Power change-based power generation equipment operation parameter anomaly detection method and device

Abstract

The invention provides a power change-based power generation equipment operation parameter anomaly detection method and device, and the method comprises the steps: obtaining the measurement values of the operation parameters of power generation equipment at different moments, and analyzing the expectation of the difference value of the operation parameters at two adjacent moments in different states according to the measurement values; on the basis of expected analysis results in different states, on-line steady state judgment is carried out on the power generation equipment through the power difference value; screening out steady-state parameters from the measured values according to an online steady-state judgment result, and constructing feature test statistics according to the expectation of a difference value of operating parameters and the expectation of a power difference value in the steady-state parameters; and judging abnormal parameters in the steady state parameters according to the characteristic test statistics. According to the method, the influence of power change on power generation equipment anomaly detection is considered, and the anomaly detection accuracy and timeliness are improved by establishing the anomaly detection model.

Description

Technical Field

[0001] The present application relates to the technical field of operation and maintenance of power generation equipment, and particularly relates to a method and device for detecting abnormal operation parameters of power generation equipment based on power change. Background Art

[0002] Currently, anomaly detection technology has been widely applied in various fields such as intrusion detection in network security, fault diagnosis of important equipment and systems, and personnel activity monitoring. The main purpose of anomaly detection is to discover abnormal behavior patterns hidden in data that are inconsistent with expected normal behaviors. Among them, for industrial equipment such as power generation equipment, faults caused by component wear and improper operation may lead to a decline in the reliability and economy of the equipment. Therefore, it is necessary to detect equipment status anomalies early and take maintenance measures to prevent further deterioration of the equipment and more serious losses.

[0003] In related technologies, when detecting anomalies in power generation equipment, a normal mode area is generally directly established, and any point falling outside this normal area is regarded as an anomaly point. However, in practical applications, the normal mode evolves with the development of the system and is not static. Taking a generator set as an example, the state of the system changes before and after each overhaul and component replacement of the unit, and with the equipment aging caused by the normal operation of the unit, the performance level of the system also continuously declines. In addition, it is very difficult to accurately define the normal mode area.

[0004] Therefore, the detection results obtained by the above anomaly detection methods in related technologies may have relatively large errors. Summary of the Invention

[0005] The present application aims to at least solve one of the technical problems in related technologies to some extent.

[0006] To this end, the first object of the present application is to propose a method for detecting abnormal operation parameters of power generation equipment based on power change. This method takes into account the impact of power change on the anomaly detection of power generation equipment, and improves the accuracy and timeliness of anomaly detection by establishing an anomaly detection model.

[0007] The second object of the present application is to propose a device for detecting abnormal operation parameters of power generation equipment based on power change;

[0008] The third object of the present application is to propose an electronic device;

[0009] The fourth object of the present application is to propose a non-transitory computer-readable storage medium.

[0010] To achieve the above object, the first aspect of the present application is to propose a method for detecting abnormal operation parameters of power generation equipment based on power change, and this method includes the following steps:

[0011] Obtain the measured values of the operating parameters of the power generation equipment at different times. According to the measured values, analyze the expectations of the differences in the operating parameters at two adjacent times in different states, wherein the expectations in different states are analyzed by using the power difference between two adjacent times.

[0012] Based on the analysis results of the expectations in different states, perform online steady-state discrimination on the power generation equipment through the power difference.

[0013] Screen out the steady-state parameters from the measured values according to the online steady-state discrimination results, and construct a characteristic test statistic according to the expectations of the differences in the operating parameters and the expectation of the power difference in the steady-state parameters.

[0014] Determine the abnormal parameters in the steady-state parameters according to the characteristic test statistic.

[0015] Optionally, in an embodiment of the present application, the analyzing the expectations of the differences in the operating parameters at two adjacent times in different states according to the measured values includes: constructing a first expression of the measured value composed of a true value and a random error, wherein the true value includes a normal value and an abnormal value, and the normal value is related to the power of the power generation equipment; combining the first expression, the function of the power difference, and the difference of the abnormal value to determine a second expression of the difference in the operating parameters, and determining a third expression of the power difference according to the steady-state process and the non-steady-state process of the power generation equipment; determining the expectation formula of the power difference according to the third expression, and determining the expectations of the differences in the operating parameters in different states according to the second expression and the expectation formula of the power difference.

[0016] Optionally, in an embodiment of the present application, the performing online steady-state discrimination on the power generation equipment through the power difference includes: determining the probability distribution of the power difference according to the third expression, and using the power difference as a characteristic index for steady-state and non-steady-state discrimination, and respectively establishing probability models of the power generation equipment in the steady state and the non-steady state by using the probability distribution; regressing the non-steady-state process through a Gaussian mixture model, and setting a steady-state discrimination result that follows a Bernoulli distribution based on the obtained non-steady-state Gaussian mixture model; calculating the Bernoulli distribution parameter by counting the proportion of the number of steady-state samples in the training samples of the statistical model, and determining the probability distribution of the steady-state discrimination result according to the Bernoulli distribution parameter; constructing a non-steady-state discrimination formula by combining the Bayesian formula and the probability distribution of the steady-state discrimination result.

[0017] Optionally, in an embodiment of the present application, constructing a characteristic test statistic according to the expectation of the difference between the operating parameters and the expectation of the power difference in the steady-state parameters includes: respectively calculating the T statistic of the difference between the operating parameters and the power difference, and respectively calculating the test P value corresponding to each T statistic; according to the influence of the expected change of the difference between the operating parameters on the expectation of the power difference, constructing the characteristic test statistic by using the test P values corresponding to each T statistic.

[0018] Optionally, in an embodiment of the present application, determining an abnormal parameter in the steady-state parameters according to the characteristic test statistic includes: calculating the characteristic test statistic of a preset normal value sample, performing regression on the characteristic test statistic of the normal value sample to obtain a probability density distribution model; determining an abnormal detection threshold according to the probability density distribution model; comparing the characteristic test statistic of the current operating parameter with the abnormal detection threshold, and judging whether the current operating parameter is an abnormal parameter according to the comparison result.

[0019] Optionally, in an embodiment of the present application, after determining the abnormal parameter in the steady-state parameters according to the characteristic test statistic, it further includes: obtaining the parameter value of the abnormal parameter at a subsequent moment, and calculating the subsequent abnormal deviation degree of the abnormal parameter according to the parameter value at the subsequent moment; verifying the abnormal parameter determination result according to the subsequent abnormal deviation degree and the current abnormal deviation degree of the abnormal parameter.

[0020] Optionally, in an embodiment of the present application, after determining the abnormal parameter in the steady-state parameters according to the characteristic test statistic, it further includes: calculating the change trend of the abnormal parameter according to the parameter value of the abnormal parameter at a subsequent moment; performing fault diagnosis on the power generation equipment by using the change trend of the abnormal parameter.

[0021] To achieve the above object, a second aspect of the present application further provides an abnormal detection device for operating parameters of a power generation equipment based on power change, including the following modules:

[0022] An analysis module, configured to obtain the measured values of the operating parameters of the power generation equipment at different moments, and analyze the expectations of the differences between the operating parameters at adjacent two moments in different states according to the measured values, wherein the expectations in different states are analyzed by using the power difference between adjacent two moments;

[0023] A discrimination module, configured to perform an online steady-state discrimination on the power generation equipment through the power difference based on the analysis result of the expectations in different states;

[0024] A construction module for screening out steady-state parameters from the measured values according to the online steady-state discrimination result, and constructing a characteristic test statistic according to the expectation of the difference between the operating parameters in the steady-state parameters and the expectation of the power difference;

[0025] A determination module for determining abnormal parameters in the steady-state parameters according to the characteristic test statistic.

[0026] To achieve the above object, a third aspect of the present application further proposes an electronic device, including:

[0027] A processor;

[0028] A memory for storing executable instructions of the processor;

[0029] Wherein, the processor is configured to execute the instructions to implement the abnormal detection method for the operating parameters of the power generation device based on power change as described in any one of the above first aspects.

[0030] To achieve the above object, an embodiment of a fourth aspect of the present application further proposes a non-transitory computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the abnormal detection method for the operating parameters of the power generation device based on power change as described in any one of the above first aspects.

[0031] The technical solutions provided by the embodiments of the present application at least bring the following beneficial effects: The present application first analyzes the expectation of the difference between operating parameters in different states by using the power difference according to the measured values of the operating parameters of the power generation device at different times; then performs online steady-state discrimination on the power generation device by using the power difference; and then constructs a characteristic test statistic according to the expectation of the difference between the operating parameters in the steady-state parameters and the expectation of the power difference, and determines the abnormal parameters in the steady-state parameters according to the characteristic test statistic. Thus, the present application compares the current level of the operating parameters with the recent reference level, which can more reasonably reflect the health status of the power generation device, combines the influence of power change to detect abnormal parameters of the power generation device, and avoids misjudgment of abnormal detection caused by power change. An abnormal detection model is established in a data-driven manner, and a new characteristic test statistic is constructed to reduce the influence of power expectation change. After substituting the real-time parameters into the abnormal detection model for calculation, the correct detection result can be output. Therefore, the present application improves the accuracy, timeliness and reliability of abnormal detection, can timely discover abnormal data of the power generation device, and is beneficial to ensuring the normal operation of the power generation device.

[0032] Additional aspects and advantages of the present invention will be given in part in the following description, become apparent in part from the following description, or be learned through the practice of the present invention. Description of the Drawings

[0033] The above and / or additional aspects and advantages of the present application will become apparent and be readily understood from the following description of embodiments in conjunction with the accompanying drawings, where:

[0034] Figure 1 is a flowchart of a method for detecting abnormal operation parameters of a power generation device based on power change proposed in an embodiment of the present application;

[0035] Figure 2 is a flowchart of a method for data analysis of operation parameters of a power generation device proposed in an embodiment of the present application;

[0036] Figure 3 is a comparison schematic diagram of the influence of primary frequency modulation and secondary frequency modulation on power proposed in an embodiment of the present application;

[0037] Figure 4 is a schematic diagram of the discrimination process of an abnormal detection model proposed in an embodiment of the present application;

[0038] Figure 5 is a schematic diagram of a detection process based on a sliding window proposed in an embodiment of the present application;

[0039] Figure 6 is a structural schematic diagram of a device for detecting abnormal operation parameters of a power generation device based on power change proposed in an embodiment of the present application. Detailed Embodiments

[0040] The embodiments of the present invention will be described in detail below. The examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to explain the present invention and should not be construed as limiting the present invention.

[0041] It should be noted that the output power of power generation devices such as generator sets is not a fixed value and will be adjusted in real time according to wind resource changes and grid load requirements. When the output power of a generator set changes, other parameters will also change accordingly. Therefore, for power generation devices, a deviation in parameters cannot be simply determined as an abnormality in the device itself. It may be caused by power changes. Different from abnormal detection problems in other fields, abnormal detection of generator set parameters needs to consider the influence of power changes. Based on this, the present application proposes a method for detecting abnormal operation parameters of a power generation device, which combines power changes for abnormal detection to improve the accuracy of abnormal detection of power generation devices.

[0042] The following will describe in detail a method and device for detecting abnormal operation parameters of a power generation device based on power change proposed in an embodiment of the present application with reference to the accompanying drawings.

[0043] Figure 1The flowchart of a method for detecting abnormal operation parameters of a power generation device based on power change proposed in an embodiment of the present application is as follows Figure 1 As shown, the method includes the following steps:

[0044] Step S101: Obtain the measured values of the operation parameters of the power generation device at different times. According to the measured values, analyze the expectations of the differences in the operation parameters between two adjacent times in different states. Among them, use the power difference between two adjacent times to analyze the expectations in different states.

[0045] Specifically, first construct a mathematical model of the operation parameters of the power generation device. Based on the relationships and change principles between different parameters reflected by the mathematical model, analyze the expectations of the differences in the operation parameters in different states. Among them, various detection devices can be used to continuously monitor the power generation device to obtain the measured values of various operation parameters at different times. The quantity and type of the obtained operation parameters are determined according to the actual application situation of the power generation device. As an example, this application can use a wind turbine generator as the power generation device to be detected for an exemplary description of abnormal detection.

[0046] To more clearly illustrate the specific implementation process of analyzing the expectations of the differences in the operation parameters between two adjacent times in this application, the following uses an analysis method proposed in an embodiment of this application for exemplary illustration. Figure 2 The flowchart of a method for data analysis of the operation parameters of a power generation device proposed in an embodiment of the present application is as follows Figure 2 As shown, the method includes the following steps:

[0047] Step S201: Construct a first expression of the measured value composed of a true value and a random error. Among them, the true value includes a normal value and an abnormal value, and the normal value is related to the power of the power generation device.

[0048] Specifically, consider the measured value of the generator set parameter as the superposition of the true value and the random error. Then, the expression (i.e., the first expression) of the measured value of the operation parameter at time t is as shown in the following formula (1):

[0049] x(t) = μ x (t) + ε x = c(p(t)) + f(t) + ε x

[0050] Among them, x(t) represents the measured value of the parameter; μ x(t) represents the true value of the parameter, which consists of two parts: the normal value and the abnormal value; c(p(t)) represents the normal value, which is related to the boundary conditions of the unit, and the power p(t) has the greatest influence; f(t) represents the abnormal value of the parameter, and its expression is shown in the following formula (2):

[0051]

[0052] Among them, t0 represents the moment when the parameter starts to show abnormalities; k(t) represents the change rate of the parameter abnormality.

[0053] ε x represents the random error. In this embodiment, the random error can be considered as Gaussian white noise, and the expression of the random error is shown in the following formula (3):

[0054]

[0055] It should be noted that in the anomaly detection of industrial equipment, the input data usually refers to sensor data. In the anomaly detection of generator sets, the operating state of the unit can be indirectly understood through the sensors arranged on site. The measured values of the sensors are generally mixed with a large amount of data noise, and there are significant differences between data anomaly points and noise points. Noise refers to the random error in the measured data, which generally follows a normal distribution, and the resulting fluctuations within a certain range of the data are normal phenomena. Noise brings interference to the in-depth excavation of data information. Therefore, in this embodiment of the present application, data denoising technology is used to preprocess the noise points before data analysis.

[0056] Step S202: Combine the first expression, the function of the power difference, and the difference of the abnormal value to determine the second expression of the difference of the operating parameter, and determine the third expression of the power difference according to the steady-state process and the non-steady-state process of the power generation equipment.

[0057] Specifically, the difference between the normal values of the operating parameters at two adjacent moments is a function of the power difference, which can be expressed by the following formula (4):

[0058] c(p(t)) - c(p(t - 1)) = g(Δp).

[0059] The difference between the abnormal values of the operating parameters at two adjacent moments is approximately equal to the abnormal change rate of the parameter at time t, which can be expressed by the following formula (5):

[0060]

[0061] Therefore, by combining the above formulas (1), (4), and (5), the expression (i.e., the second expression) of the difference between the measured values of the operating parameters at two adjacent moments can be obtained as shown in the following formula (6):

[0062] Δx = x(t) - x(t - 1) ≈ g(Δp) + k(t) + ε Δx

[0063] Wherein, Δp represents the difference in power between two adjacent moments, and its expression (i.e., the third expression) is shown in formula (7):

[0064] Δp = m1(Δf) + m2(Δl) + ε Δp

[0065] Wherein, m1(Δf) represents the influence of the change in grid frequency on power during the primary frequency regulation of the unit, and Δf represents the change in frequency; m2(Δl) represents the influence of the grid load dispatch instruction on power during the secondary frequency regulation of the unit, and Δl represents the change in the grid load instruction; ε Δp represents the random error of Δp.

[0066] It should be noted that during the secondary frequency regulation of the generator set, the change range of the output power is very large and the response is rapid. The model relationship between the input and output parameters of the generator set system cannot maintain strong consistency, and at this time, the unit is in an unsteady state. Compared with the secondary frequency regulation, the power change range in the primary frequency regulation process is smaller, and the speed regulation rate is generally about 5%, and it can be considered that the unit is in a quasi-steady state process. For example, the comparison of the influence of the primary frequency regulation and the secondary frequency regulation on the power of the unit is as Figure 3 shown. Among them, Figure 3 part (a) represents the influence of the primary frequency regulation on the power, and part (b) represents the influence of the secondary frequency regulation on the power, Figure 3 the horizontal axis in is time, and the vertical axis is power. The areas in the circles are respectively enlarged and shown as part (a) and part (b) to more clearly show the data differences.

[0067] Step S203, determine the expected formula of the power difference according to the third expression, and determine the expectations of the differences in the operating parameters in different states according to the second expression and the expected formula of the power difference.

[0068] Specifically, it can be analyzed from the above formulas (1) to (7) that the probability distribution of the difference Δx between two adjacent moment parameters follows a normal distribution and can be expressed by the following formula (8):

[0069]

[0070] Among them, according to the above third expression, the expected formula of the power difference Δp between two adjacent moments can be determined as shown in the following formula (9):

[0071] E(Δp) = m1(Δf) + m2(Δl)

[0072] Among them, in the steady-state process, the grid load scheduling instruction Δl = 0, m2(Δl) = 0, and when the parameters are normal, the change rate k(t) = 0.

[0073] Therefore, by combining the second expression and the expected formula of the power difference, the expectations of the differences Δx of the operating parameters at two adjacent moments in different states can be obtained as shown in the following formula (10):

[0074]

[0075] Wherein, steady represents the steady state, unsteady represents the non-steady state, normal represents the normal value, and abnormal represents the abnormal value.

[0076] Therefore, this embodiment analyzes the expectations of the differences of the operating parameters at two adjacent moments in different states, and the above analysis process can be used as the basis for subsequent anomaly detection.

[0077] Step S102, based on the analysis results of the expectations in different states, perform online steady-state discrimination on the power generation equipment through the power difference.

[0078] Specifically, use the expectation analysis process in the previous step to perform online steady-state discrimination on the power generation equipment. As can be seen from the above formula (10), in the two cases where the generator set is in the non-steady state and the parameters are abnormal due to equipment failure, the expectations of the operating parameter differences will change compared with the steady-state normal situation. Therefore, in order to accurately detect parameter anomalies, it is necessary to perform steady-state discrimination first.

[0079] In an embodiment of the present application, performing online steady-state discrimination on the power generation equipment through the power difference includes the following steps:

[0080] The first step is to determine the probability distribution of the power difference according to the third expression, and use the power difference as a characteristic index for steady-state and non-steady-state discrimination, and establish probability models of the power generation equipment in the steady state and the non-steady state respectively using the probability distribution.

[0081] Specifically, the probability distribution of the difference Δp between the powers at two adjacent moments is different in the steady-state and non-steady-state processes. According to the above third expression, the probability distribution of the power difference can be determined as shown in the following formula (11):

[0082]

[0083] Therefore, the difference Δp of the output power can be selected as the characteristic index for discriminating the steady state and non-steady state of the unit, and probability models in the steady state and the non-steady state are established respectively using different probability distributions.

[0084] In the second step, the non-steady state process is regressed by a Gaussian mixture model. Based on the obtained non-steady state Gaussian mixture model, a steady state discrimination result obeying the Bernoulli distribution is set.

[0085] Specifically, for the non-steady state process, the load scheduling instruction is not unique, which in turn leads to the fact that the change direction and change rate of power are not fixed values. Therefore, in this embodiment, a Gaussian mixture model is selected to regress the non-steady state process, and the Gaussian mixture model of the non-steady state process obtained is shown in the following formulas (12) and (13):

[0086]

[0087] where θ represents the expectation of Δp, and φ represents the parameter of the Bernoulli distribution.

[0088] Based on the principle shown in the above Gaussian mixture model, it can be set that y represents the steady state discrimination result, and this result obeys the Bernoulli distribution. Among them, y = 0 represents the steady state process, and y = 1 represents the non-steady state process.

[0089] In the third step, by statistically counting the proportion of the number of steady state samples in the training samples of the model, the Bernoulli distribution parameter is calculated, and the probability distribution of the steady state discrimination result is determined according to the Bernoulli distribution parameter.

[0090] Specifically, by statistically counting the ratio of the number of steady state samples to the total number of samples in the training samples of the model, the parameter φ of the Bernoulli distribution is calculated, and then the probability distribution of the steady state discrimination result is determined according to the Bernoulli distribution parameter. This operation can be represented by the following formula (14):

[0091] P(y) = B(φ)

[0092] In the fourth step, combining the Bayesian formula and the probability distribution of the steady state discrimination result, a non-steady state discrimination formula is constructed.

[0093] Specifically, since in the steady state discrimination problem, it is very difficult to directly establish the p(y|Δp) model, so in this embodiment, p(Δp|y) and p(y) can be modeled separately. Then the Bayesian formula shown in the following formula (15) is introduced:

[0094]

[0095] Furthermore, combining formula (14) and formula (15), the steady state discrimination problem is transformed into the form shown in formula (16):

[0096]

[0097] where this formula is the non-steady state discrimination formula.

[0098] Thus, after obtaining the real-time operation parameters of the power generation equipment, the present application can substitute the power difference Δp therein into the above formula (16) for calculation. If the output result is 1, it indicates that the parameter is an unsteady-state parameter.

[0099] Step S103: Screen out the steady-state parameters from the measured values according to the online steady-state discrimination result, and construct a characteristic test statistic based on the expectations of the differences of the operation parameters and the power difference in the steady-state parameters.

[0100] Specifically, according to the online steady-state discrimination operation in the previous step, non-steady-state parameters are excluded from the measured values of the operation parameters obtained in step S101, and the operation parameters in the steady-state process are screened out. Furthermore, the characteristic test statistic is calculated based on the steady-state parameters, so as to subsequently use the characteristic test statistic to judge the abnormal points in the remaining steady-state parameters.

[0101] It should be noted that the expectations of the differences of the steady-state parameters are different in the normal and abnormal modes. The expectation of Δx in the normal state is not a fixed value and is related to the expectation of Δp. If the expectation of Δp changes due to primary frequency modulation within a certain period of time, the expectation of Δx will also change accordingly. Therefore, the change of the expectation of Δx and the change of the expectation of Δp should be comprehensively considered for anomaly detection.

[0102] In an embodiment of the present application, constructing a characteristic test statistic based on the expectations of the differences of the operation parameters and the power difference in the steady-state parameters includes: calculating the T statistics of the differences of the operation parameters and the power difference respectively, and calculating the test P values corresponding to each T statistic respectively; constructing a characteristic test statistic by using the test P values corresponding to each T statistic according to the influence of the change of the expectation of the difference of the operation parameters on the expectation of the power difference.

[0103] Specifically, since the distribution functions of Δx and Δp are different, the change of their expectations cannot be directly compared. Therefore, in the embodiment of the present application, the p value of the T statistic is selected to standardize and characterize the change of the expectation.

[0104] As a possible implementation manner, the calculation formula of the T statistic of Δp is shown in the following formula (17):

[0105]

[0106] Wherein, represents the mean value of the Δp sample, which is an unbiased estimate of the expectation; represents the number of samples; S represents the standard deviation of the Δp sample; θ represents the expectation of Δp; t n-1The t value representing the degrees of freedom of the T distribution is n - 1. It can be understood that when the generator set does not participate in primary frequency modulation, m1(Δf) = 0, and the expectation of Δp is also equal to 0. The reference of the T statistic is to measure the expected change of Δp caused by frequency modulation. Therefore, in this embodiment, θ = 0.

[0107] Similarly, the T statistic of Δx can be calculated by referring to the above method.

[0108] Furthermore, in order to measure the deviation degree of the T statistic, the test P value of the T statistic is also calculated in this embodiment. Among them, the test P value corresponding to Δx is shown in the following formula (18):

[0109] p x = P(|T| > |t x |||H0)

[0110] The test P value corresponding to Δp is shown in the following formula (19):

[0111] p power = P(|T| > |t p ||H0)

[0112] Among them, H0 represents the null hypothesis. The test P value of the T statistic is used to indicate how much probability reaches the current deviation degree and even greater.

[0113] Among them, in the calculation process of the P value, the null hypothesis and the alternative hypothesis are shown in formula (20):

[0114]

[0115] t x The calculation formula is shown in the following formula (21):

[0116]

[0117] t power The calculation formula is shown in the following formula (22):

[0118]

[0119] Furthermore, according to the influence of the expected change of Δp on the expectation of Δx, a new characteristic test statistic is introduced, and the characteristic test statistic shown in the following formula (23) is constructed by using the test P values corresponding to the above two T statistics:

[0120] P-ratio = ln((p x + 1) / (p power + 1)).

[0121] It can be understood that because p x , ppower ∈(0, 1), so ln((p x + 1) / (p power + 1)) ∈ (-0.7, 0.7). In particular, when p x = p power ), P_ratio = 0. When p x << p power ), it indicates that the deviation of the expected value of Δx is much greater than the change in the expected value of Δp, and the parameter x is likely to be abnormal; when p x > p power or p x ≈ p power ), it indicates that the deviation of the expected value of Δx may be caused by the change in the expected value of Δp, and there is no sufficient reason to prove that the parameter is abnormal.

[0122] Step S104, determine the abnormal parameters in the steady-state parameters according to the characteristic test statistic.

[0123] Specifically, use the calculation formula of the characteristic test statistic obtained in the previous step to calculate the values of the characteristic test statistics corresponding to the remaining steady-state parameters in the measured values of the collected operating parameters, and then determine whether the remaining steady-state parameters are abnormal parameters.

[0124] In an embodiment of the present application, determining the abnormal parameters in the steady-state parameters according to the characteristic test statistic includes: calculating the characteristic test statistic of the preset normal value sample, performing regression on the characteristic test statistic of the normal value sample to obtain a probability density distribution model; determining an abnormal detection threshold according to the probability density distribution model; comparing the characteristic test statistic of the current operating parameter with the abnormal detection threshold, and judging whether the current operating parameter is an abnormal parameter according to the comparison result.

[0125] Specifically, in this embodiment, the P-ratio values of the respective parameters in the previously obtained normal value sample set can be calculated through the above formula (23). Then, perform regression on the probability density distribution model of P-ratio. In this embodiment, the method of kernel density estimation can be selected for regression, and the calculation formula for regression is as shown in the following formula (24):

[0126]

[0127] where n represents the sample size, K() represents the kernel smoothing function, and h represents the bandwidth value.

[0128] Furthermore, on the basis of obtaining the probability density distribution, determine the abnormal detection threshold, which can be represented by the following formula (25):

[0129] T = P(P-ratio < 0.01)

[0130] Among them, P() represents the probability distribution model of the P-ratio.

[0131] Therefore, after obtaining the anomaly detection threshold, the numerical values of the characteristic test statistics corresponding to each remaining steady-state parameter in the current measurement values obtained through the above steps are compared with the anomaly detection threshold, and each steady-state parameter is judged to be a normal value or an abnormal value according to the comparison result. Among them, when the numerical value of the characteristic test statistic of a certain parameter collected in real time is greater than the anomaly detection threshold, it is determined that the parameter is an abnormal parameter, otherwise it is a normal parameter.

[0132] Therefore, through the above steps, an anomaly detection model as shown in Figure 4 can be constructed. As shown in Figure 4 , for the data obtained from a Supervisory Information System / database, the operation is carried out according to the Figure 4 process, and the discrimination result of the abnormal parameter can be obtained.

[0133] Furthermore, in the actual application of this anomaly detection model, since the calculation of the statistic requires sample points within a period of time, this embodiment can also adopt a sliding window method for model detection. As shown in Figure 5 . The length of the sliding window is denoted as L, and the number of sample points within the window is L + 1. The statistic calculated from the sample points within the sliding window is used as the eigenvalue of the newly entered sample point in the sliding window, as shown by the blue points in Figure 5 . If the statistic of the sample points within the sliding window is determined to be abnormal, the newly entered sample point in the sliding window can be considered abnormal.

[0134] Based on the above embodiments, in an embodiment of the present application, after determining the abnormal parameter in the steady-state parameter according to the characteristic test statistic, it further includes: obtaining the parameter value of the abnormal parameter at a subsequent moment, and calculating the subsequent abnormal deviation degree of the abnormal parameter according to the parameter value at the subsequent moment; verifying the determination result of the abnormal parameter according to the subsequent abnormal deviation degree and the current abnormal deviation degree of the abnormal parameter.

[0135] Specifically, since the data of the operation parameters of the generator set has both numerical attributes and time attributes, this embodiment applies the time series analysis method to anomaly detection. The state and parameter values of the unit at the subsequent moment are always related to those at the previous moment. When the unit has a fault at the current moment, the abnormal deviation degree of the parameters at the subsequent moment is greater than or equal to the abnormal deviation degree at the current moment. After the unit has not been adjusted and repaired specifically, the degree of the fault will only deepen over time and will not weaken. Therefore, this embodiment uses this negative information to assist in anomaly decision-making to reduce the false alarm rate of anomaly detection.

[0136] During specific implementation, when outputting the abnormal detection result, continuously collect relevant operating parameters, and determine the abnormal deviation degree of the preliminarily determined abnormal parameter at the current moment through the P-value calculation method in the above embodiments, and determine the abnormal deviation degree of the abnormal parameter at subsequent moments according to the parameter values collected at subsequent moments. If the subsequent abnormal deviation degree of the abnormal parameter is greater than or equal to the current abnormal deviation degree, the determination result of the abnormal parameter can be verified as correct. If the subsequent abnormal deviation degree of the abnormal parameter is less than the current abnormal deviation degree, it can be returned for re-abnormal detection.

[0137] Based on the above embodiments, in an embodiment of the present application, after determining the abnormal parameter in the steady-state parameter according to the characteristic test statistic, it further includes: calculating the change trend of the abnormal parameter according to the parameter value of the abnormal parameter at subsequent moments; using the change trend of the abnormal parameter to perform fault diagnosis on the power generation equipment.

[0138] Specifically, after a fault occurs in the generator set equipment, some parameters show slow-varying changes, some parameters show sudden changes; some parameters show an upward trend change, some parameters show a downward trend change; the fluctuation amplitude of some parameters increases, etc. The change trends of each parameter are different, and will not be elaborated one by one in this embodiment. After detecting that a parameter is abnormal in this application, in order to accurately identify and locate the specific abnormal cause and then take relevant treatment measures, the characteristic information of the parameter change is also used for fault diagnosis of the equipment. Therefore, the abnormal detection of the generator set in this embodiment not only needs to determine whether the parameter is abnormal, but also needs to further distinguish the characteristics and types of the abnormality on this basis. For example, compare the parameter value at the current moment with the parameters at subsequent moments to determine which of the above change trends the operating parameter belongs to. Then input the determined change trend into the preset generator set fault diagnosis knowledge base to determine the fault corresponding to the change trend.

[0139] In summary, for the method for detecting abnormal operation parameters of a power generation device based on power change according to the embodiments of the present application, first, according to the measured values of the operation parameters of the power generation device at different times, the expectations of the differences in operation parameters in different states are analyzed using the power differences; then, the online steady state discrimination of the power generation device is performed using the power differences; then, according to the expectations of the differences in operation parameters and the power differences in the steady state parameters, a characteristic test statistic is constructed, and the abnormal parameters in the steady state parameters are determined according to the characteristic test statistic. Thus, this method compares the current level of the operation parameters with the recent reference level, can more reasonably reflect the health status of the power generation device, combines the influence of power change to detect abnormal parameters of the power generation device, and avoids misjudgment of abnormal detection caused by power change. An abnormal detection model is established in a data-driven manner, and a new characteristic test statistic is constructed to reduce the influence of power expectation change. After substituting the real-time parameters into the abnormal detection model for calculation, the correct detection result can be output. Therefore, this method improves the accuracy, timeliness and reliability of abnormal detection, can timely detect abnormal data of the power generation device, and is beneficial to ensuring the normal operation of the power generation device.

[0140] To implement the above embodiments, the present application also proposes a device for detecting abnormal operation parameters of a power generation device based on power change. Figure 6 As shown in the structural schematic diagram of a device for detecting abnormal operation parameters of a power generation device based on power change proposed by an embodiment of the present application, Figure 6 as shown, the device includes an analysis module 100, a discrimination module 200, a construction module 300 and a determination module 400.

[0141] Among them, the analysis module 100 is configured to obtain the measured values of the operation parameters of the power generation device at different times, and according to the measured values, analyze the expectations of the differences in the operation parameters at two adjacent times in different states, wherein the expectations in different states are analyzed using the power differences between two adjacent times.

[0142] The discrimination module 200 is configured to perform online steady state discrimination on the power generation device through the power differences based on the analysis results of the expectations in different states.

[0143] The construction module 300 is configured to screen out the steady state parameters from the measured values according to the online steady state discrimination result, and construct a characteristic test statistic according to the expectations of the differences in the operation parameters and the power differences in the steady state parameters.

[0144] The determination module 400 is configured to determine the abnormal parameters in the steady state parameters according to the characteristic test statistic.

[0145] It should be noted that the foregoing explanation of the embodiments of the method for detecting abnormal operation parameters of a power generation device based on power change also applies to the device of this embodiment. The principles for each module to implement its functions are the same and will not be elaborated here.

[0146] In summary, the device for detecting abnormal operation parameters of a power generation device based on power change according to the embodiments of the present application compares the current level of the operation parameters with the recent reference level, which can more reasonably reflect the health status of the power generation device. By combining the influence of power change for abnormal detection of the power generation device parameters, it avoids misjudgment of abnormal detection caused by power change. An abnormal detection model is established in a data-driven manner, and a new feature test statistic is constructed to reduce the influence of power expected change. After substituting the real-time parameters into the abnormal detection model for calculation, the correct detection result can be output. Thus, the device improves the accuracy, timeliness, and reliability of abnormal detection, can timely discover abnormal data of the power generation device, and is beneficial to ensuring the normal operation of the power generation device.

[0147] To implement the above embodiments, the present application also proposes an electronic device, which includes: a processor; a memory for storing processor-executable instructions; wherein, the processor is configured to execute the instructions to implement the method for detecting abnormal operation parameters of a power generation device based on power change as described in any one of the embodiments of the first aspect above.

[0148] To implement the above embodiments, the present application also proposes a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements the method for detecting abnormal operation parameters of a power generation device based on power change as described in any one of the embodiments of the first aspect above.

[0149] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. In addition, without conflict, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.

[0150] In addition, the terms "first" and "second" are for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, features defined with "first" and "second" may explicitly or implicitly include at least one such feature. In the description of this application, "a plurality of" means at least two, such as two, three, etc., unless otherwise specifically defined.

[0151] Any process or method description represented in a flowchart or described otherwise herein may be understood to represent a module, segment, or portion of code including one or more executable instructions for implementing a customized logical function or process. The scope of the preferred embodiments of this application includes additional implementations, where functions may be executed in a substantially simultaneous manner or in a reverse order according to the functions involved, rather than in the order shown or discussed, which should be understood by those skilled in the technical field to which the embodiments of this application pertain.

[0152] The logic and / or steps represented in a flowchart or described otherwise herein, for example, may be considered as a sequenced list of executable instructions for implementing a logical function and may be specifically implemented in any computer-readable medium for use by or in connection with an instruction execution system, apparatus, or device, such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device. For the purposes of this specification, a "computer-readable medium" may be any device that can contain, store, communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of the computer-readable medium include the following: an electrical connection portion with one or more wirings (electronic device), a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable medium may even be paper or other suitable medium on which the program can be printed, as the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpretation, or other appropriate processing as necessary, and then stored in a computer memory.

[0153] It should be understood that each part of the present application can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one of the following techniques known in the art or a combination thereof can be used: discrete logic circuits with logic gate circuits for implementing logical functions on data signals, application-specific integrated circuits with suitable combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0154] Those of ordinary skill in the art can understand that all or part of the steps carried by the method of implementing the above embodiments can be completed by instructing relevant hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiments.

[0155] In addition, in each embodiment of the present application, each functional unit can be integrated in a processing module, or each unit can exist physically alone, or two or more units can be integrated in a module. The above integrated module can be implemented in the form of hardware or in the form of a software functional module. When the above integrated module is implemented in the form of a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0156] The above-mentioned storage medium can be a read-only memory, a magnetic disk, an optical disc, etc. Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and should not be construed as limiting the present application. Those of ordinary skill in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present application.

Claims

1. A method for detecting abnormal operation parameters of a power generation device based on power change, characterized in that, Including the following steps: Obtain the measured values of the operating parameters of the power generation equipment at different times. According to the measured values, analyze the expectations of the differences in the operating parameters between two adjacent times in different states, where the expectations in different states are analyzed using the power difference between two adjacent times; Based on the analysis results of the expectations in different states, perform on-line steady-state discrimination on the power generation equipment through the power difference; Screen out the steady-state parameters from the measured values according to the on-line steady-state discrimination results, and construct a characteristic test statistic according to the expectations of the differences in the operating parameters and the expectations of the power differences in the steady-state parameters; Determine the abnormal parameters in the steady-state parameters according to the characteristic test statistic.

2. The method according to claim 1, wherein The analysis of the expectations of the differences in the operating parameters between two adjacent times according to the measured values includes: Construct a first expression of the measured value composed of the true value and the random error, where the true value includes the normal value and the abnormal value, and the normal value is related to the power of the power generation equipment; Combine the first expression, the function of the power difference, and the difference of the abnormal values to determine a second expression of the difference in the operating parameters, and determine a third expression of the power difference according to the steady-state process and the non-steady-state process of the power generation equipment; Determine the expected formula of the power difference according to the third expression, and determine the expectations of the differences in the operating parameters in different states according to the second expression and the expected formula of the power difference.

3. The method according to claim 2, characterized in that, The on-line steady-state discrimination of the power generation equipment through the power difference includes: Determine the probability distribution of the power difference according to the third expression, and use the power difference as a characteristic index for steady-state and non-steady-state discrimination. Establish probability models of the power generation equipment in the steady state and the non-steady state respectively using the probability distribution; Regress the non-steady-state process through a Gaussian mixture model. Based on the obtained non-steady-state Gaussian mixture model, set the steady-state discrimination result that follows a Bernoulli distribution; Calculate the Bernoulli distribution parameter by counting the proportion of the number of steady-state samples in the training samples of the statistical model, and determine the probability distribution of the steady-state discrimination result according to the Bernoulli distribution parameter; Combine Bayes' formula and the probability distribution of the steady-state discrimination result to construct a non-steady-state discrimination formula.

4. The method according to claim 1, wherein The construction of the characteristic test statistic according to the expectations of the differences in the operating parameters and the expectations of the power differences in the steady-state parameters includes: Calculate the T statistic of the difference in the operating parameters and the power difference respectively, and calculate the test P value corresponding to each T statistic; Construct the characteristic test statistic using the test P values corresponding to each T statistic according to the influence of the expected change in the difference in the operating parameters on the expectation of the power difference.

5. The method according to claim 1, wherein The determination of the abnormal parameters in the steady-state parameters according to the characteristic test statistic includes: Calculate the characteristic test statistic of the preset normal value sample, and regress the characteristic test statistic of the normal value sample to obtain a probability density distribution model; Determine the abnormal detection threshold according to the probability density distribution model; Compare the characteristic test statistic of the current operating parameter with the anomaly detection threshold, and determine whether the current operating parameter is an abnormal parameter according to the comparison result.

6. The method according to claim 1, wherein After determining the abnormal parameter in the steady-state parameter according to the characteristic test statistic, it further includes: Obtain the parameter value of the abnormal parameter at a subsequent moment, and calculate the subsequent abnormal deviation degree of the abnormal parameter according to the parameter value at the subsequent moment; Verify the determination result of the abnormal parameter according to the subsequent abnormal deviation degree and the current abnormal deviation degree of the abnormal parameter.

7. The method according to claim 6, characterized in that, After determining the abnormal parameter in the steady-state parameter according to the characteristic test statistic, it further includes: Calculate the change trend of the abnormal parameter according to the parameter value of the abnormal parameter at a subsequent moment; Use the change trend of the abnormal parameter to perform fault diagnosis on the power generation equipment.

8. An abnormal detection device for operating parameters of a power generation device based on power change, characterized in that, It includes the following modules: An analysis module, configured to obtain the measured values of the operating parameters of the power generation equipment at different moments, and analyze the expectations of the differences between the operating parameters at two adjacent moments in different states, wherein the expectations in different states are analyzed using the power difference between two adjacent moments; A discrimination module, configured to perform online steady-state discrimination on the power generation equipment through the power difference based on the analysis results of the expectations in different states; A construction module, configured to screen out the steady-state parameters from the measured values according to the online steady-state discrimination result, and construct a characteristic test statistic according to the expectations of the differences between the operating parameters in the steady-state parameters and the expectation of the power difference; A determination module, configured to determine the abnormal parameter in the steady-state parameter according to the characteristic test statistic.

9. An electronic device, including: A processor; A memory for storing instructions executable by the processor; Wherein, the processor is configured to execute the instructions to implement the method for detecting abnormal operating parameters of a power generation equipment based on power change according to any one of claims 1-7.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for detecting abnormal operating parameters of a power generation equipment based on power change according to any one of claims 1-7.

Images