Gas turbine abnormality determination method and gas turbine

Abstract

The invention discloses an abnormality judgment method for a gas turbine and the gas turbine, and relates to the field of gas turbine control. The method comprises the steps that operation data and a standard control limiting value corresponding to the current operation working condition are obtained; processing the operation data to obtain mapping data; calculating a reference index corresponding to the mapping data, and obtaining a control limit value corresponding to the reference index; when it is determined that the control limit value is larger than the standard control limit value, it is judged that the gas turbine is abnormal under the current operation working condition and the current time period; and calculating the contribution rate of each data point of the mapping data, and judging the target corresponding to the data point of which the contribution rate is greater than a preset contribution rate as abnormal. The method and the device are used for solving the problems of poor accuracy, low efficiency and long consumed time when the abnormity of the gas turbine is judged manually in the prior art, and ensuring that the abnormity of the gas turbine is judged automatically, accurately and quickly.

Description

Technical Field

[0001] The present application relates to the technical field of gas turbine control, and in particular to a method for determining an abnormality of a gas turbine and a gas turbine. Background Art

[0002] Gas turbines have the advantages of light weight, high power, small overall size, ideal acceleration and strong maneuverability. Based on these advantages, gas turbines have gradually become dominant in the power applications of large and medium-sized surface ships. The development and progress of gas turbines is one of the important signs that the navy is gradually moving towards modernization. In addition, the high efficiency, short start-up time and fast response speed of gas turbines have gradually made them one of the main power sources for many applications such as power generation, gas compression and large-scale pumping services, such as common offshore drilling platform power generation and natural gas pressurized transportation. The advantages of gas turbines such as efficient energy utilization, resource conservation and emission reduction have made great contributions to the sustainable development of the contemporary economy. With the country's increasing attention to renewable energy, the monitoring and perception requirements for gas turbines have gradually increased. Therefore, ensuring the efficient and safe operation of gas turbines has gradually become one of the key directions of current attention.

[0003] Gas turbines are generally composed of several subsystems and numerous components. By sensing the status parameters and performance indicators of each component of the gas turbine, the operating status of the gas turbine can be understood in a timely manner, thereby realizing the analysis and judgment of the overall status of the gas turbine. At present, the staff use the operating status data to analyze and judge the abnormal conditions of the gas turbine based on their own experience. This method is prone to misjudgment by personnel, and even if an accurate judgment can be made, it is time-consuming and inefficient. Summary of the invention

[0004] In response to the above-mentioned problems and technical needs, the applicant has proposed a gas turbine abnormality determination method and a gas turbine to solve the problems of poor accuracy, low efficiency and long time consumption in the prior art when manually performing abnormality determination of gas turbines, and to ensure that the determination of gas turbine abnormalities is completed automatically, accurately and quickly.

[0005] The embodiment of the present application provides a method for determining abnormality of a gas turbine, the method comprising:

[0006] Acquiring operating data and standard control limits corresponding to the current operating conditions, wherein the operating data includes: system operating data of an operating system corresponding to the gas turbine and equipment operating data of an operating equipment;

[0007] Processing the operation data based on dynamic fractal theory and kernel principal component analysis method to obtain mapping data;

[0008] Calculating a reference indicator corresponding to the mapping data, and obtaining a control limit corresponding to the reference indicator based on a pre-created correspondence between the reference indicator and the control limit;

[0009] In the case where it is determined that the control limit value is greater than the standard control limit value, determining that the gas turbine has an abnormality under the current operating condition and the current time period;

[0010] The contribution rate of each data point of the mapping data is calculated, and the target corresponding to the data point whose contribution rate is greater than a preset contribution rate is determined to be abnormal, wherein the target includes a target operating system and / or a target operating device of the gas turbine.

[0011] According to a method for determining an abnormality of a gas turbine according to an embodiment of the present application, the reference index is obtained based on the first sub-index and the second sub-index;

[0012] The first sub-indicator is obtained based on a first calculation formula, and the second sub-indicator is obtained based on a second calculation formula;

[0013] The first calculation sub-formula is obtained based on Hotelling's T square distribution, and the second calculation sub-formula is obtained based on a squared prediction error method.

[0014] According to a method for determining an abnormality of a gas turbine according to an embodiment of the present application, calculating a reference index corresponding to the mapping data includes:

[0015] Determine a mapping matrix corresponding to the mapping data, and calculate diagonal eigenvalues ​​of the mapping matrix;

[0016] Inputting the diagonal eigenvalue into the first calculation formula to obtain the first sub-index output by the first calculation formula;

[0017] Wherein, the first calculation formula includes:

[0018]

[0019] Among them, T 2 represents the first sub-index, t γ represents the diagonal eigenvalues, An inverse matrix representing the diagonal eigenvalues ​​of the mapping matrix;

[0020] Inputting the mapping matrix into the second calculation formula to obtain the second sub-indicator output by the second calculation formula;

[0021] Wherein, the second calculation formula includes:

[0022]

[0023] Among them, SPE represents the second sub-index, t k represents the mapping matrix, p represents the row dimension corresponding to the mapping data after screening, and γ represents the row dimension corresponding to the mapping matrix;

[0024] Inputting the first sub-indicator and the second sub-indicator into a third calculation formula to obtain the reference indicator output by the third calculation formula;

[0025] Wherein, the third calculation formula includes:

[0026]

[0027] Among them, F in represents the reference index, χ T and δ spe is a constant.

[0028] According to an abnormality determination method for a gas turbine according to an embodiment of the present application, calculating the contribution rate of each data point of the mapping data includes:

[0029] Inputting the mapping data into a fourth calculation formula to obtain the contribution rate output by the fourth calculation formula;

[0030] Wherein, the fourth calculation formula includes:

[0031]

[0032] Among them, C C,j represents the contribution rate of the jth data point in the mapping data, t k represents the mapping data, λ k represents the eigenvalue corresponding to the mapping data, m0 represents the number of data points after the mapping data is screened, c j represents the running data corresponding to the jth data point, and γ represents the row dimension corresponding to the filtered mapped data.

[0033] According to an abnormality determination method for a gas turbine according to an embodiment of the present application, the contribution rate includes a target contribution rate, and the preset contribution rate includes a preset abnormality contribution rate;

[0034] Determining the target corresponding to the data point whose contribution rate is greater than the preset contribution rate as abnormal includes:

[0035] determining a target contribution rate of the operation data corresponding to the mapping data based on the contribution rate of the mapping data;

[0036] The target contribution rate that is greater than the target corresponding to the preset abnormal contribution rate is determined as abnormal.

[0037] According to an abnormality determination method for a gas turbine according to an embodiment of the present application, determining that the target contribution rate is greater than a target corresponding to a preset abnormal contribution rate is abnormal, including:

[0038] Draw a contribution rate distribution diagram of the operating data and the target contribution rate;

[0039] Comparing the contribution rate distribution graph with a preset contribution rate distribution graph to obtain a comparison result;

[0040] The contribution rate distribution graph is greater than the corresponding operation data in the preset contribution rate distribution graph and is determined to be abnormal, and the target corresponding to the abnormal operation data is determined to be abnormal.

[0041] According to an abnormality determination method for a gas turbine according to an embodiment of the present application, after determining that a target corresponding to a data point whose contribution rate is greater than a preset contribution rate is abnormal, the method further includes:

[0042] Inputting abnormal operation data and targets corresponding to abnormal data points into a pre-trained abnormality discrimination model to obtain the abnormal cause output by the abnormality discrimination model;

[0043] The abnormality discrimination model is trained based on abnormal data samples, target samples and abnormal cause samples.

[0044] According to a method for determining an abnormality of a gas turbine according to an embodiment of the present application, the method further includes:

[0045] When it is determined that the control limit is less than or equal to the standard control limit, it is determined that there is no abnormality in the gas turbine under the current operating conditions and the current time period, and the step of processing the operating data based on the dynamic fractal theory and the kernel principal component analysis method to obtain mapping data is performed on the operating data of the next time period.

[0046] According to an abnormality determination method for a gas turbine according to an embodiment of the present application, the operation data is processed based on the dynamic fractal theory and the kernel principal component analysis method to obtain mapping data, including:

[0047] Processing the operating data based on the dynamic fractal theory to construct a dynamic matrix representing that the autocorrelation and cross-correlation of the operating data are minimized;

[0048] Based on the kernel principal component analysis method, the dynamic matrix is ​​mapped to a preset high-dimensional feature space to obtain the mapping data.

[0049] An embodiment of the present application further provides a gas turbine, comprising: a gas turbine body and a controller, wherein the controller is used to execute the abnormality determination method of the gas turbine as described in any of the above embodiments.

[0050] The abnormality determination method for a gas turbine and the gas turbine provided in the embodiment of the present application obtain the operating data of the gas turbine and the standard control limit corresponding to the current operating condition; process the operating data based on the dynamic fractal theory and the kernel principal component analysis method to obtain mapping data. The present application obtains mapping data for characterizing the main dynamic characteristics of the gas turbine based on the dynamic fractal theory and the kernel principal component analysis method, reduces the analysis dimension, and improves the calculation speed; further, calculates the reference index corresponding to the mapping data, and obtains the control limit corresponding to the reference index based on the correspondence relationship between the reference index and the control limit created in advance. The present application calculates the corresponding control limit for different mapping data. The reference index and control limit provide an effective data basis for subsequent abnormality determination; when it is determined that the control limit is greater than the standard control limit, it is determined that the gas turbine has an abnormality under the current operating conditions and the current time period; the contribution rate of each data point of the mapped data is calculated, and the target operating system and / or target operating equipment corresponding to the data point whose contribution rate is greater than the preset contribution rate is determined to be abnormal, thereby achieving effective determination of abnormal targets. It can be seen that the present application can automatically, accurately and quickly complete the determination of gas turbine abnormalities and abnormal targets, and solves the problems of poor accuracy, low efficiency and long time consumption in the prior art when manually determining the abnormality of the gas turbine. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0052] Figure 1 It is one of the flow charts of the abnormality determination method of the gas turbine provided in the embodiment of the present application;

[0053] Figure 2 This is a second flow chart of the abnormality determination method for a gas turbine provided in an embodiment of the present application;

[0054] Figure 3 This is a third flow chart of the abnormality determination method for a gas turbine provided in an embodiment of the present application;

[0055] Figure 4 It is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] The embodiment of the present application provides a method for determining abnormality of a gas turbine. The method can be applied to a smart terminal, a server, or a controller of a gas turbine. The present application uses the application of the method to a server as an example for illustration. Some other descriptions in the embodiment of the present application are also for illustration and are not used to limit the scope of protection of the present application. They will not be described one by one later. The specific implementation of the method is as follows: Figure 1 As shown:

[0058] Step 101, obtaining operating data and standard control limits corresponding to the current operating conditions.

[0059] The operation data includes: system operation data of the operation system corresponding to the gas turbine and equipment operation data of the operation equipment.

[0060] Specifically, it is necessary to determine the current operating condition corresponding to the current time period and obtain the standard control limit corresponding to the current operating condition.

[0061] Step 102: Process the operation data based on the dynamic fractal theory and the kernel principal component analysis method to obtain mapping data.

[0062] Step 103: Calculate the reference index corresponding to the mapping data, and obtain the control limit corresponding to the reference index based on the pre-created correspondence between the reference index and the control limit.

[0063] Step 104: When it is determined that the control limit is greater than the standard control limit, it is determined that an abnormality exists in the gas turbine under the current operating condition and the current time period.

[0064] Step 105 , calculating the contribution rate of each data point of the mapping data, and determining the target corresponding to the data point whose contribution rate is greater than a preset contribution rate as abnormal.

[0065] The target includes a target operating system and / or a target operating device of the gas turbine.

[0066] The abnormality determination method for a gas turbine provided in an embodiment of the present application obtains the operating data of the gas turbine and the standard control limit corresponding to the current operating condition; processes the operating data based on the dynamic fractal theory and the kernel principal component analysis method to obtain mapping data. The present application obtains mapping data for characterizing the main dynamic characteristics of the gas turbine based on the dynamic fractal theory and the kernel principal component analysis method, reduces the analysis dimension, and improves the calculation speed; further, calculates the reference index corresponding to the mapping data, and obtains the control limit corresponding to the reference index based on the correspondence between the reference index and the control limit created in advance. The present application calculates the reference index corresponding to different mapping data. The standard and control limits provide an effective data basis for subsequent abnormality determination; when it is determined that the control limit is greater than the standard control limit, it is determined that the gas turbine is abnormal under the current operating conditions and the current time period; the contribution rate of each data point of the mapped data is calculated, and the target operating system and / or target operating equipment corresponding to the data point with a contribution rate greater than the preset contribution rate is determined to be abnormal, thereby achieving effective determination of abnormal targets. It can be seen that the present application can automatically, accurately and quickly complete the determination of gas turbine abnormalities and abnormal targets, and solves the problems of poor accuracy, low efficiency and long time consumption in the prior art when manually determining the abnormality of the gas turbine.

[0067] In a specific embodiment, the specific implementation of processing the operation data based on the dynamic fractal theory and the kernel principal component analysis method to obtain the mapping data includes:

[0068] The operating data is processed based on the dynamic fractal theory, and a dynamic matrix is ​​constructed to characterize the operating data with minimized autocorrelation and cross-correlation. The dynamic matrix is ​​mapped to a preset high-dimensional feature space based on the kernel principal component analysis method to obtain the mapping data.

[0069] Specifically, the operation data is standardized to obtain the standardized operation data, and the standardized operation data is processed using the dynamic fractal theory and the kernel principal component analysis method.

[0070] In a specific embodiment, the specific implementation of processing the operation data based on the dynamic fractal theory and constructing a dynamic matrix that minimizes the autocorrelation and cross-correlation of the operation data is as follows: Figure 2 As shown:

[0071] Step 201, setting the initial value of the hysteresis time to zero.

[0072] Step 202, reconstructing the matrix based on the running data to obtain the target matrix.

[0073] Step 203, using the fractal theory formula to calculate the correlation dimension of the target matrix when the initial value is zero.

[0074] Among them, the fractal theory formula includes formula (1) and formula (2).

[0075]

[0076] Among them, C d (ε) represents the intermediate quantity, N s represents the number of measurements, is a constant, H represents the step function, ε represents the distance parameter, δ ij Represents the Euclidean distance between the internal row and the row state vector of the target matrix.

[0077] in, ε value and δ ij Related.

[0078] Among them, as ε changes, δ ij tends to zero and gradually decreases, for ε and C d (ε) and take the logarithm at the same time to obtain the correlation dimension, see formula (2) for details:

[0079]

[0080] Among them, C d Represents an associated dimension.

[0081] Specifically, in the actual analysis process, the value range of ε does not reach the defined limit of 0. In order to obtain the optimal C d , with lgε as the horizontal axis, lgC d (ε) is the ordinate of the calculated δ ij Make a lgε-lgC(ε) function curve with ε. According to the characteristics of the image, the slope of the straight line in the better linear area on the function curve is used as the correlation dimension.

[0082] Step 204, calculating matrix construction iteration values.

[0083] Specifically, the associated dimensions obtained above are input into the iteration value calculation formulas (3) and (4) to obtain the matrix constructed iteration value output by the iteration value calculation formula.

[0084] r=m(l+1)-C d …………………………(3)

[0085]

[0086] Among them, r represents the matrix construction iteration value before the current iteration operation, r n It represents the matrix construction iteration value after the current iteration operation, l represents the hysteresis time, and m represents the number of data points.

[0087] Step 205 , determining whether the matrix construction iteration value is less than or equal to zero, if so, executing step 206 , otherwise executing step 207 .

[0088] Step 206: output the dynamic matrix.

[0089] That is, the target matrix at the current iteration.

[0090] Step 207, a new hysteresis time is obtained based on the preset step length, and the process returns to step 202.

[0091] Here, the preset step size is taken as 1 as an example for explanation.

[0092] Specifically, the hysteresis times corresponding to different correlation dimensions are recorded once during the iteration process, the optimal hysteresis time is determined from multiple hysteresis times, and a dynamic matrix in which the autocorrelation and cross-correlation with the operating data are minimized is constructed.

[0093] The dynamic matrix is

[0094] Among them, x t represents the running data corresponding to time t, l opt Indicates the optimal hysteresis time.

[0095] In a specific embodiment, the dynamic matrix is ​​mapped to a preset high-dimensional feature space based on the kernel principal component analysis method (KPCA), and the specific implementation of obtaining the mapping data includes:

[0096] Specifically, by selecting an appropriate kernel width, the mapping data can be processed and analyzed more easily. For the mapping data, the covariance matrix can be used to represent the probability density of multidimensional random variables in the feature space. The eigenvalues ​​and eigenvectors of the covariance matrix are solved, and specifically obtained by formula (5):

[0097] Hp k =λ k p k (k=1,...,N)………………(5)

[0098] Where H represents the covariance matrix, λ k represents the eigenvalue, represents the feature vector, is the weighting coefficient of the high-dimensional feature space, is a constant, Φ(x i ) is the mapping function.

[0099] Substituting the parameters into formula (5) yields formula (6):

[0100]

[0101] Where N represents the row dimension of the input data.

[0102] For the sake of simplicity, let K ij = k(xi ,x j )=<Φ(x i ),Φ(x j )>=Φ(x i ) T Φ(x j ), introduce the radial Gaussian kernel function, k(x i ,x j ) = exp(-||x i -x j || 2 / 2ξ 2 ), simplify formula (6) to obtain formula (7):

[0103] Kα k =Nλ k α k

[0104] in,

[0105] In order to make the data more concentrated and reduce the deviation, the assumption that the data is centered at the preset point in the high-dimensional feature space (mapping space) is satisfied. In order to ensure the centralization of the mapping data, the Gaussian kernel function is required to centralize and obtain the mapping data of the running data corresponding to the kth time point on the feature vector, which is obtained by formula (8):

[0106]

[0107] Among them, t k Represents mapping data.

[0108] Specifically, in order not to affect the overall result analysis, the characteristic values ​​are screened using a preset numerical range, which is generally σ0=10 -6 , and then get the filtered mapping data p represents the row dimension of the filtered mapping data.

[0109] In order to determine the size of the final principal component and to further reduce the analysis dimension of the data, fractal theory is applied to T p Calculate and obtain the mapping principal element matrix. For details, see formula (9) and formula (10):

[0110] d x =C d (T p )……………………………………(9)

[0111] d τ =C d (τ)…………………………(10)

[0112] Among them, d x Indicates the use of T p The obtained correlation dimension, τ is a set of principal component judgment data, and τ = T p A, the matrix data A is an orthogonal matrix consisting of the eigenvectors of the covariance matrix.

[0113] By γ = max([d x ],[d τ ])[] is rounded up to get the final mapping matrix

[0114] In a specific embodiment, the reference indicator is obtained based on the first sub-indicator and the second sub-indicator; the first sub-indicator is obtained based on the first calculation formula, and the second sub-indicator is obtained based on the second calculation formula; the first calculation sub-formula is obtained based on Hotelling's T-square distribution (T-square statistic), and the second calculation sub-formula is obtained based on the square prediction error method.

[0115] Among them, Hotelling T square distribution is a univariate distribution, and the squared prediction error method (SPE) is used to monitor abnormal conditions in the production process.

[0116] In a specific embodiment, the specific implementation of calculating the reference index corresponding to the mapping data includes:

[0117] A mapping matrix corresponding to the mapping data is determined, and the diagonal eigenvalues ​​of the mapping matrix are calculated. The diagonal eigenvalues ​​are input into a first calculation formula to obtain a first sub-index output by the first calculation formula.

[0118] The first calculation formula is shown in formula (11):

[0119]

[0120] Among them, T 2 represents the first sub-index, t γ represents the diagonal eigenvalues, Represents the inverse matrix consisting of the diagonal eigenvalues ​​of the mapping matrix.

[0121] The mapping matrix is ​​input into the second calculation formula to obtain the second sub-index output by the second calculation formula.

[0122] The second calculation formula is shown in formula (12):

[0123]

[0124] Among them, SPE represents the second sub-index, t k Represents the mapping data, p represents the row dimension corresponding to the mapping data after screening, and γ represents the row dimension corresponding to the mapping matrix.

[0125] The first sub-indicator and the second sub-indicator are input into the third calculation formula to obtain a reference indicator output by the third calculation formula.

[0126] The third calculation formula is shown in formula (13):

[0127]

[0128] Among them, F in represents the reference index, χ T and δ spe is a constant.

[0129] Specifically, when monitoring abnormal situations, reference indicators can be used for abnormal analysis. The reference indicators are mainly obtained through Hotelling T square distribution and square prediction error method.

[0130] SPE is a statistic based on reconstruction error, which is used to measure the difference between the original data points and their reconstructed data. The SPE statistic represents the sum of squares of the reconstruction errors of the data points in the principal component space. A large SPE value may indicate that the data points of the gas turbine deviate from the normal mode and there are abnormalities.

[0131] The T-squared statistic is used to measure the overall deviation of the data in the principal component space. It takes into account the projection distance of the data points from the model, as well as the weights and variances of the principal components. A larger T 2 The values ​​may indicate that the operating data of the gas turbine deviates from the normal mode as a whole and there are anomalies in multiple principal component directions.

[0132] In a specific embodiment, the specific implementation of calculating the contribution rate of each data point of the mapping data includes:

[0133] The mapping data is input into the fourth calculation formula to obtain the contribution rate output by the fourth calculation formula.

[0134] The fourth calculation formula is shown in formula (14):

[0135]

[0136] Among them, C C,j represents the contribution rate of the jth data point in the mapping data, t k represents the mapping data, λ k represents the eigenvalue corresponding to the mapping data, m0 represents the number of data points after filtering the mapping data, c j represents the running data corresponding to the jth data point, and γ represents the row dimension corresponding to the filtered mapped data.

[0137] in,

[0138] In a specific embodiment, the contribution rate includes a target contribution rate, and the preset contribution rate includes a preset abnormal contribution rate.

[0139] The specific implementation of determining the target corresponding to the data point with a contribution rate greater than the preset contribution rate as abnormal includes:

[0140] Based on the contribution rate of the mapping data, a target contribution rate of the operation data corresponding to the mapping data is determined; and a target contribution rate greater than a target corresponding to a preset abnormal contribution rate is determined to be abnormal.

[0141] In a specific embodiment, the specific implementation of determining that a target corresponding to a target contribution rate greater than a preset abnormal contribution rate is abnormal includes:

[0142] Draw a contribution rate distribution graph of the operating data and the target contribution rate; compare the contribution rate distribution graph with the preset contribution rate distribution graph to obtain a comparison result; determine that the corresponding operating data in the contribution rate distribution graph that is greater than the preset contribution rate distribution graph is abnormal, and determine the target corresponding to the abnormal operating data is abnormal.

[0143] Specifically, there is a large correlation between the abnormal operating data and the nonlinear principal element in the mapping space, so the contribution rate distribution map can be used to locate the abnormality of the gas turbine. The judgment basis is that the contribution rate of the operating data is closely related to the degree of change of the operating data of the gas turbine, and the abnormality perception and positioning are achieved by analyzing the contribution rate of each monitoring target.

[0144] Specifically, the target contribution rate of the operating data corresponding to the mapping data is obtained using formula (15):

[0145]

[0146] Where R = [R1, R2, ..., R m ], represents the target contribution rate corresponding to each data point in the operation data, m represents the dimension of the operation data, OC C,i As a whole, it represents the target contribution rate corresponding to the i-th dimension in the operating data.

[0147] Specifically, through Figure 3 The calculation of target contribution rate is explained in detail:

[0148] Step 301: input the contribution rate corresponding to the mapping data, and set i and j to 1 as initial values.

[0149] Step 302, determine whether i is greater than m, if so, execute step 303, otherwise, execute step 304.

[0150] Step 303: Taking the contribution rate corresponding to the current operation as the target contribution rate.

[0151] Step 304: Set the target contribution rate corresponding to the i-th dimension in the operating data to zero.

[0152] Step 305 , determining whether j is less than the number of data points after the mapping data is screened, if so, executing step 306 , otherwise, executing step 307 .

[0153] Step 306, execute calculation formula (16).

[0154] Wherein, formula (16) is:

[0155] OC C,i =OC C,i +C C,j ………………(16)

[0156] Step 307, executing i=i+1, and outputting the target contribution rate corresponding to the i-th dimension in the operating data.

[0157] Step 308, execute j=j+m.

[0158] In a specific embodiment, after the target corresponding to the data point whose contribution rate is greater than the preset contribution rate is determined to be abnormal, the abnormal operating data and target corresponding to the abnormal data point are input into a pre-trained abnormality discrimination model to obtain the abnormal cause output by the abnormality discrimination model.

[0159] Among them, the anomaly discrimination model is trained based on abnormal data samples, target samples and abnormal cause samples.

[0160] Specifically, the present application determines the abnormal conditions and locations of abnormalities during the operation of the gas turbine based on the contribution rate distribution diagram, and then uses the expert knowledge system (abnormality discrimination model) to determine the cause of the abnormality, providing an effective solution for abnormality resolution and improving user experience.

[0161] In a specific embodiment, when it is determined that the control limit is less than or equal to the standard control limit, it is determined that there is no abnormality in the gas turbine under the current operating conditions and the current time period, and the operating data of the next time period is processed based on the dynamic fractal theory and the kernel principal component analysis method to obtain the mapping data.

[0162] Specifically, the standard control limit of the present application is obtained by processing sample data based on dynamic theory, fractal theory and kernel principal component analysis, wherein the sample data is historical operation data obtained under normal operation of the gas turbine.

[0163] Specifically, this application first combines dynamic theory and fractal theory to obtain the dynamic characteristics of multi-source data of the gas turbine system; secondly, it uses kernel principal component analysis and fractal theory to map, analyze, calculate and extract the data, and extracts the principal component feature mapping data matrix containing the main dynamic characteristics of the gas turbine; then, it applies reference indicators and contribution rate distribution diagrams to realize the evaluation and analysis of the overall system operation status, and at the same time combines with the constructed expert knowledge system to further determine the cause of the abnormality of the gas turbine and the corresponding solution. This method reduces the data analysis dimension while retaining the dynamic characteristics of the data, and the evaluation and analysis indicators can effectively realize the perception and diagnosis of the system operation status, thereby improving the efficiency of the system intelligent operation and maintenance.

[0164] This application applies fractal theory to analyze the mapping data of the gas turbine. While extracting the main features of the input data, it effectively reduces the data calculation dimension, thereby greatly reducing the difficulty of overall data analysis and improving data processing speed. The dynamic characteristics of the overall data of the gas turbine are accurately obtained through fractal theory and dynamic theory. While effectively retaining the dynamic characteristics of the system data, the accuracy and reliability of the perception of the operating status of the gas turbine can be effectively improved. The reference index, control limit and contribution rate distribution map are used to locate the abnormality of the gas turbine, and the source of the abnormality can be found accurately and quickly.

[0165] An embodiment of the present application further provides a gas turbine, comprising: a gas turbine body and a controller, wherein the controller is used to execute the abnormality determination method of the gas turbine as described in any of the above embodiments.

[0166] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4As shown, the electronic device may include: a processor (processor) 401, a communication interface (Communications Interface) 402, a memory (memory) 403 and a communication bus 404, wherein the processor 401, the communication interface 402, and the memory 403 communicate with each other through the communication bus 404. The processor 401 can call the logic instructions in the memory 403 to execute the abnormality determination method of the gas turbine, which method includes: obtaining operating data and standard control limits corresponding to the current operating conditions, wherein the operating data includes: system operating data of the operating system corresponding to the gas turbine and equipment operating data of the operating equipment; processing the operating data based on the dynamic fractal theory and the kernel principal component analysis method to obtain mapping data; calculating the reference index corresponding to the mapping data, and obtaining the control limit corresponding to the reference index based on the correspondence between the reference index and the control limit created in advance; when it is determined that the control limit is greater than the standard control limit, determining that the gas turbine has an abnormality under the current operating conditions and the current time period; calculating the contribution rate of each data point of the mapping data, and determining the target corresponding to the data point whose contribution rate is greater than the preset contribution rate as abnormal, wherein the target includes the target operating system and / or target operating equipment of the gas turbine.

[0167] In addition, the logic instructions in the above-mentioned memory 403 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0168] On the other hand, the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the gas turbine abnormality determination method provided by the above methods, the method comprising: obtaining operating data and standard control limits corresponding to the current operating conditions, wherein the operating data include: system operating data of the operating system corresponding to the gas turbine and equipment operating data of the operating equipment; processing the operating data based on dynamic fractal theory and kernel principal component analysis method to obtain mapping data; calculating a reference index corresponding to the mapping data, and obtaining a control limit corresponding to the reference index based on a correspondence between the reference index and the control limit created in advance; when it is determined that the control limit is greater than the standard control limit, determining that the gas turbine has an abnormality under the current operating conditions and the current time period; calculating the contribution rate of each data point of the mapping data, and determining the target corresponding to the data point whose contribution rate is greater than the preset contribution rate as abnormal, wherein the target includes a target operating system and / or target operating equipment of the gas turbine.

[0169] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the processor executes the method for determining an abnormality of a gas turbine provided in the above-mentioned embodiments, the method comprising: obtaining operating data and standard control limits corresponding to the current operating conditions, wherein the operating data comprises: system operating data of an operating system corresponding to the gas turbine and equipment operating data of an operating device; processing the operating data based on dynamic fractal theory and kernel principal component analysis method to obtain mapping data; calculating a reference index corresponding to the mapping data, and obtaining a control limit corresponding to the reference index based on a correspondence between the reference index and the control limit created in advance; determining that the gas turbine has an abnormality under the current operating conditions and the current time period when it is determined that the control limit is greater than the standard control limit; calculating a contribution rate of each data point of the mapping data, and determining a target corresponding to a data point whose contribution rate is greater than a preset contribution rate as abnormal, wherein the target comprises a target operating system and / or a target operating device of the gas turbine.

[0170] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0171] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0172] Finally, it should be noted that the above is only the preferred implementation of the present application, and the present application is not limited to the above embodiments. It is understood that other improvements and changes directly derived or associated by those skilled in the art without departing from the spirit and concept of the present application should be considered to be included in the scope of protection of the present application.

Claims

1. A method for determining abnormality of a gas turbine, characterized in that: The method comprises: Acquiring operating data and standard control limits corresponding to the current operating conditions, wherein the operating data includes: system operating data of an operating system corresponding to the gas turbine and equipment operating data of an operating equipment; Processing the operation data based on dynamic fractal theory and kernel principal component analysis method to obtain mapping data; Calculating a reference indicator corresponding to the mapping data, and obtaining a control limit corresponding to the reference indicator based on a pre-created correspondence between the reference indicator and the control limit; In the case where it is determined that the control limit value is greater than the standard control limit value, determining that the gas turbine has an abnormality under the current operating condition and the current time period; The contribution rate of each data point of the mapping data is calculated, and the target corresponding to the data point whose contribution rate is greater than a preset contribution rate is determined to be abnormal, wherein the target includes a target operating system and / or a target operating device of the gas turbine.

2. The gas turbine abnormality determination method according to claim 1, characterized in that: The reference indicator is obtained based on the first sub-indicator and the second sub-indicator; The first sub-indicator is obtained based on a first calculation formula, and the second sub-indicator is obtained based on a second calculation formula; The first calculation sub-formula is obtained based on Hotelling's T square distribution, and the second calculation sub-formula is obtained based on a squared prediction error method.

3. The method for determining abnormality of a gas turbine according to claim 2, wherein: Calculating a reference index corresponding to the mapping data includes: Determine a mapping matrix corresponding to the mapping data, and calculate diagonal eigenvalues ​​of the mapping matrix; Inputting the diagonal eigenvalue into the first calculation formula to obtain the first sub-index output by the first calculation formula; Wherein, the first calculation formula includes: Among them, T 2 represents the first sub-index, t γ represents the diagonal eigenvalues, An inverse matrix representing the diagonal eigenvalues ​​of the mapping matrix; Inputting the mapping matrix into the second calculation formula to obtain the second sub-indicator output by the second calculation formula; Wherein, the second calculation formula includes: Among them, SPE represents the second sub-index, t k represents the mapping matrix, p represents the row dimension corresponding to the mapping data after screening, and γ represents the row dimension corresponding to the mapping matrix; Inputting the first sub-indicator and the second sub-indicator into a third calculation formula to obtain the reference indicator output by the third calculation formula; Wherein, the third calculation formula includes: Among them, F in represents the reference index, χ T and δ spe is a constant.

4. The method for determining abnormality of a gas turbine according to any one of claims 1 to 3, characterized in that: Calculating the contribution rate of each data point of the mapping data, including: Inputting the mapping data into a fourth calculation formula to obtain the contribution rate output by the fourth calculation formula; Wherein, the fourth calculation formula includes: Among them, C C,j represents the contribution rate of the jth data point in the mapping data, t k represents the mapping data, λ k represents the eigenvalue corresponding to the mapping data, m0 represents the number of data points after the mapping data is screened, c j represents the running data corresponding to the jth data point, and γ represents the row dimension corresponding to the filtered mapped data.

5. The gas turbine abnormality determination method according to claim 4, characterized in that: The contribution rate includes a target contribution rate, and the preset contribution rate includes a preset abnormal contribution rate; Determining the target corresponding to the data point whose contribution rate is greater than the preset contribution rate as abnormal includes: determining a target contribution rate of the operation data corresponding to the mapping data based on the contribution rate of the mapping data; The target contribution rate that is greater than the target corresponding to the preset abnormal contribution rate is determined as abnormal.

6. The gas turbine abnormality determination method according to claim 4, characterized in that: The target contribution rate being greater than the target corresponding to the preset abnormal contribution rate is determined as abnormal, including: Draw a contribution rate distribution diagram of the operating data and the target contribution rate; Comparing the contribution rate distribution graph with a preset contribution rate distribution graph to obtain a comparison result; The contribution rate distribution graph is greater than the corresponding operation data in the preset contribution rate distribution graph and is determined to be abnormal, and the target corresponding to the abnormal operation data is determined to be abnormal.

7. The method for determining abnormality of a gas turbine according to any one of claims 1 to 3, characterized in that: After determining that the target corresponding to the data point whose contribution rate is greater than the preset contribution rate is abnormal, the method further includes: Inputting abnormal operation data and targets corresponding to abnormal data points into a pre-trained abnormality discrimination model to obtain the abnormal cause output by the abnormality discrimination model; The abnormality discrimination model is trained based on abnormal data samples, target samples and abnormal cause samples.

8. The method for determining abnormality of a gas turbine according to any one of claims 1 to 3, characterized in that: The method further comprises: When it is determined that the control limit is less than or equal to the standard control limit, it is determined that there is no abnormality in the gas turbine under the current operating conditions and the current time period, and the step of processing the operating data based on the dynamic fractal theory and the kernel principal component analysis method to obtain mapping data is performed on the operating data of the next time period.

9. The method for determining abnormality of a gas turbine according to any one of claims 1 to 3, characterized in that: The operation data is processed based on the dynamic fractal theory and the kernel principal component analysis method to obtain mapping data, including: Processing the operating data based on the dynamic fractal theory to construct a dynamic matrix representing that the autocorrelation and cross-correlation of the operating data are minimized; Based on the kernel principal component analysis method, the dynamic matrix is ​​mapped to a preset high-dimensional feature space to obtain the mapping data.

10. A gas turbine, characterized in that: include: A gas turbine body and a controller, wherein the controller is used to execute the abnormality determination method of the gas turbine according to any one of claims 1 to 9.

Images