Hydropower station electrical equipment fault prediction method and system

Abstract

The invention relates to a hydropower station electrical equipment fault prediction method and system, and the method comprises the following steps: obtaining the operation data of hydropower station electrical equipment, the operation data comprising current, voltage and temperature; carrying out missing value filling processing on missing data of the operation data by using the operation data through an interpolation algorithm; extracting the characteristics of each dimension of the operation data by using a wavelet transform algorithm; setting a dependency degree threshold value, calculating the dependency degree of the features and a preset tag by using a mutual information algorithm, and filtering the features of which the dependency degree is smaller than the dependency degree threshold value to obtain an equipment feature set; and constructing an electrical equipment fault prediction model by using the operation data and the equipment feature set, and outputting equipment fault information through the electrical equipment fault prediction model. By calculating the dependency degree between the features and the preset labels and setting the dependency degree threshold value to screen the features, redundant features can be effectively removed, features highly related to fault prediction are reserved, and the accuracy and efficiency of the model are improved.

Description

Technical Field

[0001] The invention relates to a method and system for predicting faults of electrical equipment in a hydropower station, belonging to the technical field of power systems and hydropower engineering. Background Art

[0002] As an important power energy production facility, the stable operation of the electrical equipment of a hydropower station is crucial to ensure the reliability and safety of power supply. With the continuous expansion of the scale of hydropower stations and the continuous upgrading of technology, the complexity of electrical equipment and the diversity of operating environments are also increasing. Traditional electrical equipment fault detection methods mainly rely on regular manual inspections and simple threshold alarm mechanisms. These methods can detect obvious faults to a certain extent, but it is often difficult to identify potential and complex fault modes in a timely and accurate manner. In recent years, with the rapid development of sensor technology, data acquisition technology, and data analysis technology, data-driven fault prediction methods have gradually become a research hotspot. These methods can predict possible equipment failures in advance through real-time collection and analysis of electrical equipment operation data, thereby providing a scientific basis for equipment maintenance and management. However, the existing fault prediction methods still have some shortcomings in practical applications, which limits their wide application in fault prediction of electrical equipment in hydropower stations. Most of the existing fault prediction models are based on a single algorithm or a fixed model structure, which is difficult to adapt to the operating characteristics and fault modes of electrical equipment in different hydropower stations. In addition, traditional models often have problems such as high computational complexity and long training time when processing large-scale data, which limits their application in real-time fault prediction.

[0003] The patent document with the patent number "CN115546558A" discloses a method, device and storage medium for classifying insulation fault states of electrical equipment. The problem with this method is that although the DBSCAN clustering algorithm can process discrete points and adaptively calculate parameters, the DBSCAN algorithm is still sensitive to the selection of parameters (such as eps and Minpts), especially in the case of uneven data distribution, which may lead to inaccurate clustering results. Although multi-source data such as leakage current and infrared images are fused, the data fusion mainly focuses on feature splicing and dimensionality reduction, and lacks in-depth analysis of the intrinsic correlation of multi-source data. Summary of the invention

[0004] In order to solve the above problems existing in the prior art, the present invention proposes a method and system for predicting faults of electrical equipment in a hydropower station.

[0005] The technical solution of the present invention is as follows:

[0006] In one aspect, the present invention provides a method for predicting faults of electrical equipment in a hydropower station, comprising the following steps:

[0007] Acquiring operating data of electrical equipment of a hydropower station, wherein the operating data includes current, voltage and temperature;

[0008] The missing data of the running data are filled with missing values ​​using the running data through interpolation algorithms;

[0009] Use wavelet transform algorithm to extract features of each dimension of running data;

[0010] A dependency threshold is set, a mutual information algorithm is used to calculate the dependency between the feature and the preset label, and features whose dependency is less than the dependency threshold are filtered to obtain a device feature set;

[0011] An electrical equipment fault prediction model is constructed using operating data and equipment feature sets, and equipment fault information is output through the electrical equipment fault prediction model.

[0012] As a preferred implementation, the method for filling missing values ​​is:

[0013]

[0014]

[0015]

[0016]

[0017] in, represents the filling value of the running data at time t, x t-i represents the data at the i-th moment before the running data at time t, x t+j represents the data at time j after the running data at time t, w j (t), w i (t) represents the adaptive weight, n, m, N represent the preset maximum number, t, j, i represent the time index, represents the mean of known running data, σ represents the standard deviation of known running data, x k represents the kth known running data, k represents the digit index, exp() represents the exponential function, |t-(ti)| 2 、|t-(t+j)| 2 Indicates time distance.

[0018] As a preferred implementation, the feature extraction method is:

[0019]

[0020]

[0021]

[0022]

[0023] Where W(a,b) represents the result of wavelet transform, a represents the preset scale parameter, b represents the preset position parameter, and x(t) represents the running data at time t. represents the complex conjugate of the wavelet function, represents the scaling and translation transformation of the wavelet function, λ(a) represents the multi-scale adaptive threshold, ρ(a) represents the noise standard deviation, represents the correlation factor between the preset scale parameter a and the preset maximum number N, represents the average wavelet transform value, W(a,b k ) indicates that the scale parameter a and the kth position parameter b are preset. k The result of wavelet transform.

[0024] As a preferred implementation, let X be the feature of the kth dimension of the running data x(t), and Y be the preset label of feature X;

[0025] The feature screening method is:

[0026]

[0027]

[0028]

[0029]

[0030] Where I(X;Y) represents the degree of dependence between feature X and preset label Y, p(x,y) represents the joint probability density function, p(x) and p(y) represent the marginal probability density functions, β represents the preset redundancy adjustment coefficient, x represents the random value of feature X, and y represents the random value of preset label Y. Indicates all possible values ​​of the preset label Y except y The sum of the marginal probability density functions, δ(xx k ),δ(yy k ) represents the Dirac function, x k Represents the kth random value x, y k represents the kth random value y;

[0031] Set the dependency threshold, filter the features whose dependency is less than the dependency threshold, and use the filtered features as the device feature set {∈ 1 ,...,∈ 2 ,...∈ k},∈ k is the kth device feature in the device feature set.

[0032] As a preferred implementation, the electrical equipment fault prediction model is expressed as:

[0033]

[0034]

[0035]

[0036]

[0037] Among them, w T represents the transpose of the weight vector w, ξ k represents the slack variable ξ of the kth running data, C represents the adjustment parameter, y k represents the kth random value of the preset label Y, Represents the prediction result, represents the kernel function, MSE train represents the mean square error of the running data, b represents the bias term, ω 2 (x) represents ∈ k The square of the adaptive bandwidth.

[0038] On the other hand, the present invention also provides a hydropower station electrical equipment fault prediction system, comprising the following steps:

[0039] Data acquisition module: obtains the operating data of the electrical equipment of the hydropower station, the operating data including current, voltage and temperature;

[0040] Data processing module: Use the running data to fill the missing data of the running data through the interpolation algorithm;

[0041] Feature extraction module: uses wavelet transform algorithm to extract features of each dimension of running data;

[0042] Feature screening module: setting a dependency threshold, using a mutual information algorithm to calculate the dependency between the feature and a preset tag, and filtering features whose dependency is less than the dependency threshold to obtain a device feature set;

[0043] Model prediction module: Use operating data and equipment feature sets to build an electrical equipment fault prediction model, and output equipment fault information through the electrical equipment fault prediction model.

[0044] The present invention has the following beneficial effects:

[0045] The present invention fills the missing values ​​in the operation data by using an interpolation algorithm. By combining forward and backward data and introducing adaptive weights and statistical features (such as mean, standard deviation, etc.), the missing values ​​can be effectively filled, the impact of data loss on fault prediction can be reduced, and the integrity and reliability of the data can be improved. By filling the missing values, the continuity and integrity of the data are ensured, and a high-quality data foundation is provided for subsequent feature extraction and model construction, thereby improving the performance of the entire fault prediction system. The wavelet transform algorithm is used to extract the features of each dimension of the operation data. The wavelet transform can decompose the data into wavelet coefficients at different scales and positions, thereby capturing the subtle changes of data such as current, voltage and temperature at different time scales, and extracting more representative and deep features. By calculating the degree of dependence between the features and the preset labels and setting the degree of dependence threshold to filter the features, the redundant features can be effectively removed, the features highly related to fault prediction can be retained, and the accuracy and efficiency of the model can be improved. The constructed fault prediction model combines the operation data and the filtered device feature set, and can dynamically adjust the model parameters (such as weight vectors, slack variables, kernel functions, etc.) according to the actual data distribution, thereby improving the adaptability and prediction accuracy of the model to different fault modes. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 The present invention is a flowchart for implementing the method. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] It should be understood that the step numbers used in this document are only for convenience of description and are not intended to limit the order in which the steps are executed.

[0049] It should be understood that the terms used in the present specification are only for the purpose of describing specific embodiments and are not intended to limit the present invention. As used in the present specification and the appended claims, unless the context clearly indicates otherwise, the singular forms "a", "an" and "the" are intended to include plural forms.

[0050] The terms “include” and “comprising” indicate the presence of described features, integers, steps, operations, elements and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or combinations thereof.

[0051] The term "and / or" means and includes any and all possible combinations of one or more of the associated listed items.

[0052] Embodiment 1:

[0053] See also Figure 1 The present invention provides a method for predicting faults of electrical equipment in a hydropower station, comprising the following steps:

[0054] Acquiring operating data of electrical equipment of a hydropower station, wherein the operating data includes current, voltage and temperature;

[0055] Using the operating data through the interpolation algorithm to fill the missing data of the operating data (known operating data);

[0056] Use wavelet transform algorithm to extract features of each dimension of running data;

[0057] A dependency threshold is set, and a mutual information algorithm is used to calculate the dependency between the feature and a preset label (including a label indicating that the feature belongs to current, voltage or temperature, and a specific numerical label in the range of [1,10]), and the features whose dependency is less than the dependency threshold are filtered to obtain a device feature set;

[0058] An electrical equipment fault prediction model is constructed using operating data and equipment feature sets, and equipment fault information is output through the electrical equipment fault prediction model.

[0059] As a preferred implementation, the method for filling missing values ​​is:

[0060]

[0061]

[0062]

[0063]

[0064] in, represents the filling value of the running data at time t, x t-i represents the data at the i-th moment before the running data at time t, x t+j represents the data at time j after the running data at time t, w j (t), w i (t) represents the adaptive weight, n, m, N represent the preset maximum number, t, j, i represent the time index, represents the mean of known running data, σ represents the standard deviation of known running data, x krepresents the kth known running data, k represents the digit index, exp() represents the exponential function, |t-(ti)| 2 、|t-(t+j)| 2 Indicates time distance.

[0065] As a preferred implementation, the feature extraction method is:

[0066]

[0067]

[0068]

[0069]

[0070] Wherein, W(a,b) represents the result of wavelet transform (it can decompose the data of each dimension of the operating data, such as current, voltage or temperature, into wavelet coefficients at different scales and positions, thereby operating the characteristics of the data at different scales), a represents the preset scale parameter, b represents the preset position parameter, and x(t) represents the operating data at time t. represents the complex conjugate of the wavelet function, represents the scaling and translation transformation of the wavelet function, λ(a) represents the multi-scale adaptive threshold, ρ(a) represents the noise standard deviation, represents the correlation factor between the preset scale parameter a and the preset maximum number N, represents the average wavelet transform value, W(a,b k ) indicates that the scale parameter a and the kth position parameter b are preset. k The result of wavelet transform.

[0071] As a preferred implementation, let X be the feature of the kth dimension of the operating data x(t) (the feature of current, voltage or temperature, if k is 2, it means the feature of voltage), and Y be the preset label of the feature X;

[0072] The feature screening method is:

[0073]

[0074]

[0075]

[0076]

[0077] Where I(X;Y) represents the degree of dependence between feature X and preset label Y, p(x,y) represents the joint probability density function, p(x) and p(y) represent the marginal probability density functions, β represents the preset redundancy adjustment coefficient, x represents the random value of feature X, and y represents the random value of preset label Y. Indicates all possible values ​​of the preset label Y except y The sum of the marginal probability density functions, δ(xx k ),δ(yy k ) represents the Dirac function, x k Represents the kth random value x, y k represents the kth random value y;

[0078] Set the dependency threshold, filter the features whose dependency is less than the dependency threshold, and use the filtered features as the device feature set {∈ 1 ,...,∈ 2 ,...∈ k},∈ k is the kth device feature in the device feature set.

[0079] As a preferred implementation, the electrical equipment fault prediction model is expressed as:

[0080]

[0081]

[0082]

[0083]

[0084] Among them, w T represents the transpose of the weight vector w, ξ k represents the slack variable ξ of the kth running data, C represents the adjustment parameter, y k represents the kth random value of the preset label Y, Represents the prediction result, represents the kernel function, MSE train represents the mean square error of the running data, b represents the bias term, ω 2 (x) represents ∈ k The square of the adaptive bandwidth is preset according to the distribution of the device feature set. Subject to represents the dependency. w, ξ, and b are solved by solving the above optimization problem min w,b,ξ {} to determine.

[0085] Embodiment 2:

[0086] The present invention also provides a hydropower station electrical equipment fault prediction system, comprising the following steps:

[0087] Data acquisition module: obtains the operating data of the electrical equipment of the hydropower station, the operating data including current, voltage and temperature;

[0088] Data processing module: Use the running data to fill the missing data of the running data through the interpolation algorithm;

[0089] Feature extraction module: uses wavelet transform algorithm to extract features of each dimension of running data;

[0090] Feature screening module: setting a dependency threshold, using a mutual information algorithm to calculate the dependency between the feature and a preset tag, and filtering features whose dependency is less than the dependency threshold to obtain a device feature set;

[0091] Model prediction module: Use operating data and equipment feature sets to build an electrical equipment fault prediction model, and output equipment fault information through the electrical equipment fault prediction model.

[0092] The system is used to implement the method in Example 1, which will not be described in detail here.

[0093] In the embodiments of the present application, "at least one" refers to one or more, and "more than one" refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. Among them, A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following" and similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b and c can be represented by: a, b, c, a and b, a and c, b and c, or a and b and c, where a, b, c can be single or multiple.

[0094] Those of ordinary skill in the art will appreciate that the various units and algorithm steps described in the embodiments disclosed herein can be implemented in a combination of electronic hardware, computer software, and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0095] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0096] In several embodiments provided in the present application, if any function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (Read-Only Memory; hereinafter referred to as: ROM), random access memory (Random Access Memory; hereinafter referred to as: RAM), disk or optical disk, and other media that can store program codes.

[0097] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for predicting faults of electrical equipment in a hydropower station, characterized in that: The following steps are involved: Acquiring operating data of electrical equipment of a hydropower station, wherein the operating data includes current, voltage and temperature; The missing data of the running data are filled with missing values ​​using the running data through interpolation algorithms; Use wavelet transform algorithm to extract features of each dimension of running data; A dependency threshold is set, a mutual information algorithm is used to calculate the dependency between the feature and the preset label, and features whose dependency is less than the dependency threshold are filtered to obtain a device feature set; An electrical equipment fault prediction model is constructed using operating data and equipment feature sets, and equipment fault information is output through the electrical equipment fault prediction model.

2. The method for predicting faults of electrical equipment in a hydropower station according to claim 1, characterized in that: The method to fill missing values ​​is: in, represents the filling value of the running data at time t, x t-i represents the data at the i-th moment before the running data at time t, x t+j represents the data at time j after the running data at time t, w j (t), w i (t) represents the adaptive weight, n, m, N represent the preset maximum number, t, j, i represent the time index, represents the mean of known running data, σ represents the standard deviation of known running data, x k represents the kth known running data, k represents the digit index, exp() represents the exponential function, |t-(ti)| 2 、|t-(t+j)| 2 Indicates time distance.

3. The method for predicting faults of electrical equipment in a hydropower station according to claim 2, characterized in that: The feature extraction method is: Where W(a,b) represents the result of wavelet transform, a represents the preset scale parameter, b represents the preset position parameter, and x(t) represents the running data at time t. represents the complex conjugate of the wavelet function, represents the scaling and translation transformation of the wavelet function, λ(a) represents the multi-scale adaptive threshold, ρ(a) represents the noise standard deviation, represents the correlation factor between the preset scale parameter a and the preset maximum number N, represents the average wavelet transform value, W(a,b k ) indicates that the scale parameter a and the kth position parameter b are preset. k The result of wavelet transform.

4. The method for predicting faults of electrical equipment in a hydropower station according to claim 3, characterized in that: Let X be the feature of the kth dimension of the running data x(t), and Y be the preset label of feature X; The feature screening method is: Where I(X;Y) represents the degree of dependence between feature X and preset label Y, p(x,y) represents the joint probability density function, p(x) and p(y) represent the marginal probability density functions, β represents the preset redundancy adjustment coefficient, x represents the random value of feature X, and y represents the random value of preset label Y. Indicates all possible values ​​of the preset label Y except y The sum of the marginal probability density functions, δ(xx k ),δ(yy k ) represents the Dirac function, x k Represents the kth random value x, y k represents the kth random value y; Set a dependency threshold, filter out features with a dependency less than the dependency threshold, and use the filtered features as the device feature set {∈1,...,∈2,...∈ k },∈ k is the kth device feature in the device feature set.

5. The method for predicting faults of electrical equipment in a hydropower station according to claim 4, characterized in that: The electrical equipment failure prediction model is expressed as: Among them, w T represents the transpose of the weight vector w, ξ k represents the slack variable ξ of the kth running data, C represents the adjustment parameter, y k represents the kth random value of the preset label Y, Represents the prediction result, represents the kernel function, MSE train represents the mean square error of the running data, b represents the bias term, ω 2 (x) represents ∈ k The square of the adaptive bandwidth.

6. A fault prediction system for electrical equipment in a hydropower station, characterized in that: The following steps are involved: Data acquisition module: obtains the operating data of the electrical equipment of the hydropower station, the operating data including current, voltage and temperature; Data processing module: Use the running data to fill the missing data of the running data through the interpolation algorithm; Feature extraction module: uses wavelet transform algorithm to extract features of each dimension of running data; Feature screening module: setting a dependency threshold, using a mutual information algorithm to calculate the dependency between the feature and a preset tag, and filtering features whose dependency is less than the dependency threshold to obtain a device feature set; Model prediction module: Use operating data and equipment feature sets to build an electrical equipment fault prediction model, and output equipment fault information through the electrical equipment fault prediction model.

7. The hydropower station electrical equipment fault prediction system according to claim 6, characterized in that: The method for filling missing values ​​in the data processing module is: in, represents the filling value of the running data at time t, x t-i represents the data at the i-th moment before the running data at time t, x t+j represents the data at time j after the running data at time t, w j (t), w i (t) represents the adaptive weight, n, m, N represent the preset maximum number, t, j, i represent the time index, represents the mean of known running data, σ represents the standard deviation of known running data, x k represents the kth known running data, k represents the digit index, exp() represents the exponential function, |t-(ti)| 2 、|t-(t+j)| 2 Indicates time distance.

8. The hydropower station electrical equipment fault prediction system according to claim 7, characterized in that: The feature extraction module, the feature extraction method is: Where W(a,b) represents the result of wavelet transform, a represents the preset scale parameter, b represents the preset position parameter, and x(t) represents the running data at time t. represents the complex conjugate of the wavelet function, represents the scaling and translation transformation of the wavelet function, λ(a) represents the multi-scale adaptive threshold, ρ(a) represents the noise standard deviation, represents the correlation factor between the preset scale parameter a and the preset maximum number N, represents the average wavelet transform value, W(a,b k ) indicates that the scale parameter a and the kth position parameter b are preset. k The result of wavelet transform.

9. The hydropower station electrical equipment fault prediction system according to claim 8, characterized in that: In the feature screening module, let X be the feature of the kth dimension of the running data x(t), and Y be the preset label of feature X; The feature screening method is: Where I(X;Y) represents the degree of dependence between feature X and preset label Y, p(x,y) represents the joint probability density function, p(x) and p(y) represent the marginal probability density functions, β represents the preset redundancy adjustment coefficient, x represents the random value of feature X, and y represents the random value of preset label Y. Indicates all possible values ​​of the preset label Y except y The sum of the marginal probability density functions, δ(xx k ),δ(yy k ) represents the Dirac function, x k Represents the kth random value x, y k represents the kth random value y; Set a dependency threshold, filter out features with a dependency less than the dependency threshold, and use the filtered features as the device feature set {∈1,...,∈2,...∈ k },∈ k is the kth device feature in the device feature set.

10. The hydropower station electrical equipment fault prediction system according to claim 9, characterized in that: The model prediction module, the electrical equipment fault prediction model is expressed as: Among them, w T represents the transpose of the weight vector w, ξ k represents the slack variable ξ of the kth running data, C represents the adjustment parameter, y k represents the kth random value of the preset label Y, Represents the prediction result, represents the kernel function, MSE train represents the mean square error of the running data, b represents the bias term, ω 2 (x) represents ∈ k The square of the adaptive bandwidth.

Images