Abnormal state detection method and device for water-turbine generator set and computer equipment

Through the abnormal state detection model, the unit working conditions and stability monitoring parameters of the hydrowheel generator set are analyzed, which solves the problem of difficulty in accurately detecting the abnormal state of the equipment in the prior art, and achieves the safe and economical operation of the hydrowheel generator set.

CN119046837AActive Publication Date: 2024-11-29CHINA THREE GORGES CORPORATION

Patent Information

Application Number
CN202411143839.0
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-20
Publication Date
2024-11-29
Estimated Expiration
2044-08-20

AI Technical Summary

Technical Problem

It is difficult to accurately detect abnormal conditions of equipment when monitoring the status of existing water turbine generators, resulting in the inability to detect hidden dangers and defects in time, which may lead to the occurrence of destructive accidents.

Method used

Using an abnormal state detection model, by obtaining the unit working conditions and stability monitoring parameters of the water turbine generator set, using the encoder and mapping network to map the feature data, calculate the distance from the mapping features to the single-class center, and determine whether the equipment is in normal condition. state.

Benefits of technology

Accurate detection of abnormal states of hydrowheel generator sets is achieved, and destructive accidents are avoided, which is conducive to guiding the safe and economical operation of the unit.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119046837A_ABST
    Figure CN119046837A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of water and electricity, and discloses a water-turbine generator set abnormal state detection method and device and computer equipment, and the method comprises the steps: obtaining the current feature data of a water-turbine generator set, the feature data comprising set working condition parameters and stability monitoring parameters; inputting the feature data into a pre-trained abnormal state detection model to obtain mapping features of the feature data; calculating the distance from the mapping feature to a single-class center, wherein the single-class center is determined according to positive sample data collected when the water-turbine generator set is in a normal operation state; if the distance is smaller than a preset threshold value, judging that the water-turbine generator set is in a normal state. When the abnormal state of the water turbine generator set is detected, the influence of different working conditions on the running state of the water turbine generator set is considered, so that the abnormal state of the water turbine generator set can be accurately detected.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of hydropower technology, and in particular to a method and device for detecting abnormal state of a hydro-generator set, and computer equipment. Background Art

[0002] The hydro-turbine generator set is the core key equipment of the hydropower station. The condition monitoring technology of the hydro-turbine generator set can grasp and analyze its operating status at any time, discover the existing hidden dangers and defects in time, avoid the occurrence of destructive accidents, and is conducive to guiding the operation and maintenance of the unit, thereby ensuring the safe and economical operation of the unit.

[0003] The status monitoring data of the hydro-turbine generator set is multivariate time series data, which is high-dimensional and massive, and reflects the status information of the hydro-turbine generator set in real time. The current hydro-turbine generator monitoring system of the hydropower station collects information such as the swing and peak-to-peak value of the equipment, and uses a simple threshold method and slope method to warn of abnormal conditions. However, practice shows that the parameter characteristics of the unit are closely related to its operating conditions. How to scientifically mine monitoring data and accurately detect abnormal equipment conditions is a difficulty in monitoring the status of hydro-turbine generators. Summary of the invention

[0004] In view of this, the present invention provides a method, device and computer equipment for detecting abnormal state of a hydro-generator set, so as to solve the problem that the abnormal state of the equipment cannot be accurately detected.

[0005] In a first aspect, the present invention provides a method for detecting an abnormal state of a hydro-turbine generator set, comprising: obtaining current characteristic data of the hydro-turbine generator set, the characteristic data including unit operating parameters and stability monitoring parameters; inputting the characteristic data into a pre-trained abnormal state detection model to obtain mapping features of the characteristic data; calculating the distance from the mapping features to a single-class center, the single-class center being determined based on positive sample data collected when the hydro-turbine generator set is in a normal operating state; if the distance is less than a preset threshold, determining that the hydro-turbine generator set is in a normal state.

[0006] The abnormal state detection method of the hydro-turbine generator set provided by the embodiment of the present invention obtains the unit operating parameters and stability monitoring parameters of the hydro-turbine generator set, analyzes the unit operating parameters and stability monitoring parameters using the abnormal state detection model, obtains the mapping characteristics of the characteristic data, and determines whether the hydro-turbine generator set is in a normal state according to the distance between the mapping characteristics and the single class center. Since the operating state of the hydro-turbine generator set will be different under different working conditions, the abnormal state detection model fully learns the relationship between the different working conditions and monitoring parameters of the hydro-turbine generator set. When detecting the abnormal state of the hydro-turbine generator set, it can take into account the impact of different working conditions on the operating state of the hydro-turbine generator set, so that the abnormal state of the hydro-turbine generator set can be accurately detected, and the occurrence of destructive accidents can be avoided, which is conducive to guiding the operation and maintenance of the unit, thereby ensuring the safe and economical operation of the unit.

[0007] In an optional embodiment, the abnormal state detection model includes an encoder and a mapping network, the encoder is used to map the feature data to obtain initial multidimensional features, and send the initial multidimensional features to the mapping network; the mapping network is used to perform secondary mapping on the initial multidimensional features to obtain mapping features.

[0008] In an optional embodiment, the steps of training the abnormal state detection model include: obtaining training feature data of the hydro-turbine generator set at different time nodes; constructing multiple groups of sample data based on the training feature data of the hydro-turbine generator set at different time nodes; using the multiple groups of sample data to establish positive sample data and negative sample data; using the positive sample data to train the initial encoder to obtain the encoder; using the encoder, positive sample data and negative sample data to train the initial mapping network to obtain the mapping network.

[0009] In an optional embodiment, multiple groups of sample data are constructed based on the training features of the hydro-turbine generator set at different time nodes, including: constructing a first two-dimensional feature matrix based on the training feature data at different time nodes; normalizing the first two-dimensional feature matrix to obtain a second two-dimensional feature matrix; and dividing the second two-dimensional feature matrix in the time dimension by overlapping sampling to obtain multiple groups of sample data.

[0010] In an optional embodiment, the step of establishing negative sample data using multiple groups of sample data includes: keeping the unit operating parameters in each sample data unchanged, randomly replacing the original data of the stability parameters in the sample data with exceeded values ​​to obtain exceeded sample data; and normalizing the exceeded sample data to obtain negative sample data.

[0011] In an optional embodiment, the initial encoder is trained using positive sample data to obtain an encoder, including: keeping the unit operating parameters in each sample data unchanged, and randomly setting the stability parameters in the sample data to 0 according to a preset probability to obtain enhanced sample data; using the initial encoder to map the enhanced sample data to a first multidimensional feature; using a decoder network to reconstruct the first multidimensional feature into an original sample; each sample data and its corresponding original sample are respectively input into a first loss function to obtain a first loss value; if the sum of the first loss value does not meet a first preset condition, optimizing the parameters of the initial encoder and the decoder network, and returning to the step of mapping the enhanced sample data to the first multidimensional feature using the initial encoder, until the first loss value meets the first preset condition, and obtaining the encoder.

[0012] In an optional embodiment, an encoder, positive sample data and negative sample data are used to train an initial mapping network to obtain a mapping network, including: using the encoder to map the positive sample data and the negative sample data to a second multidimensional feature and a third multidimensional feature, respectively; using the initial mapping network to map the second multidimensional feature and the third multidimensional feature to a fourth multidimensional feature and a fifth multidimensional feature, respectively; using the distance between the fourth multidimensional feature and the single-class center to calculate the second loss value of the initial mapping network; using the fifth multidimensional feature, the single-class center and the preset radius of the hypersphere of the single-class center to calculate the third loss value of the initial mapping network; if the sum of the second loss value and the third loss value does not meet the second preset condition, optimizing the parameters of the initial mapping network, returning to the step of using the encoder to map the positive sample data and the negative sample data to the second multidimensional feature and the third multidimensional feature, respectively, until the sum of the second loss value and the third loss value meets the second preset condition, and obtaining the mapping network.

[0013] In an optional embodiment, the sample data includes multiple groups of data, and the step of determining the preset threshold includes: inputting each group of data in the positive sample data into the abnormal state detection model respectively, and obtaining the sixth multidimensional features corresponding to each group of data; calculating the distance between each sixth multidimensional feature and the single class center respectively; determining the preset threshold according to the distance corresponding to each sixth multidimensional feature, so that the proportion of distances less than the preset threshold meets the preset ratio.

[0014] In a second aspect, the present invention provides an abnormal state detection device for a hydro-turbine generator set, comprising: a data acquisition module, used to obtain current characteristic data of the hydro-turbine generator set, the characteristic data including unit operating parameters and stability monitoring parameters; a mapping feature acquisition module, used to input the characteristic data into a pre-trained abnormal state detection model to obtain mapping features of the characteristic data; a distance calculation module, used to calculate the distance from the mapping feature to a single class center, the single class center is determined based on positive sample data collected when the hydro-turbine generator set is in a normal operating state; a state detection module, if the distance is less than a preset threshold, the state detection module is used to determine that the hydro-turbine generator set is in a normal state.

[0015] In a third aspect, the present invention provides a computer device, comprising: a memory and a processor, the memory and the processor being communicatively connected to each other, the memory storing computer instructions, and the processor executing the method for detecting abnormal state of a hydro-turbine generator set according to the first aspect or any corresponding embodiment thereof by executing the computer instructions. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 is a flow chart of a method for detecting abnormal state of a hydro-generator set according to an embodiment of the present invention;

[0018] Figure 2 is a schematic diagram of a processing flow of data processing of feature data by an abnormal state detection model in an embodiment of the present invention;

[0019] Figure 3 is a schematic diagram of a process for training an abnormal state detection model in an embodiment of the present invention;

[0020] Figure 4 is a schematic diagram of obtaining multiple groups of sample data by dividing the second two-dimensional feature matrix in an overlapping sampling manner in an embodiment of the present invention;

[0021] Figure 5 is a schematic diagram of a flow chart of training an encoder in an embodiment of the present invention;

[0022] Figure 6 is a schematic diagram of a process of training a mapping network in an embodiment of the present invention;

[0023] Figure 7is a structural block diagram of a device for detecting abnormal state of a hydro-generator set according to an embodiment of the present invention;

[0024] Figure 8 It is a schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0026] According to an embodiment of the present invention, an embodiment of a method for detecting abnormal state of a hydro-turbine generator set is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0027] In this embodiment, a method for detecting abnormal state of a hydro-turbine generator set is provided, which can be used in the above-mentioned terminals, such as mobile phones, tablet computers, etc. Figure 1 is a flow chart of a method for detecting abnormal state of a hydro-generator set according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0028] Step S101, obtaining current characteristic data of the hydro-generator set, the characteristic data including the set operating parameters and stability monitoring parameters.

[0029] In an optional embodiment, the unit operating parameters include one or more of working water head, relay stroke, active power, reactive power, etc.

[0030] The working head refers to the water level difference between the inlet and outlet of the turbine generator set, that is, the head that the turbine can use. The greater the working head, the greater the output power of the turbine.

[0031] The servomotor stroke refers to the moving distance of the servomotor piston, which is also the opening of the turbine guide vane. The larger the servomotor stroke, the larger the opening of the turbine guide vane, the greater the flow through the turbine, and the output power will increase accordingly.

[0032] Active power refers to the actual power output by the turbine generator set, that is, the power on the turbine shaft. The size of active power depends on factors such as the turbine speed, water head and flow rate.

[0033] Reactive power refers to the reactive power output by the turbine unit, that is, the reactive power on the turbine shaft. The size of reactive power depends on factors such as the turbine speed, water head and flow rate.

[0034] The stability monitoring parameters are vibration and displacement indicators of the hydro-turbine generator set during operation, and are used to monitor and evaluate the operating status of the unit. Specifically, they include one or more of the upper guide X-direction swing, upper guide Y-direction swing, lower guide X-direction swing, lower guide Y-direction swing, water guide X-direction swing, water guide Y-direction swing, upper frame X-direction horizontal vibration peak-to-peak value, upper frame Y-direction horizontal vibration peak-to-peak value, upper frame Z-direction vertical vibration peak-to-peak value, lower frame X-direction horizontal vibration peak-to-peak value, lower frame Y-direction horizontal vibration peak-to-peak value, lower frame Z-direction vertical vibration peak-to-peak value, top cover X-direction horizontal vibration peak-to-peak value, top cover Y-direction horizontal vibration peak-to-peak value, top cover Z-direction vertical vibration peak-to-peak value, axial displacement peak-to-peak value, etc. during the operation of the hydro-turbine generator set.

[0035] Excessive swing, vibration or axial displacement may indicate that the unit is faulty or unbalanced and needs to be repaired and maintained in a timely manner. Through real-time monitoring and analysis of these parameters, problems can be discovered in advance and corresponding measures can be taken to avoid equipment damage and downtime accidents. Specific monitoring and evaluation standards are usually provided by the manufacturer of the hydro-turbine generator set or relevant technical specifications.

[0036] Under different operating conditions, the values ​​of the corresponding stability monitoring parameters of the hydro-generator set will also change due to different operating conditions. For example, the working water head and the relay stroke will affect the output power of the hydro-generator set, thereby affecting the vibration and displacement indicators of the set. For example, when the working water head increases, the output power of the hydro-generator set will also increase, resulting in an increase in the vibration and displacement indicators of the set. Active power and reactive power will also affect the operating state of the hydro-generator set, thereby affecting the vibration and displacement indicators of the set. For example, when the active power increases, the output power of the hydro-generator set will also increase, resulting in an increase in the vibration and displacement indicators of the set.

[0037] Therefore, in the embodiment of the present invention, when obtaining the characteristic parameters of the hydro-turbine generator set, the unit operating parameters and stability monitoring parameters are obtained, and the unit operating parameters and stability monitoring parameters are combined to analyze whether the hydro-turbine generator set is abnormal. The operating parameters of the hydro-turbine generator set can be fully considered, and the obtained abnormality detection results are more accurate, reducing misjudgment situations.

[0038] Step S102: input the feature data into a pre-trained abnormal state detection model to obtain mapping features of the feature data.

[0039] In an optional embodiment, the acquired feature data includes multiple parameters, which can be combined to form two-dimensional data. The two-dimensional data is input into an abnormal state detection model, and the two-dimensional data is mapped to a hypersphere centered on a single class through the abnormal state detection model to obtain mapping features.

[0040] Step S103, calculate the distance from the mapping feature to the single class center, the single class center is determined based on the positive sample data collected when the hydro-generator set is in normal operation. If the distance is less than the preset threshold, execute step S104; if the distance is greater than or equal to the preset threshold, execute step S105.

[0041] In an optional embodiment, the distance from the mapping feature to the single-class center may be Euclidean distance, cosine distance, or the like.

[0042] Step S104, determining whether the hydro-generator set is in a normal state.

[0043] Step S105, determining whether the hydro-generator set is in an abnormal state.

[0044] In an optional embodiment, the single-class center is obtained by clustering or other similarity measurement methods on the positive sample data collected when the hydro-turbine generator set is in normal operation. If the mapping feature of the hydro-turbine generator set is closer to the single-class center, it means that the hydro-turbine generator set is closer to normal operation. Therefore, the distance from the mapping feature to the single-class center can be combined to determine whether the hydro-turbine generator set is abnormal.

[0045] In an optional embodiment, the preset threshold may be determined based on relevant experience.

[0046] The abnormal state detection method of the hydro-turbine generator set provided by the embodiment of the present invention obtains the unit operating parameters and stability monitoring parameters of the hydro-turbine generator set, analyzes the unit operating parameters and stability monitoring parameters using the abnormal state detection model, obtains the mapping characteristics of the characteristic data, and determines whether the hydro-turbine generator set is in a normal state according to the distance between the mapping characteristics and the single class center. Since the operating state of the hydro-turbine generator set will be different under different working conditions, the abnormal state detection model fully learns the relationship between the different working conditions and the monitoring parameters of the hydro-turbine generator set. When detecting the abnormal state of the hydro-turbine generator set, it can take into account the influence of different working conditions on the operating state of the hydro-turbine generator set, so that the abnormal state of the hydro-turbine generator set can be accurately detected, and the occurrence of destructive accidents can be avoided, which is conducive to guiding the operation and maintenance of the unit, thereby ensuring the safe and economical operation of the unit.

[0047] In an optional embodiment, in the abnormal state detection model used in the above step S102, if Figure 2As shown, it includes an encoder and a mapping network. The encoder is used to map the feature data to obtain initial multi-dimensional features and send the initial multi-dimensional features to the mapping network; the mapping network is used to perform secondary mapping on the initial multi-dimensional features to obtain mapping features.

[0048] In an alternative embodiment, if Figure 3 As shown, the steps of training the abnormal state detection model include:

[0049] Step S301, obtaining training feature data of the hydro-generator set at different time nodes.

[0050] In an optional embodiment, the training feature data obtained when training the abnormal state detection model is of the same type as the feature data obtained in the above step S101.

[0051] Step S302, constructing multiple groups of sample data based on the training feature data of the hydro-generator set at different time nodes.

[0052] In an optional embodiment, after obtaining the training feature data of different time nodes, the feature data is divided into multiple groups of sample data with the time node as a dimension, and the number of time nodes corresponding to each group of sample data is the same.

[0053] Step S303: Create positive sample data and negative sample data using multiple groups of sample data.

[0054] In an optional embodiment, if the hydro-turbine generator set operates according to the parameters in the positive sample data, it indicates that the hydro-turbine generator set is in a normal operating state; if the hydro-turbine generator set operates according to the parameters in the negative sample data, it indicates that the hydro-turbine generator set is in an abnormal operating state.

[0055] In an optional embodiment, the positive sample data and the negative sample data may be determined in any manner. For example, when the hydro-generator set is in normal operation, step S301 may be executed to obtain training feature data of the hydro-generator set at different time nodes, and after forming multiple groups of sample data according to the training feature data, the multiple groups of sample data are used as positive sample data. The values ​​of one or more parameters in each group of sample data are replaced with abnormal values ​​to obtain negative sample data.

[0056] Step S304: train the initial encoder using the positive sample data to obtain an encoder.

[0057] In an optional embodiment, an autoencoder may be constructed and self-supervised training may be performed to obtain an encoder.

[0058] Step S305, using the encoder, positive sample data and negative sample data, the initial mapping network is trained to obtain a mapping network.

[0059] In an optional embodiment, the above step S302 specifically includes the following steps:

[0060] Step a1, constructing a first two-dimensional feature matrix according to the training feature data at different time nodes.

[0061] Exemplarily, after reading the feature data of multiple time nodes under normal conditions from the hydropower station computer monitoring system, a first two-dimensional feature matrix is ​​constructed with the time nodes as row dimensions and the features as column dimensions.

[0062] For example, if information in L nodes is collected, and 20 feature data are collected at each time node, the number of row dimensions is L, the number of column dimensions is 20, and the first two-dimensional feature matrix constructed is: The i-th feature sequence is recorded as row vector a i (1≤i≤20), the length of the vector is L, then the matrix

[0063] Step a2, normalize the first two-dimensional feature matrix to obtain a second two-dimensional feature matrix:

[0064]

[0065] a ij is the data of the i-th feature at the j-th time node, a imax represents the maximum value of the i-th feature, a imin represents the minimum value of the i-th feature, ε is a positive value, and ε is usually selected to be no greater than 10 -5 Positive value.

[0066] Step a3, in the time dimension, the second two-dimensional feature matrix is ​​divided in an overlapping sampling manner to obtain multiple groups of sample data.

[0067] In an optional embodiment, if the length of each sample is M, the sampling step length is S, that is, the overlap length is MS, then the number of samples obtained is:

[0068]

[0069] The function floor() represents rounding down, and the data set of the sample data is recorded as Among the samples

[0070] For example, Figure 4 As shown, for a two-dimensional data matrix A containing 5 time nodes, assuming that the length of the sample is 3 and the length of the sampling step is 1, the overlap length is 2, and a total of 3 samples can be obtained.

[0071] In an optional embodiment, after executing the above steps a1 to a3 to obtain multiple groups of sample data, the unit operating parameters in each sample data are kept unchanged, the original data of the stability parameter in the sample data is randomly replaced with the exceeded value to obtain the exceeded sample data, and the exceeded sample data is normalized to obtain the negative sample data. When generating negative sample data, the fewer the replacements of the exceeded values ​​of the stability parameter, the greater the difficulty of training the anomaly detection neural network, and the better the detection effect.

[0072] In an alternative embodiment, if Figure 5 As shown, the above step S304 specifically includes the following steps:

[0073] Step b1, keep each sample data x i The unit operating parameters in remain unchanged, and the sample data x is converted into i The stability parameters in are randomly set to 0, and the enhanced sample data x is obtained. i ′ .

[0074] In an optional embodiment, the larger the preset probability p is, the more difficult it is to perform self-supervised training of the autoencoder; and when p is too large, all stability monitoring parameters of the sample data are close to 0, making it difficult for the model to capture the connections between the parameters, and the performance is reduced; therefore, the effect of model training tends to first increase and then decrease as p increases.

[0075] Step b2: Use the initial encoder E to enhance the sample data x i ′ Mapped to the first multidimensional feature Z i , for example, if the enhanced sample data x i ′ Mapped to d-dimensional features, then,

[0076] Step b3, using the decoder network D to transform the first multidimensional feature Z i Reconstruct the original sample For example,

[0077] Step b4: convert the sample data x i and the original sample Input into the first loss function to get the first loss value.

[0078] In an optional embodiment, the first loss function is constructed based on the sum of the characteristic differences of each group of positive sample data, wherein the characteristic difference of a group of positive sample data refers to the group of positive sample data x i With the first multidimensional feature Z i difference.

[0079] Specifically, the first loss function is:

[0080]

[0081] Where N represents the number of positive sample data, represents the original sample, x i Represents positive sample data.

[0082] If the first loss value does not meet the first preset condition, the parameters of the initial encoder and decoder network are optimized, and the process returns to step b2 until the first loss value meets the first preset condition to obtain the encoder.

[0083] In an optional embodiment, the optimization function can be expressed as:

[0084]

[0085] If the above optimization function is used to optimize the initial encoder, the first preset condition is: the first loss value reaches the minimum value, and when the first loss value is the minimum value, the current initial encoder is used as the encoder.

[0086] In an optional embodiment, the first preset condition may be that the first loss value is less than a preset value, that is, when the first loss value is less than the preset value, the current initial encoder is used as the encoder in the abnormal state detection model.

[0087] In the above embodiment, when training the encoder in the abnormal state detection model, the operating condition parameters in the sample data are kept unchanged, and the stability monitoring parameters in the sample data are randomly set to 0 to form enhanced sample data. The enhanced sample data is used to train the initial encoder, and the monitoring data is randomly set to 0 with a certain probability. The encoder can automatically correct the data parameters set to 0 and learn the intrinsic connection between the operating condition and the monitoring parameters. Therefore, the abnormal state detection model can analyze the stability monitoring parameters in combination with the unit operating condition, thereby obtaining an abnormal detection result, and the result obtained in combination with the unit operating condition is more accurate.

[0088] In an alternative embodiment, if Figure 6 As shown, the above step S305 specifically includes the following steps:

[0089] Step c1, using an encoder to map the positive sample data and the negative sample data into a second multidimensional feature and a third multidimensional feature, respectively, where the second multidimensional feature and the third multidimensional feature are both d-dimensional features.

[0090] Step c2, using the initial mapping network to map the second multidimensional feature and the third multidimensional feature into the fourth multidimensional feature and the fifth multidimensional feature, respectively, the fourth multidimensional feature and the fifth multidimensional feature are both d ′ Dimensional features.

[0091] Step c3, calculating the second loss value of the initial mapping network using the distance between the fourth multidimensional feature and the single class center.

[0092] In an optional embodiment, the mapping features of the positive sample data are located on a hypersphere with a center c, and the smaller the distance from the center, the better. The loss function for calculating the second loss value is:

[0093]

[0094] Where N represents the number of positive sample data, G(E(x i )) represents the fourth multidimensional feature of the i-th group of positive sample data, c is d ′ Learnable single-class centers in dimensional feature space, λ1 represents the adjustable balance coefficient, J represents the number of layers of the neural network in the mapping network G, and W j is the j-th layer parameter of the mapping network G, ‖*‖ F In an optional embodiment, the neural network in the mapping network includes but is not limited to a convolutional neural network, a fully connected network, a Transformer network, a recurrent neural network, and the like.

[0095] Step c4, calculating the third loss value of the initial mapping network using the fifth multidimensional feature, the single-class center and the preset radius of the hypersphere of the single-class center.

[0096] In an optional embodiment, negative sample data is used to train the mapping network G. The mapping features of the negative sample data should be far away from the single-class center c, but need to be restricted to a hypersphere with a radius of r to reduce the open risk of the mapping feature space. The loss function for calculating the third loss value is:

[0097]

[0098] Among them, K represents the number of negative sample data, Represents the fifth multidimensional feature of the i-th group of negative sample data, c is d ′ Learnable single-class centers in dimensional feature space, r represents the preset radius of the hypersphere of the single-type center, λ2 is an adjustable parameter, ‖*‖ F represents the Frobenius norm of *.

[0099] If the sum of the second loss value and the third loss value does not satisfy the second preset condition, the parameters of the initial mapping network are optimized, and the process returns to step c1 until the sum of the second loss value and the third loss value satisfies the second preset condition, thereby obtaining a mapping network.

[0100] In an optional embodiment, the optimization function of the mapping network G is

[0101]

[0102] If the above optimization function is used to optimize the initial encoder, the second preset condition is: the sum of the second loss value and the third loss value reaches a minimum value. When the sum of the second loss value and the third loss value is a minimum value, the current initial mapping network is used as the mapping network.

[0103] In an optional embodiment, the second preset condition may be that the sum of the second loss value and the third loss value is less than a preset value, that is, when the sum of the second loss value and the third loss value is less than the preset value, the current initial mapping network is used as the mapping network.

[0104] In the above embodiment, when training the mapping network in the abnormal state detection model, a method of using positive and negative samples to train a single classifier is proposed, and the mapping network is increased. In addition to normal samples, artificially generated negative samples are used to train the network to increase the distance between positive and negative samples in the mapping feature space to improve the model's abnormality detection ability.

[0105] In an optional embodiment, the preset threshold used in judging whether the hydro-generator set is in a normal state is determined by the following steps:

[0106] Step d1, inputting each group of data in the positive sample data into the abnormal state detection model respectively, and obtaining the sixth multi-dimensional features corresponding to each group of data respectively.

[0107] Step d2, respectively calculate the distance between each sixth multidimensional feature and the single class center: d(G(E(x i )),c)=||G(E(x i ))-c||2.

[0108] Step d3: determining a preset threshold value according to the distances corresponding to the sixth multi-dimensional features, so that the proportion of distances smaller than the preset threshold value satisfies a preset ratio.

[0109] In this embodiment, a device for detecting abnormal state of a hydro-generator set is also provided, and the device is used to implement the above-mentioned embodiments and preferred implementation modes, and the descriptions that have been made are not repeated. As used below, the term "module" can implement a combination of software and / or hardware of a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also conceivable.

[0110] This embodiment provides a device for detecting abnormal state of a hydro-generator set. Figure 7 As shown, including:

[0111] The data acquisition module 701 is used to obtain the current characteristic data of the hydro-generator set, and the characteristic data includes the unit operating parameters and stability monitoring parameters.

[0112] A mapping feature acquisition module 702 is used to input feature data into a pre-trained abnormal state detection model to obtain mapping features of the feature data;

[0113] The distance calculation module 703 is used to calculate the distance from the mapping feature to the single class center, and the single class center is determined based on the positive sample data collected when the hydro-generator set is in normal operation;

[0114] The state detection module 704 is used to determine that the hydro-generator set is in a normal state if the distance is less than a preset threshold.

[0115] In some optional implementations, the normal state detection model includes an encoder and a mapping network,

[0116] The encoder is used to map the feature data to obtain initial multidimensional features and send the initial multidimensional features to the mapping network; the mapping network is used to perform secondary mapping on the initial multidimensional features to obtain mapping features.

[0117] In some optional implementations, the abnormal state detection model is trained by a model training module, and the model training module includes:

[0118] A characteristic data acquisition unit, used to acquire characteristic data of the hydro-generator set at different time nodes;

[0119] A sample construction unit, used to construct multiple groups of sample data according to characteristic data of the hydro-generator set at different time nodes;

[0120] A positive and negative sample construction unit, used to establish positive sample data and negative sample data using multiple groups of sample data;

[0121] An encoder training unit, used for training an initial encoder using positive sample data to obtain an encoder;

[0122] The mapping network training unit is used to train the initial mapping network using the encoder, positive sample data and negative sample data to obtain the mapping network.

[0123] In some optional embodiments, the sample construction unit includes:

[0124] A first two-dimensional feature matrix construction subunit, used to construct a first two-dimensional feature matrix according to feature data at different time nodes;

[0125] A normalization processing subunit, used for performing normalization processing on the first two-dimensional feature matrix to obtain a second two-dimensional feature matrix;

[0126] The feature division subunit is used to divide the second two-dimensional feature matrix in a time dimension in an overlapping sampling manner to obtain multiple groups of sample data.

[0127] In some optional embodiments, the positive and negative sample construction unit includes:

[0128] The negative sample construction subunit is used to keep the unit operating parameters in each sample data unchanged, randomly replace the original data of the stability parameter in the sample data with an exceeded value to obtain exceeded sample data, and normalize the exceeded sample data to obtain negative sample data.

[0129] In some optional implementations, the encoder training unit includes:

[0130] The enhanced sample construction subunit is used to keep the unit operating parameters in each sample data unchanged, and randomly set the stability parameters in the sample data to 0 according to a preset probability to obtain enhanced sample data.

[0131] A first data mapping subunit, configured to map the enhanced sample data into a first multi-dimensional feature using an initial encoder;

[0132] A feature reconstruction subunit, used to reconstruct the first multi-dimensional feature into an original sample using a decoder network;

[0133] A first loss value calculation unit, used for inputting each sample data and its corresponding original sample into a first loss function to obtain a first loss value;

[0134] The first optimization subunit optimizes the parameters of the initial encoder and decoder networks if the sum of the first loss values ​​does not meet the first preset condition, and returns to the step of mapping the enhanced sample data into the first multidimensional feature using the initial encoder until the first loss value meets the first preset condition to obtain the encoder.

[0135] In some optional implementations, the mapping network training unit includes:

[0136] A second data mapping subunit, used to map the positive sample data and the negative sample data into a second multidimensional feature and a third multidimensional feature respectively using an encoder;

[0137] A third data mapping subunit is used to map the second multi-dimensional feature and the third multi-dimensional feature into a fourth multi-dimensional feature and a fifth multi-dimensional feature respectively by using an initial mapping network;

[0138] A second loss value calculation subunit, used to calculate a second loss value of the initial mapping network using the fourth multi-dimensional feature and the distance between the single class center;

[0139] A third loss value calculation subunit, used to calculate a third loss value of the initial mapping network using the fifth multidimensional feature, the single-class center, and a preset radius of the hypersphere of the single-class center;

[0140] The second optimization subunit optimizes the parameters of the initial mapping network if the sum of the second loss value and the third loss value does not meet the second preset condition, and returns to the step of using the encoder to map the positive sample data and the negative sample data to the second multidimensional feature and the third multidimensional feature, respectively, until the sum of the second loss value and the third loss value meets the second preset condition, thereby obtaining the mapping network.

[0141] In some optional embodiments, the preset threshold is calculated by a preset threshold calculation module, and the preset threshold calculation module is used to input each group of data in the positive sample data into the abnormal state detection model respectively, to obtain the sixth multidimensional features corresponding to each group of data; to calculate the distance between each sixth multidimensional feature and the single class center respectively; to determine the preset threshold according to the distance corresponding to each sixth multidimensional feature, so that the proportion of distances less than the preset threshold satisfies the preset ratio.

[0142] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0143] The abnormal state detection device of the hydro-turbine generator set in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0144] The embodiment of the present invention also provides a computer device having the above Figure 8 The abnormal state detection device of the hydro-turbine generator set shown.

[0145] See also Figure 8 , Figure 8 is a schematic diagram of the structure of a computer device provided by an optional embodiment of the present invention, such as Figure 8As shown, the computer device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components are connected to each other using different buses for communication, and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple computer devices can be connected, and each device provides some necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 8 A processor 10 is taken as an example.

[0146] The processor 10 may be a central processing unit, a network processor or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be a dedicated integrated circuit, a programmable logic device or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic or any combination thereof.

[0147] The memory 20 stores instructions executable by at least one processor 10, so that at least one processor 10 executes the method shown in the above embodiment.

[0148] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely arranged relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0149] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid state drive; the memory 20 may also include a combination of the above types of memory.

[0150] The computer device also includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 8 The example of connecting through bus is taken in the following.

[0151] The input device 30 can receive input digital or character information, and generate key signal input related to the user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a track pad, a touch pad, an indicator bar, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 may include a display device, an auxiliary lighting device (e.g., an LED) and a tactile feedback device (e.g., a vibration motor), etc. The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0152] The embodiment of the present invention also provides a computer-readable storage medium. The method according to the embodiment of the present invention can be implemented in hardware, firmware, or can be implemented as a computer code that can be recorded in a storage medium, or can be implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and will be stored in a local storage medium through a network download, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state hard disk, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor, or hardware, the method shown in the above embodiment is implemented.

[0153] A part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of the computer program instruction in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc., and accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0154] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for detecting abnormal state of a hydro-generator set, characterized in that: The method comprises: Acquire the current characteristic data of the hydro-generator set, which includes the unit operating parameters and stability monitoring parameters; Inputting the feature data into a pre-trained abnormal state detection model to obtain mapping features of the feature data; Calculating the distance from the mapping feature to the single class center, wherein the single class center is determined based on positive sample data collected when the hydro-generator set is in a normal operating state; If the distance is less than a preset threshold, it is determined that the hydro-generator set is in a normal state.

2. The method according to claim 1, characterized in that The abnormal state detection model includes an encoder and a mapping network, The encoder is used to map the feature data to obtain initial multi-dimensional features, and send the initial multi-dimensional features to the mapping network; The mapping network is used to perform secondary mapping on the initial multi-dimensional features to obtain the mapping features.

3. The method according to claim 2, characterized in that The steps of training the abnormal state detection model include: Obtain training feature data of the hydro-generator set at different time nodes; Constructing multiple groups of sample data according to the training feature data of the hydro-generator set at different time nodes; Using the multiple groups of sample data to establish positive sample data and negative sample data; Using the positive sample data to train an initial encoder to obtain the encoder; The encoder, positive sample data and negative sample data are used to train an initial mapping network to obtain the mapping network.

4. The method according to claim 3, characterized in that The constructing of multiple groups of sample data according to the training features of the hydro-generator set at different time nodes includes: Constructing a first two-dimensional feature matrix according to the training feature data at different time nodes; Normalizing the first two-dimensional feature matrix to obtain a second two-dimensional feature matrix; In the time dimension, the second two-dimensional feature matrix is ​​divided in an overlapping sampling manner to obtain multiple groups of sample data.

5. The method according to claim 3 or 4, characterized in that: The step of establishing negative sample data using the multiple groups of sample data comprises: Keeping the unit operating condition parameters in each of the sample data unchanged, randomly replacing the original data of the stability parameters in the sample data with the exceeded values, to obtain the exceeded sample data; The excessive sample data is normalized to obtain the negative sample data.

6. The method according to claim 3, characterized in that The using the positive sample data to train the initial encoder to obtain the encoder includes: Keeping the unit operating condition parameters in each of the sample data unchanged, randomly setting the stability parameters in the sample data to 0 according to a preset probability, to obtain enhanced sample data; Mapping the enhanced sample data into a first multidimensional feature using an initial encoder; Reconstructing the first multidimensional features into original samples using a decoder network; Input each of the sample data and its corresponding original sample into a first loss function to obtain a first loss value; If the sum of the first loss values ​​does not meet the first preset condition, the parameters of the initial encoder and the decoder network are optimized, and the step of mapping the enhanced sample data into the first multidimensional feature using the initial encoder is returned to until the first loss value meets the first preset condition to obtain the encoder.

7. The method according to claim 3 or 6, characterized in that: The method of training the initial mapping network by using the encoder, the positive sample data and the negative sample data to obtain the mapping network comprises: Mapping the positive sample data and the negative sample data into a second multidimensional feature and a third multidimensional feature respectively using the encoder; Mapping the second multidimensional feature and the third multidimensional feature into a fourth multidimensional feature and a fifth multidimensional feature respectively using an initial mapping network; Calculating a second loss value of the initial mapping network using the distance between the fourth multidimensional feature and the single-class center; Calculating a third loss value of the initial mapping network using the fifth multidimensional feature, the single-class center, and a preset radius of the hypersphere of the single-class center; If the sum of the second loss value and the third loss value does not meet the second preset condition, the parameters of the initial mapping network are optimized, and the step of using the encoder to map the positive sample data and the negative sample data into second multidimensional features and third multidimensional features, respectively, is returned to until the sum of the second loss value and the third loss value meets the second preset condition, thereby obtaining the mapping network.

8. The method according to claim 3, characterized in that The sample data includes multiple groups of data, and the step of determining the preset threshold includes: Inputting each group of data in the positive sample data into the abnormal state detection model respectively to obtain the sixth multi-dimensional features corresponding to each group of data; Calculating the distance between each of the sixth multidimensional features and the single class center respectively; The preset threshold is determined according to the distances corresponding to the sixth multi-dimensional features, so that the proportion of distances less than the preset threshold satisfies a preset ratio.

9. A device for detecting abnormal state of a hydro-generator set, characterized in that: The device comprises: The data acquisition module is used to obtain the current characteristic data of the hydro-generator set, which includes the unit operating parameters and stability monitoring parameters; A mapping feature acquisition module, used for inputting the feature data into a pre-trained abnormal state detection model to obtain mapping features of the feature data; A distance calculation module, used to calculate the distance from the mapping feature to the single class center, wherein the single class center is determined based on positive sample data collected when the hydro-generator set is in normal operation; A state detection module is used to determine that the hydro-generator set is in a normal state if the distance is less than a preset threshold.

10. A computer device, characterized in that: include: A memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the method for detecting abnormal state of a hydro-turbine generator set according to any one of claims 1 to 8 by executing the computer instructions.

Citation Information

Patent Citations

  • Semi-supervised anomaly intrusion detection method

    CN101980480A

  • Hydropower unit health evaluation method

    CN108375476A

  • Hydroelectric generating set multi-dimensional vibration area fine division method based on decision tree model

    CN111092442A

  • Water turbine runner real-time state evaluation method and application thereof

    CN113027658A

  • Online fault diagnosis method for aero-engine

    CN114741945A

Cited By

  • Method for monitoring state of water turbine generator set of industrial circulating water power generation system

    CN122792267A

  • Method for monitoring state of water turbine generator set of industrial circulating water power generation system

    CN122792267B

  • Abnormal state detection method and apparatus for hydro-turbine generator set, and computer device

    WO2026040288A1