systems such as gas turbines and rocket engines, and can seriously affect the stability and service life of the system. Therefore, to ensure the stability of the combustion system and prevent it from entering an oscillatory state, it is crucial to develop rapid detection methods for combustion oscillation faults.

[0003] In recent years, domestic and foreign scholars have done a lot of work on the real-time monitoring of combustion oscillations. In existing technologies, most of them rely on physical sensors to directly measure the fluctuations of parameters such as pressure and temperature. However, these methods mostly rely on complex sensor arrangements and complex algorithm processing, and have problems such as high cost, complex calculations, and untimely warnings. In addition, they may not be able to adapt to the diversity of operating conditions in certain extreme environments, which will inevitably affect the safe operation of the equipment.

[0004] Therefore, it is urgent to propose a method and device for rapid detection of combustion oscillation faults to solve the technical problems that the combustion oscillation monitoring methods in the existing technology mostly rely on complex sensor arrangements and complex algorithm processing, and have high costs, complex calculations, and untimely warnings. Summary of the Invention

[0005] In view of this, it is necessary to provide a method and device for rapid detection of combustion oscillation faults, which mostly rely on complex sensor arrangements and complex algorithm processing, and have problems such as high cost, complex calculations, and untimely warnings.

[0006] In order to solve the above problems, the present invention provides a method for rapid detection of combustion oscillation faults, comprising:

[0007] Acquire key signals of the combustion chamber during the combustion process and determine multiple combustion states; the key signals include time series data;

[0008] Performing symbol segmentation on the time series data based on a maximum entropy segmentation method to obtain a symbol time series;

[0009] obtaining a reference state according to the symbol time sequence and the plurality of combustion states;

[0010] determining a current combustion state among the plurality of combustion states based on the reference state;

[0011] When the current combustion state is the combustion oscillation mode, the early warning mechanism is activated.

[0012] In a possible implementation, preprocessing the time series data based on a digital filtering technology to obtain the preprocessed time series data includes:

[0013] Performing subsequence extraction on the time series data according to a preset window length to obtain a multidimensional trajectory matrix;

[0014] According to the multidimensional trajectory matrix, a plurality of groups and a matrix corresponding to each group are obtained;

[0015] Reconstructing the matrix of each group according to a diagonal averaging algorithm to obtain a time series corresponding to each group;

[0016] According to all time series, the time series data after preprocessing is obtained.

[0017] In a possible implementation, obtaining a plurality of groups and a matrix corresponding to each group according to the multidimensional trajectory matrix includes:

[0018] Performing singular value decomposition on the multidimensional trajectory matrix to obtain a plurality of eigenvalues ​​and an eigenvector corresponding to each eigenvalue;

[0019] The multi-dimensional trajectory matrix is ​​divided according to all eigenvectors to obtain a plurality of groups and a matrix corresponding to each group.

[0020] In a possible implementation, performing symbol segmentation on the time series data based on the maximum entropy segmentation method to obtain a symbol time series includes:

[0021] Arranging the time series in the time series data in ascending order to obtain a sorted time series;

[0022] Determine the equal division points according to the preset symbol set;

[0023] Determining, according to the equal-division points, corresponding equal-division point values ​​in the sorted time series;

[0024] The equally divided point values ​​and the sorted time series are segmented based on a maximum entropy segmentation method to obtain a symbolic time series.

[0025] In a possible implementation, obtaining a reference state according to the symbol time sequence and the multiple combustion states includes:

[0026] performing probability calculation on the plurality of combustion states according to the symbol time series to obtain a state probability of each combustion state;

[0027] Obtaining a state probability histogram according to the state probabilities of the multiple combustion states;

[0028] An analysis is performed based on the state probability histogram to obtain a reference state.

[0029] In a possible implementation, determining a current combustion state among the multiple combustion states according to the reference state includes:

[0030] Performing similarity measurement on each combustion state according to the reference state to obtain a similarity measurement value for each combustion state;

[0031] Based on all similarity metrics, the current combustion state is determined.

[0032] In a possible implementation, when the current combustion state is in the combustion oscillation mode, starting the early warning mechanism includes:

[0033] When the similarity metric value of the current combustion state is greater than a preset threshold, the current combustion state is determined to be a combustion oscillation mode, and an early warning mechanism is activated.

[0034] In a possible implementation, when the current combustion state is the combustion oscillation mode, after the early warning mechanism is activated, the method further includes:

[0035] determining a fuel or air ratio to be adjusted according to the key signal;

[0036] The combustion state in the combustion chamber is adjusted according to the fuel to be adjusted or the air ratio.

[0037] In one possible implementation, the similarity measure is calculated as:

[0038]

[0039] Where, represents the threshold value, represents the Euclidean norm, represents the probability distribution of stable combustion state, The probability distribution of the oscillatory state, is the current combustion state probability distribution, and 、 and have the same number of symbols.

[0040] On the other hand, the present invention also provides a combustion oscillation fault rapid detection device, comprising:

[0041] A signal acquisition module is used to acquire key signals of the combustion chamber during the combustion process and determine multiple combustion states; the key signals include time series data;

[0042] A symbol segmentation module, configured to perform symbol segmentation on the time series data based on a maximum entropy segmentation method to obtain a symbol time series;

[0043] a state determination module, configured to obtain a reference state based on the symbol time sequence and the plurality of combustion states;

[0044] a state determination module, configured to determine a current combustion state among the plurality of combustion states according to the reference state;

[0045] The state warning module is used to start the warning mechanism when the current combustion state is the combustion oscillation mode.

[0046] The beneficial effects of the present invention are: obtaining key signals of the combustion chamber during the combustion process and determining multiple combustion states; the key signals include time series data; symbol segmentation of the time series data is performed based on the maximum entropy segmentation method to obtain a symbol time series; a reference state is obtained based on the symbol time series and multiple combustion states; the current combustion state among multiple combustion states is determined based on the reference state; when the current combustion state is a combustion oscillation mode, an early warning mechanism is activated; the present invention can process the time series data and the combustion state of the combustion chamber during the combustion process through the maximum entropy segmentation method, and determine the current combustion state through the combustion state and the reference state, thereby performing real-time monitoring of the combustion oscillation through the current combustion state, without the need for complex sensor arrangement and algorithm processing, thereby reducing costs, reducing calculation amount, and improving early warning efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A schematic flow chart of an embodiment of a method for rapid detection of combustion oscillation faults provided by the present invention;

[0048] Figure 2 A schematic diagram of a process flow of an embodiment of the pre-processing provided by the present invention;

[0049] Figure 3 A signal schematic diagram of an embodiment of a signal close to a combustion oscillation operating condition provided by the present invention;

[0050] Figure 4 For the present invention Figure 1 A schematic flow chart of an embodiment of step S102;

[0051] Figure 5 A schematic diagram of the structure of an embodiment of data segmentation provided by the present invention;

[0052] Figure 6 A signal diagram of an embodiment of the combustion oscillation formation process provided by the present invention;

[0053] Figure 7 A schematic structural diagram of an embodiment of a combustion oscillation fault rapid detection device provided by the present invention;

[0054] Figure 8 This is a schematic structural diagram of an embodiment of the electronic device provided by the present invention. DETAILED DESCRIPTION

[0055] The preferred embodiments of the present invention will be described in detail below in conjunction with the accompanying drawings, wherein the accompanying drawings constitute a part of this application and are used together with the embodiments of the present invention to illustrate the principles of the present invention, and are not used to limit the scope of the present invention.

[0056] like Figure 1 As shown, a specific embodiment of the present invention discloses a method for rapid detection of combustion oscillation faults, comprising:

[0057] S101, obtaining key signals of the combustion chamber during the combustion process and determining multiple combustion states; the key signals include time series data;

[0058] S102, performing symbol segmentation on the time series data based on the maximum entropy segmentation method to obtain a symbol time series;

[0059] S103, obtaining a reference state according to the symbol time sequence and multiple combustion states;

[0060] S104, determining a current combustion state among a plurality of combustion states according to a reference state;

[0061] S105: When the current combustion state is the combustion oscillation mode, start the early warning mechanism.

[0062] In a specific embodiment of the present invention, it can be applied to the combustion chambers of systems such as gas turbines and rocket engines, and can collect key signals from the combustion chamber during the combustion process. The key signals may include time series data of relevant parameters such as upstream and downstream pressure signals of the combustion chamber and sound field signals near the combustion chamber. Multiple combustion states can also be set, and the time series data can be symbolically segmented based on the maximum entropy segmentation method to obtain a symbol-time series. The specific maximum entropy segmentation method can be set according to actual conditions and is not limited by the embodiment of the present invention. The symbol-time series and multiple combustion states can also be processed to obtain a reference state. Based on the reference state, multiple combustion states can be determined to obtain the current combustion state of the combustion chamber. The combustion chamber in the current combustion state can then be monitored in real time. When the current combustion state is in a combustion oscillation mode, an early warning mechanism can be activated, and the activation of the early warning mechanism can provide a prompt to the staff.

[0063] Compared with the prior art, the present embodiment provides a method for obtaining key signals of a combustion chamber during a combustion process and determining multiple combustion states; the key signals include time series data; symbol segmentation is performed on the time series data based on the maximum entropy segmentation method to obtain a symbol time series; a reference state is obtained based on the symbol time series and multiple combustion states; a current combustion state among multiple combustion states is determined based on the reference state; when the current combustion state is a combustion oscillation mode, an early warning mechanism is activated; the present invention can process the time series data and the combustion state of the combustion chamber during the combustion process based on the maximum entropy segmentation method, and determine the current combustion state through the combustion state and the reference state, thereby performing real-time monitoring of the combustion oscillation through the current combustion state, without the need for complex sensor arrangement and algorithm processing, thereby reducing costs, reducing calculation amount, and improving early warning efficiency.

[0064] It should be noted that the collected key signals may contain different data. In some embodiments of the present invention, before step S102, the following steps are further included:

[0065] The time series data is preprocessed based on digital filtering technology to obtain the preprocessed time series data.

[0066] In a specific embodiment of the present invention, after obtaining the time series data, the time series data can be preprocessed using digital filtering technology to obtain the preprocessed time series data. The specific preprocessing process can be set according to actual conditions, and the embodiment of the present invention is not limited here.

[0067] In some embodiments of the present invention, Figure 2 As shown in the figure, the time series data is preprocessed based on the digital filtering technology, and the time series data after preprocessing includes:

[0068] S201, extracting subsequences from time series data according to a preset window length to obtain a multidimensional trajectory matrix;

[0069] S202, obtaining a plurality of groups and a matrix corresponding to each group according to the multidimensional trajectory matrix;

[0070] S203, reconstructing the matrix of each group according to the diagonal averaging algorithm to obtain the time series corresponding to each group;

[0071] S204: Obtain pre-processed time series data based on all time series.

[0072] In a specific embodiment of the present invention, a preset window length can be set L , the time series data is , we can extract subsequences from time series data according to the preset window length and obtain a multi-dimensional trajectory matrix, as shown in formula (1):

[0073] (1)

[0074] For example, the time series data is {0, 1, 1, 0}, and the preset window length is L is 3, If 4, K =4-3+1=2, then the multidimensional trajectory matrix is ​​shown in formula (2):

[0075] (2)

[0076] In some embodiments of the present invention, step S202 includes:

[0077] Perform singular value decomposition on the multidimensional trajectory matrix to obtain multiple eigenvalues ​​and the eigenvector corresponding to each eigenvalue;

[0078] The multidimensional trajectory matrix is ​​divided according to all eigenvectors to obtain multiple groups and a matrix corresponding to each group.

[0079] In a specific embodiment of the present invention, the multidimensional trajectory matrix X Perform singular value decomposition, as shown in formula (3):

[0080] (3)

[0081] Where, , , .

[0082] In order to obtain the singular value decomposition of the matrix X, it is necessary to solve the orthogonal matrix and .

[0083] Orthogonal Matrix and Satisfying formulas (4) and (5):

[0084] (4)

[0085] (5)

[0086] Where, is the identity matrix.

[0087] Calculated Perform singular value decomposition as formula (6):

[0088] (6)

[0089] If we write it as the sum of column vectors multiplied by row vectors, we can get formula (7):

[0090] (7)

[0091] Where, is the rank of the matrix, which is also the number of nonzero singular values.

[0092] Then, the multidimensional trajectory matrix is ​​transformed into Divide into several different groups, where each non-zero singular value is a separate group, let For the The matrix contained in the group, the truncated singular value decomposition is shown in formulas (8) and (9):

[0093] (8)

[0094] (9)

[0095] Then we can get multiple groups and matrices corresponding to each group. Then we can reconstruct each matrix into m The transformation rule of the time series corresponding to the components is shown in formula (10):

[0096] (10)

[0097] Where, Represents the matrix at position elements.

[0098] Then all time series can be integrated to obtain the preprocessed time series data.

[0099] The first m-order modes are selected for reconstruction to achieve the purpose of noise reduction. Since the energy contained in the pressure sequence mainly comes from the oscillation signal when combustion oscillation occurs, the singular spectrum analysis is used to process the pressure signal, and the components that contribute most to the energy of the original signal are selected for reconstruction. The weak oscillation signal in the early stage of combustion oscillation is extracted. The results after filtering are as follows: Figure 3 As shown, Figure 3 (a1) is the original signal close to the combustion oscillation condition. It can be seen that the original signal is seriously interfered with by the signal. (a2) is the image of the original signal after filtering. It can be seen that after filtering, a signal with an approximate periodic oscillation is obtained. (b1) is the original signal under the oscillation condition. The original signal has shown obvious oscillation characteristics, but there are also many noise points, indicating that the oscillation intensity at this time is much greater than the noise intensity. (b2) is the filtered image. It can be seen that after filtering, the signal becomes smoother and the noise has been basically removed.

[0100] In some embodiments of the present invention, Figure 4 As shown, step S102 includes:

[0101] S401, arranging the time series in the time series data in ascending order to obtain a sorted time series;

[0102] S402: Determine equal division points according to a preset symbol set;

[0103] S403: Determine the corresponding equal-division point value in the sorted time series according to the equal-division point;

[0104] S404 , segmenting the equally divided point values ​​and the sorted time series based on the maximum entropy segmentation method to obtain a symbolic time series.

[0105] In a specific embodiment of the present invention, it is assumed that the preset symbol set of different symbols in the symbol sequence is: , and its corresponding probability is: , and meet the following requirements:

[0106] (a) ;

[0107] (b)

[0108] Then the information entropy value is calculated as shown in formula (11):

[0109] (11)

[0110] All time series in the preprocessed time series data can be arranged in ascending order to obtain the sorted time series, and then the equal division points of the preset symbol set can be taken out. The equal division points can be , we can extract the corresponding equal-division point values ​​in the sorted time series according to the equal-division points, and then we can segment the sorted time series using the maximum entropy segmentation method to obtain the symbolic time series. The symbolic time series is shown in formula (12):

[0111] (12)

[0112] Where, 、 、 and is the time series obtained by segmentation in the symbol time series; is the minimum value of the time series data; is the maximum value of the time series data.

[0113] For example, Figure 5, the data can be divided into 4 segments, that is, the preset symbol set is , choose the length of the symbol string to be 1, then the state set is isomorphic to the symbol set, that is, the combustion state set After segmentation, the symbol time series corresponding to the original sound pressure sequence is obtained, such as Figure 5 In the 122344231, the finite state machine generally refers to the finite state automaton, which represents The relationship between them.

[0114] In some embodiments of the present invention, step S103 includes:

[0115] Probability calculation is performed on multiple combustion states according to the symbol time series to obtain the state probability of each combustion state;

[0116] According to the state probabilities of multiple combustion states, a state probability histogram is obtained;

[0117] Analyze the state probability histogram to obtain the reference state.

[0118] In a specific embodiment of the present invention, the number of times each combustion state appears in the symbol time sequence can be counted, and the probability of its appearance can be calculated, so as to obtain the state probability of each combustion state. This probability is the relative frequency of the state in the symbol sequence. Then, a state probability histogram can be drawn based on the state probabilities of multiple combustion states. Figure 5 In the state probability histogram, the x-axis represents the state set and the y-axis represents the state probability. For the combustion system under specific conditions, the state probability histogram of the stable combustion state set and the limit cycle oscillation state has stable characteristics and can reflect the sound pressure fluctuation of each combustion state. It is called the reference state.

[0119] In some embodiments of the present invention, step S104 includes:

[0120] Performing similarity measurement on each combustion state according to the reference state to obtain a similarity measurement value of each combustion state;

[0121] Based on all similarity metrics, the current combustion state is determined.

[0122] In a specific embodiment of the present invention, a similarity measurement is performed on each combustion state according to a reference state, so that a similarity measurement value of each combustion state can be obtained. The similarity measurement value is calculated as shown in formula (13):

[0123] (13)

[0124] Where, represents the threshold value, represents the Euclidean norm, represents the probability distribution of stable combustion state, The probability distribution of the oscillatory state, is the current combustion state probability distribution, and 、 and have the same number of symbols.

[0125] Then, the state corresponding to the largest similarity measure value among all similarity measure values ​​may be determined as the current combustion state.

[0126] In some embodiments of the present invention, step S105 includes:

[0127] When the similarity metric value of the current combustion state is greater than a preset threshold, the current combustion state is determined to be a combustion oscillation mode, and the early warning mechanism is activated.

[0128] In a specific embodiment of the present invention, when the similarity metric value of the current combustion state is greater than a preset threshold, it can be determined that the current combustion state is a combustion oscillation mode, and the system activates the early warning mechanism, wherein the preset threshold can be set according to actual conditions, and the embodiment of the present invention is not limited here. Figure 6 As shown, Figure 6 The middle figure shows the formation process of a combustion oscillation. The upper figure is the time series of dynamic pressure, and the lower figure is the combustion oscillation detection process based on the threshold. By calculating the degree of proximity between the current state and the reference state, it can be seen that in the early stage of the formation of combustion oscillation, the pressure amplitude does not change much, but If the preset threshold is exceeded, an early warning signal is issued to detect combustion oscillations.

[0129] In some embodiments of the present invention, after step S105, the method further includes:

[0130] Determine the fuel or air ratio to be adjusted based on key signals;

[0131] The combustion state in the combustion chamber is adjusted according to the fuel or air ratio to be adjusted.

[0132] In a specific embodiment of the present invention, after the early warning mechanism is activated, the fuel or air ratio to be adjusted can be determined based on the key signal to eliminate or alleviate the oscillation phenomenon in the combustion chamber.

[0133] In the embodiment of the present invention, symbolic dynamics analysis is used to simplify the data processing process, reduce the computational complexity, and reveal the hidden structures and patterns in the time series, thereby improving the detection speed and accuracy of combustion oscillation faults. The symbolic processing method has a strong anti-interference ability and can greatly reduce the influence of dynamic noise and measurement noise. The method mainly collects sound field signals without affecting the combustion field, can monitor the combustion state in real time, and does not require complex hardware equipment. It has strong adaptability and versatility and is suitable for different types of combustion equipment and different operating conditions.

[0134] In order to better implement the combustion oscillation fault rapid detection method in the embodiment of the present invention, based on the combustion oscillation fault rapid detection method, the embodiment of the present invention also provides a combustion oscillation fault rapid detection device, such as Figure 7 As shown, the combustion oscillation fault rapid detection device 700 includes:

[0135] The signal acquisition module 701 is used to obtain key signals of the combustion chamber during the combustion process and determine multiple combustion states; the key signals include time series data;

[0136] The symbol segmentation module 702 is used to perform symbol segmentation on the time series data based on the maximum entropy segmentation method to obtain a symbol time series;

[0137] A state determination module 703 is configured to obtain a reference state based on the symbol time sequence and multiple combustion states;

[0138] a state determination module 704 for determining a current combustion state among a plurality of combustion states according to a reference state;

[0139] The state warning module 705 is used to start the warning mechanism when the current combustion state is the combustion oscillation mode.

[0140] The combustion oscillation fault rapid detection device 700 provided in the above embodiment can implement the technical solution described in the above embodiment of the combustion oscillation fault rapid detection method. The specific implementation principles of the above modules or units can refer to the corresponding contents in the above embodiment of the combustion oscillation fault rapid detection method, which will not be repeated here.

[0141] like Figure 8 As shown, the present invention also provides an electronic device 800. The electronic device 800 includes a processor 801, a memory 802 and a display 803. Figure 8 Only some of the components of the electronic device 800 are shown, but it should be understood that implementation of all of the shown components is not required, and more or fewer components may be implemented instead.

[0142] In some embodiments, the memory 802 may be an internal storage unit of the electronic device 800, such as a hard disk or memory of the electronic device 800. In other embodiments, the memory 802 may also be an external storage device of the electronic device 800, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device 800.

[0143] Furthermore, the memory 802 may include both an internal storage unit of the electronic device 800 and an external storage device. The memory 802 is used to store application software installed in the electronic device 800 and various data.

[0144] In some embodiments, the processor 801 may be a central processing unit (CPU), a microprocessor, or other data processing chip, configured to execute program codes or process data stored in the memory 802 , such as the combustion oscillation fault rapid detection method of the present invention.

[0145] In some embodiments, display 803 can be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen. Display 803 is used to display information about electronic device 800 and to display a visual user interface. Components 801-803 of electronic device 800 communicate with each other via a system bus.

[0146] In some embodiments of the present invention, when the processor 801 executes the combustion oscillation fault rapid detection program in the memory 802, the following steps may be implemented:

[0147] Acquire key signals from the combustion chamber during the combustion process and determine multiple combustion states; key signals include time series data;

[0148] Perform symbol segmentation on time series data based on the maximum entropy segmentation method to obtain symbol time series;

[0149] According to the symbolic time series and multiple combustion states, a reference state is obtained;

[0150] determining a current combustion state among a plurality of combustion states based on a reference state;

[0151] When the current combustion state is the combustion oscillation mode, the early warning mechanism is activated.

[0152] It should be understood that, when the processor 801 executes the combustion oscillation fault rapid detection program in the memory 802 , in addition to the above functions, it can also implement other functions. For details, please refer to the description of the corresponding method embodiment above.

[0153] Furthermore, the embodiments of the present invention do not specifically limit the type of electronic device 800 mentioned. The electronic device 800 may be a portable electronic device such as a mobile phone, tablet computer, personal digital assistant (PDA), wearable device, or laptop computer. Exemplary embodiments of portable electronic devices include, but are not limited to, portable electronic devices running iOS, Android, Microsoft, or other operating systems. The portable electronic devices mentioned above may also be other portable electronic devices, such as a laptop computer with a touch-sensitive surface (e.g., a touch panel). It should also be understood that in some other embodiments of the present invention, the electronic device 800 may not be a portable electronic device, but rather a desktop computer with a touch-sensitive surface (e.g., a touch panel).

[0154] Accordingly, an embodiment of the present application also provides a computer-readable storage medium, which is used to store computer-readable programs or instructions. When the program or instructions are executed by a processor, the steps or functions of the combustion oscillation fault rapid detection method provided in the above-mentioned method embodiments can be implemented.

[0155] Those skilled in the art will appreciate that all or part of the process flow of the above-described method embodiment can be implemented by instructing related hardware (such as a processor, controller, etc.) through a computer program, and the computer program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a magnetic disk, an optical disk, a read-only memory, or a random access memory.

[0156] The above is a detailed introduction to the method and device for rapid detection of combustion oscillation faults provided by the present invention. Specific examples are used herein to illustrate the principles and implementation methods of the present invention. The description of the above embodiments is only used to help understand the method of the present invention and its core ideas. At the same time, for those skilled in the art, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scopes. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for rapid detection of combustion oscillation faults, characterized in that: include: Obtain key signals from the combustion chamber during the combustion process and determine multiple combustion states; The key signals include time series data; Performing symbol segmentation on the time series data based on a maximum entropy segmentation method to obtain a symbol time series; obtaining a reference state according to the symbol time sequence and the plurality of combustion states; determining a current combustion state among the plurality of combustion states based on the reference state; When the current combustion state is a combustion oscillation mode, starting an early warning mechanism; Preprocessing the time series data based on digital filtering technology to obtain preprocessed time series data includes: Performing subsequence extraction on the time series data according to a preset window length to obtain a multidimensional trajectory matrix; According to the multidimensional trajectory matrix, a plurality of groups and a matrix corresponding to each group are obtained; Reconstructing the matrix of each group according to a diagonal averaging algorithm to obtain a time series corresponding to each group; According to all time series, the time series data after preprocessing is obtained.

2. The method for rapid detection of combustion oscillation faults according to claim 1, characterized in that: The step of obtaining a plurality of groups and a matrix corresponding to each group according to the multidimensional trajectory matrix includes: Performing singular value decomposition on the multidimensional trajectory matrix to obtain a plurality of eigenvalues ​​and an eigenvector corresponding to each eigenvalue; The multi-dimensional trajectory matrix is ​​divided according to all eigenvectors to obtain a plurality of groups and a matrix corresponding to each group.

3. The method for rapid detection of combustion oscillation faults according to claim 1, characterized in that: The performing symbol segmentation on the time series data based on the maximum entropy segmentation method to obtain a symbol time series includes: Arranging the time series in the time series data in ascending order to obtain a sorted time series; Determine the equal division points according to the preset symbol set; Determining, according to the equal-division points, corresponding equal-division point values ​​in the sorted time series; The equally divided point values ​​and the sorted time series are segmented based on a maximum entropy segmentation method to obtain a symbolic time series.

4. The method for rapid detection of combustion oscillation faults according to claim 1, characterized in that: The obtaining of a reference state according to the symbol time sequence and the plurality of combustion states includes: performing probability calculation on the plurality of combustion states according to the symbol time series to obtain a state probability of each combustion state; Obtaining a state probability histogram according to the state probabilities of the multiple combustion states; An analysis is performed based on the state probability histogram to obtain a reference state.

5. The method for rapid detection of combustion oscillation faults according to claim 1, characterized in that: Determining a current combustion state among the plurality of combustion states according to the reference state includes: Performing similarity measurement on each combustion state according to the reference state to obtain a similarity measurement value for each combustion state; Based on all similarity metrics, the current combustion state is determined.

6. The method for rapid detection of combustion oscillation faults according to claim 5, characterized in that: When the current combustion state is in the combustion oscillation mode, starting the early warning mechanism includes: When the similarity metric value of the current combustion state is greater than a preset threshold, the current combustion state is determined to be a combustion oscillation mode, and an early warning mechanism is activated.

7. The method for rapid detection of combustion oscillation faults according to claim 1, characterized in that: When the current combustion state is the combustion oscillation mode, after the early warning mechanism is activated, the method further includes: determining a fuel or air ratio to be adjusted according to the key signal; The combustion state in the combustion chamber is adjusted according to the fuel to be adjusted or the air ratio.

8. The method for rapid detection of combustion oscillation faults according to claim 5, characterized in that: The similarity metric is calculated as: Where, represents the threshold value, represents the Euclidean norm, represents the probability distribution of stable combustion state, The probability distribution of the oscillatory state, is the current combustion state probability distribution, and 、 and have the same number of symbols.

9. A combustion oscillation fault rapid detection device, characterized in that: include: A signal acquisition module is used to obtain key signals of the combustion chamber during the combustion process and determine multiple combustion states; The key signals include time series data; A symbol segmentation module, configured to perform symbol segmentation on the time series data based on a maximum entropy segmentation method to obtain a symbol time series; a state determination module, configured to obtain a reference state based on the symbol time sequence and the plurality of combustion states; a state determination module, configured to determine a current combustion state among the plurality of combustion states according to the reference state; A state warning module, configured to activate a warning mechanism when the current combustion state is a combustion oscillation mode; The symbol segmentation module is further configured to perform subsequence extraction on the time series data according to a preset window length to obtain a multidimensional trajectory matrix; obtain a plurality of groups and a matrix corresponding to each group based on the multidimensional trajectory matrix; reconstruct the matrix of each group according to a diagonal averaging algorithm to obtain a time series corresponding to each group; and obtain the preprocessed time series data based on all the time series.

Images