Thermal generator set fault detection method and system

Abstract

The invention relates to the technical field of electrical fault detection devices, in particular to a thermal generator set fault detection method and system, and the method comprises the steps: obtaining the numerical value of each parameter at each moment in the operation process of a thermal generator set, forming a parameter vector at each moment through the numerical values of all parameters at each moment, according to the method, a parameter estimation vector at each moment is determined, the suspected fault possibility at each moment is determined by combining the correlation between any two parameters so as to distinguish the suspected fault moment and the normal moment, and the fault occurrence time period is finally determined according to the numerical values of the parameters at the suspected fault moment and the normal moment. According to the method, the suspected fault time is determined through correlation analysis among various parameters in the operation process of the thermal generator set, so that the accuracy of fault occurrence time period detection is guaranteed.

Description

Technical Field

[0001] The invention relates to the technical field of electrical fault detection devices, and in particular to a method and system for detecting faults of a thermal power generator set. Background Art

[0002] Thermal power generation is the main form of power production at present, and plays a role in ensuring the supply of power in the power system. Due to its large capacity and high parameters, once a thermal power generator fails and causes unplanned downtime, it will have a serious adverse impact on the stable operation of the power grid, reliable power supply, and safe operation of equipment, causing unnecessary economic losses at the least and casualties at the worst.

[0003] Existing problem: In the current fault detection of thermal power generating units, judgment is often made based on the set threshold of each parameter. However, in actual operation, multiple parameters may jointly affect the state of the thermal power generating unit. The threshold based only on the parameter size may not fully reflect this complex relationship and may lead to underreporting of thermal power generating unit faults. Summary of the invention

[0004] The present invention provides a method and system for detecting faults of a thermal power generating set to solve the existing problems.

[0005] A method and system for detecting a fault of a thermal power generating set of the present invention adopts the following technical solutions:

[0006] An embodiment of the present invention provides a method for detecting a fault of a thermal power generator set, the method comprising the following steps:

[0007] Obtain the value of each parameter at every moment during the operation of the thermal power generating unit;

[0008] The parameter vector at each moment is formed by the values ​​of all parameters at each moment; the dynamic process matrix of the normal operation of the thermal power generating set is formed according to the parameter vectors at consecutive moments; the parameter estimation vector at each moment is determined according to the parameter vector at each moment and the dynamic process matrix of the normal operation of the thermal power generating set;

[0009] Determine the test statistic of the partial correlation analysis of any two parameters at each moment according to the correlation between all any two parameters in the continuous moments; determine the possibility of suspected fault at each moment according to the test statistic of the partial correlation analysis of any two parameters at each moment and the parameter estimation vector at each moment;

[0010] According to the possibility of the suspected fault at each moment, the suspected fault moment and the normal moment are determined; according to the values ​​of the parameters at the suspected fault moment and the normal moment, the fault occurrence time period is determined.

[0011] Furthermore, the dynamic process matrix constituting the normal operation of the thermal power generating set includes the following specific steps:

[0012] A first constant n is preset, a second constant m is preset, and in the period from the nth moment to the mth moment, the parameter vectors at each moment are arranged from top to bottom in chronological order to form a dynamic process matrix D for the normal operation of the thermal power generating set.

[0013] Furthermore, the specific steps of determining the parameter estimation vector at each moment are as follows:

[0014] The transposed vector of the parameter vector at time t is used as the state evaluation matrix X obs ;

[0015] Obtain the transposed matrix D of the dynamic process matrix D of the normal operation of the thermal power generating unit T , calculate the D and the D T The Euclidean distance L between them is calculated, and then the product of the reciprocal of L and D is calculated to obtain a first matrix D1; wherein T represents a transposition operation;

[0016] Calculate the D and X obs The product of , obtains the first matrix D2;

[0017] Get the transposed matrix D1 of D1 T , calculate the D1 T Multiply it with D2 to get the parameter estimation matrix X est ;

[0018] The X est The transposed matrix of is used as the parameter estimation vector at the tth moment.

[0019] Furthermore, the specific steps of determining the test statistic of the partial correlation analysis of any two parameters at each moment are as follows:

[0020] A third constant z is preset, and the period from the tzth moment to the tth moment is recorded as the reference period at the tth moment;

[0021] In the reference period at the t-th moment, the numerical values ​​of each parameter at all moments are used to form a numerical reference sequence of each parameter in chronological order, and the Pearson correlation coefficient of the numerical reference sequences of any two parameters is used as the correlation coefficient of any two parameters at the t-th moment;

[0022] Get any two parameters A and B, and based on the correlation coefficients of all the two parameters at the tth time, get the partial correlation coefficient between the parameters A and B at the tth time after eliminating the influence of all other parameters that are not the parameters A and B;

[0023] According to the partial correlation coefficient between parameters A and B at the tth moment after eliminating the influence of all other parameters that are not parameters A and B, the test statistic of the partial correlation analysis of parameters A and B at the tth moment is obtained.

[0024] Furthermore, the determination of the possibility of a suspected fault at each moment includes the following specific steps:

[0025] The parameter estimation vector at each moment includes: an estimated value of each parameter at each moment;

[0026] In the reference period at the t-th moment, the estimated values ​​of each parameter at all moments are used to form a reference sequence of estimated values ​​of each parameter in chronological order, and the Pearson correlation coefficient of the reference sequence of estimated values ​​of any two parameters is used as the correlation coefficient of the estimated values ​​of any two parameters at the t-th moment;

[0027] Calculate the correlation coefficient P of parameters A and B at time t t The correlation coefficient P between (A,B) and the estimated values ​​of parameters A and B at time t t ′ The absolute value of the difference between (A, B), and the normalized value of the product of the absolute value of the difference and the test statistic of the partial correlation analysis of the parameters A and B at the t-th time as the abnormal factor of the parameters A and B at the t-th time;

[0028] The mean of the abnormal factors of all two arbitrary parameters at the tth moment is taken as the suspected fault possibility at the tth moment.

[0029] Furthermore, the determination of the suspected fault time and the normal time includes the following specific steps:

[0030] When the suspected fault possibility at the tth moment is greater than the preset judgment threshold, the tth moment is recorded as the suspected fault moment;

[0031] When the suspected fault possibility at the tth moment is less than or equal to the preset judgment threshold, the tth moment is recorded as a normal moment.

[0032] Furthermore, the specific steps of determining the fault occurrence period are as follows:

[0033] The time period consisting of the consecutive suspected fault moments is recorded as the suspected fault time period;

[0034] Get the duration L of the gth suspected fault period g , starting from the g-th suspected fault period, traverse the time before the g-th suspected fault period one by one, and obtain the previous L g A normal moment, as a control normal moment;

[0035] The parameters at least include: shaft vibration; in chronological order, the values ​​of shaft vibration at all times within the g-th suspected fault period constitute a suspected fault shaft vibration sequence, and the values ​​of shaft vibration at all normal times constitute a control normal shaft vibration sequence;

[0036] Determining the fault factor of the g-th suspected fault period according to the suspected fault shaft vibration sequence and the reference normal shaft vibration sequence;

[0037] According to the fault factor of the g-th suspected fault period, it is determined whether the g-th suspected fault period is a fault occurrence period.

[0038] Furthermore, the determination of the fault factor of the g-th suspected fault period includes the following specific steps:

[0039] In the suspected fault shaft vibration sequence and the control normal shaft vibration sequence, the absolute value of the difference between the shaft vibration values ​​under the same sequence value is calculated, and the sum of the absolute value of the difference between the shaft vibration values ​​under all the same sequence values ​​is taken as the fault factor of the gth suspected fault period.

[0040] Furthermore, the determining whether the g-th suspected fault period is a fault occurrence period comprises the following specific steps:

[0041] When the normalized value of the fault factor of the g-th suspected fault period is greater than the preset second threshold, the g-th suspected fault period is recorded as the fault occurrence period.

[0042] The present invention also proposes a thermal power generator set fault detection system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program stored in the memory to implement the steps of the aforementioned thermal power generator set fault detection method.

[0043] The beneficial effects of the technical solution of the present invention are:

[0044] In an embodiment of the present invention, the value of each parameter at each moment in the operation process of the thermal power generator set is obtained, and the parameter vector at each moment is formed with the values ​​of all parameters at each moment, so as to determine the parameter estimation vector at each moment, thereby determining the best estimate of the current operating state of the thermal power generator set according to the comprehensive analysis of the current moment parameters and the normal operating parameters of the thermal power generator set, so as to ensure the accuracy of the subsequent suspected fault time distinction. Combined with the correlation between any two parameters, the suspected fault possibility at each moment is determined to distinguish the suspected fault moment from the normal moment, and the fault occurrence period is finally determined according to the values ​​of the parameters at the suspected fault moment and the normal moment, thereby further ensuring the accuracy of the suspected fault time distinction by analyzing the mutual influence between different parameters, so that the parameter difference at the suspected fault moment and the normal moment can be better used to determine whether it is a fault occurrence period. So far, the present invention determines the suspected fault moment by analyzing the correlation between multiple parameters during the operation of the thermal power generator set, thereby ensuring the accuracy of the fault occurrence period detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the embodiments of the present invention 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 only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 A flow chart of the steps of a method for detecting a fault of a thermal power generating set according to the present invention;

[0047] Figure 2 This is a schematic diagram of a bearing in a thermal power generator set;

[0048] Figure 3 This is a schematic diagram of the changes in power generation during the unplanned shutdown of a thermal power generator set. DETAILED DESCRIPTION

[0049] In order to further explain the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the following is a detailed description of a method and system for detecting a fault of a thermal power generator set proposed by the present invention, its specific implementation, structure, features and effects, in conjunction with the accompanying drawings and preferred embodiments. In the following description, different "one embodiment" or "another embodiment" does not necessarily refer to the same embodiment. In addition, specific features, structures or characteristics in one or more embodiments may be combined in any suitable form.

[0050] Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs.

[0051] The specific scheme of a method and system for detecting faults of a thermal power generating set provided by the present invention is described in detail below with reference to the accompanying drawings.

[0052] See also Figure 1 , which shows a flowchart of a method for detecting a fault of a thermal power generator set provided by an embodiment of the present invention, the method comprising the following steps:

[0053] Step S001: Obtain the value of each parameter at each moment during the operation of the thermal power generating set.

[0054] During the operation of the thermal power generating set, the value of each parameter is collected at each moment, where the parameters include: active power, line voltage, phase current, excitation current, reactive power, shaft vibration and bearing vibration.

[0055] It should be noted that the data at each moment mentioned above is collected when each parameter does not exceed the set threshold range. If any parameter exceeds the set threshold range, the machine will be shut down immediately. The collection frequency is once a minute. The active power and reactive power are measured by the power meter installed at the generator set outlet. The line voltage and phase current are measured by voltage transformers and current transformers. The excitation current is measured by the current sensor installed in the excitation system. Shaft vibration and bearing vibration are measured by vibration sensors installed on the unit shaft and bearing seat.

[0056] It should be further explained that: a thermal power generating set is composed of multiple systems, and each system is composed of several subsystems and complex equipment. Therefore, in this embodiment, the above-mentioned 7 representative parameters are selected for analysis. Among them, active power and reactive power are key indicators for measuring the power generation efficiency of a thermal power generating set. Line voltage is crucial to ensuring the stability and efficiency of power transmission. Phase current refers to the current flowing through each phase. Its stability and balance are crucial to the safe operation of the generator set. The excitation current is the current supplied to the rotor winding of the synchronous generator, which is used to establish a magnetic field and directly affects the voltage regulation and reactive power output of the generator. Shaft vibration refers to the vibration in the axial direction of the generator set. Excessive vibration may cause mechanical damage and reduced efficiency. Bearing vibration refers to the vibration of the bearing seat, which is also an important indicator for evaluating the mechanical state of the unit. Since there are many rotating equipment in a thermal power generating set, such as a rotating separator, a coal mill, etc., bearings are indispensable components in these rotating equipment, so shaft vibration and bearing vibration can intuitively reflect mechanical failures in the thermal power generating set. A schematic diagram of a bearing in a thermal power generating set, such as Figure 2When an important device or system in a thermal power generating unit fails, the unit will trip and stop working. This process is an unplanned shutdown of the thermal power generating unit. This is a process from normal load state to rapid load shedding until shutdown. The schematic diagram of power generation change during unplanned shutdown of the thermal power generating unit is shown in Figure 3 shown. Figure 3 The horizontal axis is time, in units of one second (1s), and the vertical axis is power generation, in units of megawatts (MW). The solid line on the axis represents the change of power generation over time. Figure 3 It can be seen that during normal operation, the power generation fluctuates within a small range. When a fault occurs, the power generation drops rapidly to 0, i.e., the system shuts down for protection.

[0057] Step S002: construct a parameter vector at each moment with the values ​​of all parameters at each moment; construct a dynamic process matrix for the normal operation of the thermal power generating set based on the parameter vectors at consecutive moments; determine a parameter estimation vector at each moment based on the parameter vector at each moment and the dynamic process matrix for the normal operation of the thermal power generating set.

[0058] Preferably, in one embodiment of the present invention, a method for obtaining a parameter estimation vector at each moment includes:

[0059] The parameter vector at each moment is constructed with the values ​​of all parameters at each moment.

[0060] For example: the parameter vector X at the i-th moment i =[x i,1 ,x i,2 ,x i,3 ,x i,4 ,x i,5 ,x i,6 ,x i,7 ], where x i,1 、x i,2 、x i,3 、x i,4 、x i,5 、x i,6 and x i,7 They are respectively the value of active power, the value of line voltage, the value of phase current, the value of excitation current, the value of reactive power, the value of shaft vibration and the value of bearing vibration at the i-th moment.

[0061] The first constant n is preset to be 10, and the second constant m is preset to be 60, wherein m>n, and this is taken as an example for description.

[0062] In the period from the nth moment to the mth moment, the parameter vectors of each moment are arranged from top to bottom in chronological order to form the dynamic process matrix D of the normal operation of the thermal power generating unit.

[0063] in, X n , X n+1 and X m are the parameter vectors at the nth moment, the n+1th moment, and the mth moment respectively.

[0064] It should be noted that the thermal power generating set in this embodiment is used as base load power generation, and the thermal power generating set is mainly responsible for providing the continuous basic load of the power grid. Therefore, the thermal power generating set will operate at a state close to full load for a long time to ensure the stability and reliability of the power supply. It can be seen that the failure is often caused by a long period of full load operation. Therefore, in this embodiment, when each parameter does not exceed the set threshold range, no further failure analysis is performed on the first hour after the thermal power generating set starts to operate. It is considered that the first hour after the thermal power generating set starts to operate is a normal operating state. Therefore, the parameter vectors at each moment in the period from the nth moment to the mth moment are taken to form the matrix D. The first 10 minutes after the start-up is the start-up stage, and each parameter will fluctuate and be unstable, so it is not included in the matrix D.

[0065] Taking the tth moment as an example, where t>m, that is, when each parameter does not exceed the set threshold range, further fault analysis is performed one hour after the thermal power generating unit starts to operate.

[0066] The parameter vector X at time t t The transposed vector X t T , as the state evaluation matrix X obs .

[0067] Wherein, T represents the transposition operation, which is a well-known technique for converting the row vector X t becomes a column vector X t T .

[0068] Get the transposed matrix D of D T , calculate D and D T The Euclidean distance L between them, and then calculate the reciprocal of L -1 The product of D is the first matrix D1.

[0069] Calculate D and X obs The product of , obtains the first matrix D2.

[0070] Get the transposed matrix D1 of the first matrix D1 T , calculate D1 T The product of D2 gives the parameter estimation matrix X est .

[0071] The parameter estimation matrix Xest The transposed matrix X est T , as the parameter estimation vector X at time t t ′ .

[0072] Among them, X t ′ =[x i ′ ,1 ,x i ′ ,2 ,x i ′ ,3 ,x i ′ ,4 ,x i ′ ,5 ,x i ′ ,6 ,x i ′ ,7 ], where x i ′ ,1 、x i ′ ,2 、x i ′ ,3 、x i ′ ,4 、x i ′ ,5 、x i ′ ,6 and x i ′ ,7 They are respectively the estimated value of active power, the estimated value of line voltage, the estimated value of phase current, the estimated value of excitation current, the estimated value of reactive power, the estimated value of shaft vibration and the estimated value of bearing vibration at the i-th moment.

[0073] It should be noted that the operation of obtaining the Euclidean distance between two matrices is well known. First, the matrix is ​​flattened into a one-dimensional vector. Since D and D T The number of elements in is the same, so we can get D and D T Flattened into the Euclidean distance between one-dimensional vectors. Since D is a matrix with 7 columns and 51 rows, L -1 is a data value, namely L -1 The specific operation of multiplying L with D is: -1Multiply each element in D respectively, so D1 is also a matrix with 7 columns and 51 rows. obs is a matrix with 1 column and 7 rows. It is known that the number of columns of the first matrix in matrix multiplication must be equal to the number of rows of the second matrix. Then D2 is a matrix with 1 column and 51 rows, and D1 T is a matrix with 51 columns and 7 rows, then X est is a matrix with 1 column and 7 rows, namely X est T is a matrix X with 1 column and 7 rows est Transpose to a matrix with 7 columns and 1 row.

[0074] It should be further explained that: Based on the multivariate state assessment modeling method, the parameter estimation matrix X is obtained by comprehensive analysis of the parameter vector at the current time t (current observation data) and the dynamic process matrix D of the normal operation of the thermal power generating unit (dynamic information of historical normal operation). est , which serves as the best estimate of the current operating status of the thermal power generating unit.

[0075] According to the above method, the parameter estimation vector at each moment is obtained, thereby obtaining the estimated value of each parameter at each moment.

[0076] Step S003: Determine the test statistic of the partial correlation analysis of any two parameters at each moment based on the correlation between all any two parameters in continuous moments; determine the possibility of suspected faults at each moment based on the test statistic of the partial correlation analysis of any two parameters at each moment and the parameter estimation vector at each moment.

[0077] The above obtains the estimated values ​​of various parameters at each moment during the operation of the thermal power generator set. Furthermore, since various parameters affect each other during the operation of the thermal power generator set, it is necessary to analyze the correlation between different parameters during the operation of the thermal power generator set, and according to the correlation between different parameters, combine the parameter vector at each moment with the parameter estimation vector to obtain the possibility of suspected faults at each moment.

[0078] Preferably, in one embodiment of the present invention, the method for obtaining the possibility of a suspected fault at each moment includes:

[0079] The third constant z is preset to be 30, and this is taken as an example for description.

[0080] Still taking the tth moment as an example, the period from the tzth moment to the tth moment is recorded as the reference period of the tth moment.

[0081] In the reference period at the tth moment, the numerical values ​​of each parameter at all moments are used in chronological order to form a numerical reference sequence of each parameter, and the Pearson correlation coefficient of the numerical reference sequences of any two parameters is used as the correlation coefficient of the any two parameters at the tth moment.

[0082] Taking any two parameters A and B as an example, according to the correlation coefficients of all any two parameters at the tth time, the partial correlation coefficient between parameters A and B is obtained when the influence of all other parameters other than parameters A and B is eliminated at the tth time.

[0083] According to the partial correlation coefficient between parameters A and B at the tth time after eliminating the influence of all other parameters that are not parameters A and B, the test statistic of the partial correlation analysis of parameters A and B at the tth time is obtained.

[0084] It should be noted that the calculation method of the Pearson correlation coefficient is a well-known technique. The calculation method of the partial correlation coefficient is also well-known. It is a technique commonly used in statistics and data analysis to measure the strength of the linear relationship between two variables after controlling the influence of one or more additional variables. The calculation method of the test statistic of partial correlation analysis is also well-known. The test statistic of partial correlation analysis is usually used to determine whether the partial correlation coefficient between two variables is statistically significant. In statistics, partial correlation analysis is used to evaluate the correlation between two variables after controlling the influence of one or more other variables. In order to determine whether this correlation is statistically significant, a test statistic is usually calculated.

[0085] According to the above method, the test statistic of the partial correlation analysis of any two parameters at the tth time is obtained.

[0086] In the reference period at the tth moment, the estimated values ​​of each parameter at all moments are used in chronological order to form a reference sequence of estimated values ​​of each parameter, and the Pearson correlation coefficient of the reference sequence of estimated values ​​of any two parameters is used as the correlation coefficient of the estimated values ​​of the any two parameters at the tth moment.

[0087] Still taking any two parameters A and B as an example, calculate the correlation coefficient P of parameters A and B at time t t The correlation coefficient P between (A,B) and the estimated values ​​of parameters A and B at time t t ′ The absolute value of the difference between (A, B) and the test statistic R of the partial correlation analysis of parameters A and B at time t t The normalized value of the product of (A, B) is taken as the abnormal factor of parameters A and B at the tth time, and the average of the abnormal factors of all arbitrary two parameters at the tth time is taken as the suspected fault possibility at the tth time.

[0088] It should be noted that the normalized value of the above product is normalized to between 0 and 1 using the norm() linear normalization function in this embodiment. The larger the test statistic of the partial correlation analysis, the more significant the linear relationship between the two variables is after controlling the influence of other variables, that is, R t The larger (A, B) is, the more significant the linear relationship between parameters A and B is, indicating that parameters A and B affect each other during the operation of the thermal power generating unit, and the more important parameters A and B are, the more significant the linear relationship between parameters A and B is. t (A,B) is the weight, when P t (A,B) and P t ′ The larger the difference between (A, B), the greater the difference between the values ​​of parameters A and B at time t and the estimated values. The estimated values ​​are obtained based on the data during the normal operation of the thermal power generating unit. Therefore, the larger the abnormal factor, the more likely it is that a fault has occurred at time t.

[0089] Step S004: Determine the suspected fault moment and the normal moment according to the suspected fault possibility at each moment; determine the fault occurrence time period according to the values ​​of the parameters at the suspected fault moment and the normal moment.

[0090] Preferably, in one embodiment of the present invention, the method for obtaining the fault occurrence time period includes:

[0091] The preset judgment threshold is 0.5, and this is used as an example for description.

[0092] When the possibility of a suspected fault at the tth moment is greater than a preset judgment threshold, the tth moment is recorded as the suspected fault moment.

[0093] When the suspected fault possibility at the tth moment is less than or equal to the preset judgment threshold, the tth moment is recorded as a normal moment.

[0094] According to the above method, it is determined whether each moment is a suspected fault moment or a normal moment. In this embodiment, when each parameter does not exceed the set threshold range, the first hour after the thermal power generating set starts to operate is regarded as a normal moment.

[0095] Since the thermal power generating set usually maintains a relatively stable state in the base load power generation mode, the present embodiment uses shaft vibration (vibration in the axis direction of the thermal power generating set) as the main parameter for fault judgment.

[0096] The time period consisting of the consecutive suspected fault moments is recorded as the suspected fault time period. Thus, a plurality of suspected fault time periods are obtained.

[0097] It should be noted that: in this embodiment, the duration of the suspected fault period can be 1.

[0098] Taking the g-th suspected fault period as an example, obtain the duration L of the g-th suspected fault period g , starting from the g-th suspected fault period, traverse the time before the g-th suspected fault period one by one, and obtain the previous L g A normal moment is used as a comparison normal moment.

[0099] In chronological order, the values ​​of shaft vibration at all times within the g-th suspected fault period constitute a suspected fault shaft vibration sequence, and the values ​​of shaft vibration at all control normal times constitute a control normal shaft vibration sequence.

[0100] In the suspected fault shaft vibration sequence and the normal shaft vibration sequence, the absolute value of the difference between the shaft vibration values ​​under the same sequence value is calculated, and the sum of the absolute values ​​of the difference between the shaft vibration values ​​under all the same sequence values ​​is taken as the fault factor K of the g-th suspected fault period. g .

[0101] It should be noted that the failure of thermal power generating units is often caused by a long period of full-load operation. Therefore, in this embodiment, the normal time before the suspected fault period is used as a reference to the normal time to compare the difference in shaft vibration between the suspected fault period and the normal time. The greater the difference in shaft vibration, the more likely it is that the suspected fault period is a fault.

[0102] The second threshold is preset to 0.75, and this is taken as an example for description.

[0103] When K g When the normalized value of is greater than the preset second threshold, the g-th suspected fault period is recorded as the fault occurrence period.

[0104] It should be noted that: in this embodiment, norm(K g ) as K g Normalized value of K, where norm() is a linear normalization function. g Normalized to between 0 and 1.

[0105] According to the above method, it is determined whether each suspected fault period is a fault occurrence period.

[0106] It should be noted that: when each parameter does not exceed the set threshold range, the fault occurrence period is judged every hour after the thermal power generating unit starts running. If there is a fault occurrence period within the current hour, an alarm will be immediately issued and maintenance will be notified.

[0107] The present invention also provides a thermal power generator set fault detection system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program stored in the memory to implement the steps of the aforementioned thermal power generator set fault detection method.

[0108] So far, the present invention is completed.

[0109] In summary, in the embodiment of the present invention, the value of each parameter at each moment in the operation process of the thermal power generator set is obtained, and the parameter vector at each moment is formed with the values ​​of all parameters at each moment, so as to determine the parameter estimation vector at each moment, and the correlation between any two parameters is combined to determine the possibility of suspected faults at each moment, so as to distinguish the suspected fault moment from the normal moment, and finally determine the fault occurrence period according to the values ​​of the parameters at the suspected fault moment and the normal moment. The present invention determines the suspected fault moment by analyzing the correlation between multiple parameters during the operation of the thermal power generator set, thereby ensuring the accuracy of the fault occurrence period detection.

[0110] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for detecting a fault of a thermal power generating set, characterized in that: The method comprises the following steps: Obtain the value of each parameter at every moment during the operation of the thermal power generating unit; The parameter vector at each moment is formed by the values ​​of all parameters at each moment; the dynamic process matrix of the normal operation of the thermal power generating set is formed according to the parameter vectors at consecutive moments; the parameter estimation vector at each moment is determined according to the parameter vector at each moment and the dynamic process matrix of the normal operation of the thermal power generating set; Determine the test statistic of the partial correlation analysis of any two parameters at each moment according to the correlation between all any two parameters in the continuous moments; determine the possibility of suspected fault at each moment according to the test statistic of the partial correlation analysis of any two parameters at each moment and the parameter estimation vector at each moment; According to the possibility of the suspected fault at each moment, the suspected fault moment and the normal moment are determined; according to the values ​​of the parameters at the suspected fault moment and the normal moment, the fault occurrence time period is determined.

2. A method for detecting a fault in a thermal power generating set according to claim 1, characterized in that: The dynamic process matrix constituting the normal operation of the thermal power generating set includes the following specific steps: A first constant n is preset, a second constant m is preset, and in the period from the nth moment to the mth moment, the parameter vectors at each moment are arranged from top to bottom in chronological order to form a dynamic process matrix D for the normal operation of the thermal power generating set.

3. A method for detecting a fault in a thermal power generating set according to claim 1, characterized in that: The specific steps of determining the parameter estimation vector at each moment are as follows: The transposed vector of the parameter vector at time t is used as the state evaluation matrix X obs ; Obtain the transposed matrix D of the dynamic process matrix D of the normal operation of the thermal power generating unit T , calculate the D and the D T The Euclidean distance L between them is calculated, and then the product of the reciprocal of L and D is calculated to obtain a first matrix D1; wherein T represents a transposition operation; Calculate the D and X obs The product of , obtains the first matrix D2; Get the transposed matrix D1 of D1 T , calculate the D1 T Multiply it with D2 to get the parameter estimation matrix X est ; The X est The transposed matrix of is used as the parameter estimation vector at the tth moment.

4. A method for detecting a fault of a thermal power generating set according to claim 1, characterized in that: The specific steps of determining the test statistic of the partial correlation analysis of any two parameters at each moment are as follows: A third constant z is preset, and the period from the tzth moment to the tth moment is recorded as the reference period at the tth moment; In the reference period at the t-th moment, the numerical values ​​of each parameter at all moments are used to form a numerical reference sequence of each parameter in chronological order, and the Pearson correlation coefficient of the numerical reference sequences of any two parameters is used as the correlation coefficient of any two parameters at the t-th moment; Get any two parameters A and B, and based on the correlation coefficients of all the two parameters at the tth time, get the partial correlation coefficient between the parameters A and B at the tth time after eliminating the influence of all other parameters that are not the parameters A and B; According to the partial correlation coefficient between parameters A and B at the tth moment after eliminating the influence of all other parameters that are not parameters A and B, the test statistic of the partial correlation analysis of parameters A and B at the tth moment is obtained.

5. A method for detecting a fault of a thermal power generating set according to claim 4, characterized in that: The specific steps of determining the possibility of a suspected fault at each moment are as follows: The parameter estimation vector at each moment includes: an estimated value of each parameter at each moment; In the reference period at the t-th moment, the estimated values ​​of each parameter at all moments are used to form a reference sequence of estimated values ​​of each parameter in chronological order, and the Pearson correlation coefficient of the reference sequence of estimated values ​​of any two parameters is used as the correlation coefficient of the estimated values ​​of any two parameters at the t-th moment; Calculate the correlation coefficient P of parameters A and B at time t t The correlation coefficient P between (A,B) and the estimated values ​​of parameters A and B at time t t ′ The absolute value of the difference between (A, B), and the normalized value of the product of the absolute value of the difference and the test statistic of the partial correlation analysis of the parameters A and B at the t-th time as the abnormal factor of the parameters A and B at the t-th time; The mean of the abnormal factors of all two arbitrary parameters at the tth moment is taken as the suspected fault possibility at the tth moment.

6. A method for detecting faults in a thermal power generating set according to claim 1, characterized in that: The specific steps of determining the suspected fault time and the normal time are as follows: When the suspected fault possibility at the tth moment is greater than the preset judgment threshold, the tth moment is recorded as the suspected fault moment; When the suspected fault possibility at the tth moment is less than or equal to the preset judgment threshold, the tth moment is recorded as a normal moment.

7. A method for detecting faults in a thermal power generating set according to claim 1, characterized in that: The specific steps of determining the fault occurrence period are as follows: The time period consisting of the consecutive suspected fault moments is recorded as the suspected fault time period; Get the duration L of the gth suspected fault period g , starting from the g-th suspected fault period, traverse the time before the g-th suspected fault period one by one, and obtain the previous L g A normal moment, as a control normal moment; The parameters at least include: shaft vibration; in chronological order, the values ​​of shaft vibration at all times within the g-th suspected fault period constitute a suspected fault shaft vibration sequence, and the values ​​of shaft vibration at all normal times constitute a control normal shaft vibration sequence; Determining the fault factor of the g-th suspected fault period according to the suspected fault shaft vibration sequence and the reference normal shaft vibration sequence; According to the fault factor of the g-th suspected fault period, it is determined whether the g-th suspected fault period is a fault occurrence period.

8. A method for detecting faults in a thermal power generating set according to claim 7, characterized in that: The specific steps of determining the failure factor of the g-th suspected failure period are as follows: In the suspected fault shaft vibration sequence and the control normal shaft vibration sequence, the absolute value of the difference between the shaft vibration values ​​with the same sequence value is calculated, and the sum of the absolute value of the difference between the shaft vibration values ​​with all the same sequence values ​​is taken as the fault factor of the gth suspected fault period.

9. A method for detecting faults in a thermal power generating set according to claim 7, characterized in that: The specific steps of determining whether the g-th suspected fault period is a fault occurrence period are as follows: When the normalized value of the fault factor of the g-th suspected fault period is greater than the preset second threshold, the g-th suspected fault period is recorded as the fault occurrence period.

10. A thermal power generator fault detection system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is executed by a processor, the steps of a method for detecting a fault of a thermal power generator set as described in any one of claims 1 to 9 are implemented.

Images