Generator set active power low frequency oscillation early warning system

Abstract

The invention belongs to the technical field of power system monitoring, and discloses a generator set active power low-frequency oscillation early warning system, which comprises a sampling data processing and analyzing unit used for collecting and preprocessing active power sampling data of a generator set, and labeling and classifying the preprocessed active power sampling data according to waveform characteristics; the hidden Markov model construction unit is used for constructing and training a hidden Markov model according to the low-frequency oscillation characteristic data set and in combination with different working states of the generator set; and the low-frequency oscillation early warning analysis unit is used for acquiring active power data of the generator set in real time and calculating early warning of low-frequency oscillation of the generator set by using a forward-backward algorithm. According to the invention, high-precision prediction and evaluation can be carried out on low-frequency oscillation of the generator set, so that efficient, safe and reliable operation support is provided for the generator set.

Description

Technical Field

[0001] The invention relates to the technical field of power system monitoring, and in particular to a low-frequency oscillation early warning system for active power of a generator set. Background Art

[0002] With the rapid development of smart grid technology, the scale of the power grid has continued to expand, and has gradually developed from a regional power grid to a national grid. The operation mode and structure of regional interconnected power grids have become increasingly complex, and the interconnection between regions is relatively weak, which greatly increases the difficulty of power grid security and stability. Low-frequency oscillation has become an important issue in the stable operation of power systems. The frequency of low-frequency oscillation is generally between 0.2 and 2.5 Hz. Its occurrence may cause overcurrent tripping of the tie line, loss of step between systems or between units and systems, and seriously threaten the stability and safety of the power grid.

[0003] Timely and accurate identification and analysis of low-frequency oscillation characteristics are crucial to avoiding the potential hazards caused by low-frequency oscillations. By issuing early warning information, the power sector can take corresponding measures to suppress the occurrence of serious low-frequency oscillations, thereby effectively improving the stability and security of the power grid. However, current low-frequency oscillation detection technology usually relies on manually set thresholds and simple rules for oscillation identification. This approach is easily affected by noise and external disturbances, resulting in false alarms or missed alarms, which affects the accuracy of detection and, in turn, the security and stability of the system.

[0004] Therefore, how to provide a low-frequency oscillation early warning system for the active power of a generator set is a problem that needs to be solved urgently. Summary of the invention

[0005] The embodiment of the present invention provides a generator set active power low-frequency oscillation early warning system to solve the problems in the prior art.

[0006] In order to have a basic understanding of some aspects of the disclosed embodiments, a brief summary is given below. This summary is not a general review, nor is it intended to identify key / important components or to describe the scope of protection of these embodiments. Its only purpose is to present some concepts in a simple form as a prelude to the detailed description that follows.

[0007] According to a first aspect of an embodiment of the present invention, a generator set active power low-frequency oscillation early warning system is provided.

[0008] In one embodiment, a generator set active power low-frequency oscillation early warning system includes:

[0009] The sampling data processing and analysis unit is used to collect the active power sampling data of the generator set and perform preprocessing, and use the sliding bilateral cumulative sum algorithm to mark and classify the preprocessed active power sampling data according to the waveform characteristics to obtain a low-frequency oscillation feature data set;

[0010] A hidden Markov model building unit is used to build and train a hidden Markov model through transition probability analysis based on a low-frequency oscillation feature data set and in combination with different working states of the generator set;

[0011] The low-frequency oscillation early warning analysis unit is used to obtain the active power data of the generator set in real time, and calculate the early warning of the low-frequency oscillation of the generator set based on the trained hidden Markov model and the forward-backward algorithm.

[0012] In one embodiment, the sampling data processing and analysis unit includes:

[0013] A data preprocessing module is used to preprocess the active power sampling data of the generator set by using filtering technology and normalization method;

[0014] The low-frequency oscillation feature judgment module is used to input the pre-processed active power sampling data into the sliding bilateral cumulative sum algorithm, calculate the cumulative deviation of each window according to the preset sliding window length, and judge whether there is a low-frequency oscillation feature in each window according to the cumulative deviation;

[0015] The data set construction module is used to perform waveform feature annotation according to the judgment result, and based on the waveform feature annotation result, extract the characteristic parameters of the low-frequency oscillation to form a low-frequency oscillation characteristic data set.

[0016] In one embodiment, the low frequency oscillation characteristic determination module includes:

[0017] The mean value calculation submodule is used to initialize the window length of the active power sampling data and calculate the mean value of the active power sampling data points in each window;

[0018] The cumulative deviation calculation submodule is used to calculate the deviation based on the active power sampling data points in each window and its mean value, and accumulate the absolute values ​​of all deviations to obtain the cumulative deviation of each window;

[0019] The low-frequency oscillation feature judgment submodule is used to compare the cumulative deviation of each window with the threshold set in advance. If the cumulative deviation exceeds the threshold, it means that the window has a low-frequency oscillation feature. Otherwise, it is considered that the window does not have a low-frequency oscillation feature.

[0020] In one embodiment, the deviation includes an upward deviation and a downward deviation;

[0021] The rising deviation represents the increase of the value of the active power sampling data point relative to the mean value in the window;

[0022] The downward deviation represents a decrease in the value of the active power sampling data point relative to the mean value within the window.

[0023] In one embodiment, the data set construction module includes:

[0024] The time marking submodule is used to mark the windows with low-frequency oscillation characteristics as low-frequency oscillation events and record the start and end time of the low-frequency oscillation events;

[0025] An oscillation period calculation submodule, used to calculate the oscillation period of each low-frequency oscillation event according to the start and end time of the low-frequency oscillation event;

[0026] The oscillation frequency calculation submodule is used to calculate the difference between the maximum and minimum values ​​of the active power data during each low-frequency oscillation event to obtain the oscillation amplitude; calculate the oscillation frequency by the inverse of the oscillation period, and record the duration of the low-frequency oscillation event; use the oscillation amplitude and duration as characteristic parameters of the low-frequency oscillation, and form a low-frequency oscillation characteristic data set.

[0027] In one embodiment, the low frequency oscillation early warning analysis unit includes:

[0028] An initial state configuration module, used to determine the hidden state of the low-frequency oscillation of the generator set based on the operating characteristics of the generator set, and configure the initial state probability of the hidden state;

[0029] The forward probability calculation module is used to calculate the hidden state transition probability and observation probability according to the initial state probability of the hidden state through the active power data obtained in real time, and to perform forward recursive calculation according to the transition probability and observation probability to obtain the forward probability at each time point;

[0030] The comprehensive probability calculation module is used to recursively calculate the backward probability of each hidden state at each time point using the forward-backward algorithm, and to calculate the comprehensive probability of each hidden state at each time point by combining the forward probability and the backward probability through the comprehensive probability calculation formula;

[0031] The risk level judgment module is used to judge the low-frequency oscillation risk level of the generator set in real time according to the comprehensive probability of each hidden state at each time point, and output the warning type of the low-frequency oscillation.

[0032] In one embodiment, the latent state of the low frequency oscillation of the generator set corresponds to an operating state of the generator set.

[0033] In one embodiment, the forward recursive calculation is performed based on the transition probability and the observation probability to obtain the forward probability at each time point as follows:

[0034] ;

[0035] In the formula, g t (j) represents the forward probability of hidden state j at time point t, g t-1 represents the forward probability of hidden state i at time point t-1, d ij represents the transition probability from hidden state i to hidden state j, f j (O t ) indicates that the observed data O is observed in the hidden state j t The observation probability of , N represents the number of types of hidden states.

[0036] In one embodiment, the comprehensive probability calculation formula is:

[0037] ;

[0038] In the formula, Q t (i) represents the comprehensive probability of hidden state i at time point t, g t (i) represents the forward probability of hidden state j at time point t, m t (i) represents the backward probability of hidden state i at time point t, g t (j) represents the forward probability of hidden state j at time point t, m t (j) represents the backward probability of hidden state j at time point t, and N represents the number of types of hidden states.

[0039] In one embodiment, the risk level determination module includes:

[0040] The hidden state selection submodule is used to select the hidden state with the largest comprehensive probability based on the comprehensive probability of each hidden state at each time point;

[0041] A risk level determination submodule is used to determine the risk level of the current low frequency oscillation of the generator set according to the selected hidden state and using a pre-established risk level standard set;

[0042] The warning type output submodule is used to output the warning type of low-frequency oscillation based on the current risk level of low-frequency oscillation of the generator set.

[0043] According to a second aspect of an embodiment of the present invention, a method for early warning of low-frequency oscillation of active power of a generator set is provided.

[0044] In one embodiment, the generator set active power low-frequency oscillation early warning method comprises:

[0045] The active power sampling data of the generator set is collected and preprocessed. The preprocessed active power sampling data is labeled and classified according to the waveform characteristics using the sliding bilateral cumulative sum algorithm to obtain a low-frequency oscillation feature data set.

[0046] Based on the low-frequency oscillation feature data set and combined with the different working states of the generator set, a hidden Markov model is constructed and trained through transition probability analysis;

[0047] The active power data of the generator set is obtained in real time, and the early warning of low-frequency oscillation of the generator set is calculated based on the trained hidden Markov model and the forward-backward algorithm.

[0048] According to a third aspect of an embodiment of the present invention, a computer device is provided.

[0049] In some embodiments, the computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the steps of the above method when executing the computer program.

[0050] According to a fourth aspect of embodiments of the present invention, a computer-readable storage medium is provided.

[0051] In one embodiment, the computer-readable storage medium stores a computer program, and the computer program implements the steps of the above method when executed by a processor.

[0052] The technical solution provided by the embodiment of the present invention may have the following beneficial effects:

[0053] By using a sliding bilateral cumulative sum algorithm on active power data, the present invention can accurately mark whether each time window has low-frequency oscillation characteristics, can effectively capture the signal of low-frequency oscillation, and distinguish it from other noise or normal fluctuations. By constructing and training a hidden Markov model, it can analyze the transition probability between hidden states according to the low-frequency oscillation feature data set and the different working states of the generator set, and can comprehensively capture the behavior mode and transition law of the generator set under different operating states, so that the early warning system has stronger dynamic adaptability. Through real-time monitoring and dynamic analysis, the system can predict and evaluate the low-frequency oscillation of the generator set with high precision, thereby providing efficient, safe and reliable operation support for the generator set.

[0054] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0056] Figure 1 It is a principle block diagram of a generator set active power low-frequency oscillation early warning system according to an exemplary embodiment;

[0057] Figure 2 is a flow chart of a method for early warning of low-frequency oscillation of active power of a generator set according to an exemplary embodiment;

[0058] Figure 3 It is a schematic diagram showing the structure of a computer device according to an exemplary embodiment. DETAILED DESCRIPTION

[0059] The following description and accompanying drawings fully illustrate the specific embodiments of this article so that those skilled in the art can practice them. Parts and features of some embodiments may be included in or replace parts and features of other embodiments. The scope of the embodiments of this article includes the entire scope of the claims, as well as all available equivalents of the claims. Herein, the terms "first", "second", etc. are only used to distinguish one element from another, without requiring or implying any actual relationship or order between these elements. In fact, the first element can also be called the second element, and vice versa. Moreover, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that the structure, device or equipment including a series of elements includes not only those elements, but also other elements that are not explicitly listed, or also include elements inherent to such structure, device or equipment. In the absence of more restrictions, the elements defined by the sentence "including one..." do not exclude the existence of other identical elements in the structure, device or equipment including the elements. Each embodiment is described in a progressive manner herein, and each embodiment focuses on the differences from other embodiments, and the same and similar parts between the embodiments can be referred to each other.

[0060] The terms "longitudinal", "lateral", "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc. in this document indicate the orientation or position relationship based on the orientation or position relationship shown in the drawings, and are only for the convenience of describing this document and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation, and therefore cannot be understood as a limitation on the present invention. In the description of this document, unless otherwise specified and limited, the terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a mechanical connection or an electrical connection, it can also be the internal communication of two elements, it can be a direct connection, or it can be an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0061] As used herein, the term "plurality" means two or more than two, unless otherwise specified.

[0062] In this document, the character " / " indicates that the preceding and following objects are in an "or" relationship. For example, A / B means: A or B.

[0063] In this article, the term "and / or" is a description of the association relationship between objects, indicating that three relationships may exist. For example, A and / or B means: A or B, or, A and B.

[0064] It should be understood that, although the various steps in the flow chart are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear description in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a portion of the steps in the figure may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these sub-steps or stages is not necessarily to be carried out in sequence, but can be executed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.

[0065] Each module in the device or system of the present application can be implemented in whole or in part by software, hardware, or a combination thereof. The above modules can be embedded in or independent of a processor in a computer device in the form of hardware, or can be stored in a memory in a computer device in the form of software, so that the processor can call and execute operations corresponding to the above modules.

[0066] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0067] Figure 1 An embodiment of the generator set active power low-frequency oscillation early warning system of the present invention is shown.

[0068] In this optional embodiment, the generator set active power low-frequency oscillation early warning system includes:

[0069] The sampling data processing and analysis unit 101 is used to collect and pre-process the active power sampling data of the generator set, and use the sliding bilateral cumulative sum algorithm to mark and classify the pre-processed active power sampling data according to the waveform characteristics to obtain a low-frequency oscillation feature data set;

[0070] In this optional embodiment, the sampling data processing and analysis unit includes:

[0071] A data preprocessing module is used to preprocess the active power sampling data of the generator set by using filtering technology and normalization method;

[0072] It should be noted that filtering technology is mainly used to remove noise and unnecessary high-frequency components in the data and retain low-frequency useful signals. In the active power data of the generator set, there may be noise caused by measurement equipment, environmental factors or power grid fluctuations.

[0073] Normalization methods are mainly used to scale data to a specific range (usually 0 to 1 or -1 to 1) to eliminate the dimension and scale differences of the data. In the active power data of generator sets, there may be large numerical differences in data from different time periods or different units.

[0074] The low-frequency oscillation feature judgment module is used to input the pre-processed active power sampling data into the sliding bilateral cumulative sum algorithm, calculate the cumulative deviation of each window according to the preset sliding window length, and judge whether there is a low-frequency oscillation feature in each window according to the cumulative deviation;

[0075] In this optional embodiment, the low-frequency oscillation characteristic judgment module includes:

[0076] The mean value calculation submodule is used to initialize the window length of the active power sampling data and calculate the mean value of the active power sampling data points in each window;

[0077] It should be noted that the window length is the time span of a continuous data point used for analysis. Choosing the right window length is crucial for accurately identifying low-frequency oscillations. When initializing the window length, the following factors need to be considered:

[0078] Generator set operating characteristics: Different generator sets have different operating characteristics and response speeds, so different window lengths may be required.

[0079] Frequency range of the LFO: The frequency range of the LFO will determine the minimum window length. In order to include at least one complete oscillation period, the window length should be greater than or equal to the oscillation period.

[0080] Data sampling rate: The data sampling rate determines the time interval between each data point, which in turn affects the number of data points in the window.

[0081] The cumulative deviation calculation submodule is used to calculate the deviation based on the active power sampling data points in each window and its mean value, and accumulate the absolute values ​​of all deviations to obtain the cumulative deviation of each window;

[0082] In this optional embodiment, the deviation includes an upward deviation and a downward deviation;

[0083] The rising deviation represents the increase of the value of the active power sampling data point relative to the mean value in the window;

[0084] The downward deviation represents a decrease in the value of the active power sampling data point relative to the mean value within the window.

[0085] It should be noted that the cumulative deviation, as an important feature of low-frequency oscillation detection, can reflect the degree of fluctuation of the active power data of the generator set in each window. A larger cumulative deviation may mean the presence of low-frequency oscillation, while a smaller cumulative deviation may indicate that the data is relatively stable.

[0086] The low-frequency oscillation feature judgment submodule is used to compare the cumulative deviation of each window with the threshold set in advance. If the cumulative deviation exceeds the threshold, it means that the window has a low-frequency oscillation feature. Otherwise, it is considered that the window does not have a low-frequency oscillation feature.

[0087] The data set construction module is used to perform waveform feature annotation according to the judgment result, and based on the waveform feature annotation result, extract the characteristic parameters of the low-frequency oscillation to form a low-frequency oscillation characteristic data set.

[0088] In this optional embodiment, the data set construction module includes:

[0089] The time marking submodule is used to mark the windows with low-frequency oscillation characteristics as low-frequency oscillation events and record the start and end time of the low-frequency oscillation events;

[0090] It should be noted that for windows that are identified as having low-frequency oscillation features, they will be marked as low-frequency oscillation events, and the start and end times of the events will be recorded. This can be achieved by recording the start index and end index of the window, or if the data is timestamp indexed, then the corresponding timestamp is recorded.

[0091] An oscillation period calculation submodule, used to calculate the oscillation period of each low-frequency oscillation event according to the start and end time of the low-frequency oscillation event;

[0092] The oscillation frequency calculation submodule is used to calculate the difference between the maximum and minimum values ​​of the active power data during each low-frequency oscillation event to obtain the oscillation amplitude; calculate the oscillation frequency by the inverse of the oscillation period, and record the duration of the low-frequency oscillation event; use the oscillation amplitude and duration as characteristic parameters of the low-frequency oscillation, and form a low-frequency oscillation characteristic data set.

[0093] A hidden Markov model building unit 103 is used to build and train a hidden Markov model through transition probability analysis according to the low-frequency oscillation feature data set and in combination with different working states of the generator set;

[0094] It should be noted that relevant features (such as oscillation amplitude, oscillation period, frequency, etc.) are extracted from the low-frequency oscillation feature data set, and combined with the working status of the generator set (such as start, run, stop, etc.) to form a complete training data set. The working status of the generator set will affect its low-frequency oscillation characteristics, so different working statuses need to be marked and classified in the data.

[0095] Building and training a hidden Markov model involves the following steps:

[0096] Initial state probability distribution: Determine the probability that the generator set is in each state at the initial moment.

[0097] State transition probability matrix: Based on historical data or expert experience, the probability of a generator set transferring from one state to another is estimated to form a state transition probability matrix.

[0098] Observation probability matrix: For each state, determine the probability of observing a specific characteristic parameter (or observation value) in that state to form an observation probability matrix.

[0099] Data preparation: Divide the low-frequency oscillation feature dataset into a training set and a test set. The training set is used for model training, and the test set is used for model verification.

[0100] Model training: Using the training set data, the expectation-maximization (EM) algorithm or similar algorithms is used to iteratively optimize the model parameters (including the initial state probability distribution, state transition probability matrix, and observation probability matrix) so that the model can best explain the observed data.

[0101] Model verification: Use the test set data to evaluate the performance of the model, such as by calculating indicators such as accuracy, recall, and F1 score to measure the model's ability to identify low-frequency oscillation events.

[0102] The low frequency oscillation warning analysis unit 105 is used to obtain the active power data of the generator set in real time, and calculate the warning of the low frequency oscillation of the generator set according to the trained hidden Markov model and using the forward-backward algorithm.

[0103] In this optional embodiment, the low-frequency oscillation early warning analysis unit includes:

[0104] An initial state configuration module, used to determine the hidden state of the low-frequency oscillation of the generator set based on the operating characteristics of the generator set, and configure the initial state probability of the hidden state;

[0105] In this optional embodiment, the latent state of the low-frequency oscillation of the generator set corresponds to the operating state of the generator set.

[0106] Hidden state represents the internal state of the system that cannot be directly observed in the hidden Markov model. For the hidden state of the low-frequency oscillation of the generator set, it can be defined as follows based on the operating characteristics of the generator set:

[0107] Normal operating state: The generator set is in a stable operating state and shows no signs of low-frequency oscillation.

[0108] Initial state of low-frequency oscillation: The generator set begins to show signs of low-frequency oscillation, but the oscillation amplitude and frequency are still gradually increasing.

[0109] Continuous low-frequency oscillation state: The generator set is in a low-frequency oscillation state, and the oscillation amplitude and frequency are relatively stable.

[0110] End state of low-frequency oscillation: The amplitude and frequency of the low-frequency oscillation of the generator set begin to decrease, and the oscillation gradually subsides.

[0111] The forward probability calculation module is used to calculate the hidden state transition probability and observation probability according to the initial state probability of the hidden state through the active power data obtained in real time, and to perform forward recursive calculation according to the transition probability and observation probability to obtain the forward probability at each time point;

[0112] It should be noted that the initial hidden state transition probabilities have been configured for the hidden Markov model based on historical data or expert experience. In real-time analysis, these transition probabilities need to be dynamically adjusted according to actual conditions, but they are usually learned based on historical data and have been determined during the model training phase.

[0113] For each hidden state, it is necessary to calculate the degree of match between the features observed at the current time point (i.e., the features of the active power data) and the state, which is usually achieved through the observation probability matrix. In real-time analysis, the corresponding probability in the observation probability matrix is ​​updated or queried based on the current active power data features.

[0114] In this optional embodiment, the forward recursive calculation is performed based on the transition probability and the observation probability to obtain the forward probability at each time point as follows:

[0115] ;

[0116] In the formula, g t (j) represents the forward probability of hidden state j at time point t, g t-1 represents the forward probability of hidden state i at time point t-1, d ij represents the transition probability from hidden state i to hidden state j, f j (O t ) indicates that the observed data O is observed in the hidden state j tThe observation probability of , N represents the number of types of hidden states.

[0117] The comprehensive probability calculation module is used to recursively calculate the backward probability of each hidden state at each time point using the forward-backward algorithm, and to calculate the comprehensive probability of each hidden state at each time point by combining the forward probability and the backward probability through the comprehensive probability calculation formula;

[0118] It should be noted that the calculation formula for the backward probability of each hidden state is:

[0119] ;

[0120] In this optional embodiment, the comprehensive probability calculation formula is:

[0121] ;

[0122] In the formula, Q t (i) represents the comprehensive probability of hidden state i at time point t, g t (i) represents the forward probability of hidden state j at time point t, m t (i) represents the backward probability of hidden state i at time point t, g t (j) represents the forward probability of hidden state j at time point t, m t (j) represents the backward probability of hidden state j at time point t, N represents the number of types of hidden states, and f j (O t+1 ) indicates that the observed data O is observed in the hidden state j t+1 The observation probability, m t+1 (j) represents the backward probability of hidden state j at time point t.

[0123] The risk level judgment module is used to judge the low-frequency oscillation risk level of the generator set in real time according to the comprehensive probability of each hidden state at each time point, and output the warning type of the low-frequency oscillation.

[0124] In this optional embodiment, the risk level determination module includes:

[0125] The hidden state selection submodule is used to select the hidden state with the largest comprehensive probability based on the comprehensive probability of each hidden state at each time point;

[0126] A risk level determination submodule is used to determine the risk level of the current low frequency oscillation of the generator set according to the selected hidden state and using a pre-established risk level standard set;

[0127] The warning type output submodule is used to output the warning type of low-frequency oscillation based on the current risk level of low-frequency oscillation of the generator set.

[0128] Figure 2An embodiment of the method for early warning of low-frequency oscillation of active power of a generator set of the present invention is shown.

[0129] In this optional embodiment, the generator set active power low-frequency oscillation early warning method includes:

[0130] Step 201, collecting and preprocessing the active power sampling data of the generator set, using the sliding bilateral cumulative sum algorithm, labeling and classifying the preprocessed active power sampling data according to waveform characteristics, and obtaining a low-frequency oscillation feature data set;

[0131] Step 203, constructing and training a hidden Markov model through transition probability analysis based on the low-frequency oscillation feature data set and in combination with different working states of the generator set;

[0132] Step 205 , obtaining the active power data of the generator set in real time, and calculating the early warning of the low frequency oscillation of the generator set according to the trained hidden Markov model and using the forward-backward algorithm.

[0133] In one embodiment, a computer device is provided. The computer device may be a server, and its internal structure diagram may be as follows: Figure 3 As shown. The computer device includes a processor, a memory and a network interface connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store static information and dynamic information data. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the steps in the above method embodiment are implemented.

[0134] Those skilled in the art will understand that Figure 3 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0135] In addition, the present invention also provides a computer device, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiment when executing the computer program.

[0136] In addition, the present invention further provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps in the above method embodiment are implemented.

[0137] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided by the present invention can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory or optical memory, etc. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0138] The present invention is not limited to the structures which have been described above and shown in the drawings, and various modifications and changes may be made without departing from the scope thereof. The scope of the present invention is limited only by the appended claims.

Claims

1. A generator set active power low-frequency oscillation early warning system, characterized in that: include: The sampling data processing and analysis unit is used to collect the active power sampling data of the generator set and perform preprocessing, and use the sliding bilateral cumulative sum algorithm to mark and classify the preprocessed active power sampling data according to the waveform characteristics to obtain a low-frequency oscillation feature data set; A hidden Markov model building unit is used to build and train a hidden Markov model through transition probability analysis based on a low-frequency oscillation feature data set and in combination with different working states of the generator set; The low-frequency oscillation early warning analysis unit is used to obtain the active power data of the generator set in real time, and calculate the early warning of the low-frequency oscillation of the generator set based on the trained hidden Markov model and the forward-backward algorithm.

2. The generator set active power low-frequency oscillation early warning system according to claim 1, characterized in that: The sampling data processing and analysis unit comprises: A data preprocessing module is used to preprocess the active power sampling data of the generator set by using filtering technology and normalization method; The low-frequency oscillation feature judgment module is used to input the pre-processed active power sampling data into the sliding bilateral cumulative sum algorithm, calculate the cumulative deviation of each window according to the preset sliding window length, and judge whether there is a low-frequency oscillation feature in each window according to the cumulative deviation; The data set construction module is used to perform waveform feature annotation according to the judgment result, and based on the waveform feature annotation result, extract the characteristic parameters of the low-frequency oscillation to form a low-frequency oscillation characteristic data set.

3. The generator set active power low-frequency oscillation early warning system according to claim 2, characterized in that: The low-frequency oscillation characteristic judgment module includes: The mean value calculation submodule is used to initialize the window length of the active power sampling data and calculate the mean value of the active power sampling data points in each window; The cumulative deviation calculation submodule is used to calculate the deviation based on the active power sampling data points in each window and its mean value, and accumulate the absolute values ​​of all deviations to obtain the cumulative deviation of each window; The low-frequency oscillation feature judgment submodule is used to compare the cumulative deviation of each window with the threshold set in advance. If the cumulative deviation exceeds the threshold, it means that the window has a low-frequency oscillation feature. Otherwise, it is considered that the window does not have a low-frequency oscillation feature.

4. The generator set active power low-frequency oscillation early warning system according to claim 3, characterized in that: The deviation includes an upward deviation and a downward deviation; The rising deviation represents the increase of the value of the active power sampling data point relative to the mean value in the window; The downward deviation represents a decrease in the value of the active power sampling data point relative to the mean value within the window.

5. The generator set active power low-frequency oscillation early warning system according to claim 3, characterized in that: The data set construction module includes: The time marking submodule is used to mark the windows with low-frequency oscillation characteristics as low-frequency oscillation events and record the start and end time of the low-frequency oscillation events; An oscillation period calculation submodule, used to calculate the oscillation period of each low-frequency oscillation event according to the start and end time of the low-frequency oscillation event; The oscillation frequency calculation submodule is used to calculate the difference between the maximum and minimum values ​​of the active power data during each low-frequency oscillation event to obtain the oscillation amplitude; calculate the oscillation frequency by the inverse of the oscillation period, and record the duration of the low-frequency oscillation event; use the oscillation amplitude and duration as characteristic parameters of the low-frequency oscillation, and form a low-frequency oscillation characteristic data set.

6. The generator set active power low-frequency oscillation early warning system according to claim 1, characterized in that: The low frequency oscillation early warning analysis unit comprises: An initial state configuration module, used to determine the hidden state of the low-frequency oscillation of the generator set based on the operating characteristics of the generator set, and configure the initial state probability of the hidden state; The forward probability calculation module is used to calculate the hidden state transition probability and observation probability according to the initial state probability of the hidden state through the real-time acquired active power data, and to perform forward recursive calculation according to the transition probability and observation probability to obtain the forward probability at each time point; The comprehensive probability calculation module is used to recursively calculate the backward probability of each hidden state at each time point using the forward-backward algorithm, and to calculate the comprehensive probability of each hidden state at each time point by combining the forward probability and the backward probability through the comprehensive probability calculation formula; The risk level judgment module is used to judge the low-frequency oscillation risk level of the generator set in real time according to the comprehensive probability of each hidden state at each time point, and output the warning type of the low-frequency oscillation.

7. The generator set active power low-frequency oscillation early warning system according to claim 6, characterized in that: The latent state of the low-frequency oscillation of the generator set corresponds to the operating state of the generator set.

8. The generator set active power low-frequency oscillation early warning system according to claim 7, characterized in that: The forward recursive calculation is performed based on the transition probability and the observation probability to obtain the forward probability calculation formula for each time point: ； In the formula, g t (j) represents the forward probability of hidden state j at time point t, g t-1 represents the forward probability of hidden state i at time point t-1, d ij represents the transition probability from hidden state i to hidden state j, f j (O t ) indicates that the observed data O is observed in the hidden state j t The observation probability of , N represents the number of types of hidden states.

9. The generator set active power low-frequency oscillation early warning system according to claim 7, characterized in that: The comprehensive probability calculation formula is: ； In the formula, Q t (i) represents the comprehensive probability of hidden state i at time point t, g t (i) represents the forward probability of hidden state j at time point t, m t (i) represents the backward probability of hidden state i at time point t, g t (j) represents the forward probability of hidden state j at time point t, m t (j) represents the backward probability of hidden state j at time point t, and N represents the number of types of hidden states.

10. The generator set active power low-frequency oscillation early warning system according to claim 9, characterized in that: The risk level judgment module includes: The hidden state selection submodule is used to select the hidden state with the largest comprehensive probability based on the comprehensive probability of each hidden state at each time point; A risk level determination submodule is used to determine the risk level of the current low frequency oscillation of the generator set according to the selected hidden state and using a pre-established risk level standard set; The warning type output submodule is used to output the warning type of low-frequency oscillation based on the current risk level of low-frequency oscillation of the generator set.

Images