Reactive power optimization control method for global wind power plant
By conducting monitoring log correlation analysis and reactive power demand analysis on K equipment in the whole-domain wind farm, combining the dynamic adjustment range and output tolerance range, reactive power optimization control function is built, which solves the problem that traditional methods are difficult to cope with dynamically changing reactive power requirements, and achieves efficient reactive power distribution and grid stability improvement.
Patent Information
- Application Number
- CN202510056479.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-14
- Publication Date
- 2025-05-13
Smart Images
Figure CN119995063A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data processing, and in particular to a reactive power optimization control method for a global wind farm. Background Art
[0002] With the continuous expansion of wind farms and the increase in the complexity of power systems, the optimization and control of reactive power in wind farms across the region has become an important issue that needs to be urgently addressed in the operation of power systems. The rational allocation of reactive power is crucial to improving the power quality and stability of the power grid and the operating efficiency of wind farms. However, in the prior art, the reactive power demand of each device in a wind farm presents a dynamic change characteristic. Traditional reactive power optimization methods often rely on static parameter settings or single optimization of local equipment, which makes it difficult to cope with the complex correlations between devices across the region and the real-time changing reactive power demand, resulting in low overall operating efficiency of wind farms, and may even cause grid fluctuations or equipment failures. Summary of the invention
[0003] The present application provides a reactive power optimization control method for a whole-region wind farm, which solves the technical problem in the prior art that the reactive power distribution is not accurate enough, resulting in low wind farm operation efficiency and grid stability.
[0004] In view of the above problems, the present application provides a reactive power optimization control method for a global wind farm.
[0005] The present application provides a reactive power optimization control method for a global wind farm, the method comprising:
[0006] Traverse the K devices of the interactive target global wind farm, and obtain the K device window operation log sequences in the monitoring logs within the preset monitoring window, where K is an integer greater than or equal to 1; perform associated iterative identification on the K device window operation log sequences in chronological order from front to back according to the preset monitoring indicator group, and determine the K iterative device window operation status indicator groups; call the pre-built device reactive power demand analyzer to analyze the K iterative device window operation status indicator groups to obtain K device reactive power demand information; construct a reactive power demand distribution map of the target global wind farm based on the K device reactive power demand information; obtain the dynamic adjustment range of the SVG device of the target global wind farm, and use it as the first optimization constraint; obtain the K reactive power output tolerance ranges of the K devices, and generate the second optimization constraint; construct a reactive power optimization control function, combine the first optimization constraint and the second optimization constraint, perform reactive power optimization control analysis on the reactive power demand distribution map, and when the reactive power optimization control function reaches the minimum value, obtain the target reactive power optimization control scheme.
[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:
[0008] First, traverse the K devices of the interactive target global wind farm, and obtain the K device window operation log sequences in the monitoring logs within the preset monitoring window, where K is an integer greater than or equal to 1. Then, according to the preset monitoring indicator group, the K device window operation log sequences are associated and iteratively identified in the order from front to back in time, and the K iterative device window operation status indicator groups are determined. Further, call the pre-built device reactive demand analyzer to analyze the K iterative device window operation status indicator groups to obtain the reactive demand information of the K devices. Then, based on the reactive demand information of the K devices, construct the reactive power demand distribution map of the target global wind farm. At the same time, obtain the dynamic adjustment range of the SVG equipment of the target global wind farm and use it as the first optimization constraint; obtain the K reactive power output tolerance ranges of the K devices and generate the second optimization constraint. Finally, construct the reactive power optimization control function, combine the first optimization constraint and the second optimization constraint, perform reactive power optimization control analysis on the reactive power demand distribution map, and when the reactive power optimization control function reaches the minimum value, obtain the target reactive power optimization control scheme. The technical problem that the reactive power distribution in the prior art is not accurate enough, resulting in low wind farm operating efficiency and grid stability, is solved, and the technical effect of improving the wind farm operating efficiency and grid stability is achieved. BRIEF DESCRIPTION OF THE DRAWINGS
[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.
[0010] Figure 1 A schematic flow chart of a reactive power optimization control method for a global wind farm provided in an embodiment of the present application;
[0011] Figure 2 A schematic diagram of a flow chart for determining K iterative device window operating status indicator groups in a reactive power optimization control method for a global wind farm provided in an embodiment of the present application. DETAILED DESCRIPTION
[0012] The present application solves the technical problem in the prior art that the reactive power distribution is not accurate enough, resulting in low wind farm operating efficiency and grid stability, by providing a reactive power optimization control method for the entire wind farm.
[0013] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.
[0014] It should be noted that the terms "including" and "having" are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or are inherent to these processes, methods, products or devices.
[0015] Examples, such as Figure 1 As shown, the embodiment of the present application provides a reactive power optimization control method for a global wind farm, wherein the method includes:
[0016] Traverse the K devices of the interactive target global wind farm, and obtain the K device window operation log sequences in the monitoring logs within the preset monitoring window, where K is an integer greater than or equal to 1.
[0017] A global wind farm is usually composed of multiple wind turbines, transformers, and reactive compensation devices such as SVG devices for adjusting reactive power); by traversing K devices in the global wind farm, an interactive connection between devices is established, where K is an integer greater than or equal to 1; for each device, within a preset monitoring time window, the monitoring log data of the device is extracted, and the monitoring log includes key parameters such as the reactive power, voltage, current, power factor, and operating status of the device; the monitoring logs of each device within the time window are arranged in chronological order, and a corresponding device window operation log sequence is generated, and K device window operation log sequences are obtained.
[0018] According to the preset monitoring indicator group, the K device window operation log sequences are associated and iterated and identified in a time-ordered order to determine K iterative device window operation status indicator groups.
[0019] Define a preset monitoring indicator group, which includes but is not limited to indicators related to the equipment operating status, such as reactive power, voltage, current, power factor, load fluctuation, and equipment status identification; for the window operation log sequence of K devices, iteratively process the log sequence of each device in chronological order from front to back, and extract the data corresponding to the monitoring indicator group; in the iterative process, perform correlation analysis on the indicator data in the time series, including but not limited to calculating the change trend between related indicators, detecting abnormal points in the equipment operating status, etc.; through iterative analysis, generate an operating status indicator group for each device in the monitoring window, and the operating status indicator group contains all data in the preset monitoring indicator group and associated dynamic status information; output the iterative operating status indicator group of K devices to provide input data for subsequent reactive power demand analysis.
[0020] Furthermore, the preset monitoring indicator group includes active power generation, reactive power generation, grid voltage, grid frequency and reactive compensation equipment operation indicators.
[0021] The preset monitoring indicator group includes active power generation, reactive power generation, grid voltage, grid frequency and reactive compensation equipment operation indicators. Among them, active power generation reflects the actual output power of the equipment, reactive power generation indicates the reactive power currently provided by the equipment, grid voltage records the real-time voltage value of the grid to which the equipment is connected, grid frequency refers to the frequency stability of the grid operation, and reactive compensation equipment operation indicators refer to the adjustment range, switching status and other information of the SVG equipment.
[0022] Furthermore, if Figure 2 As shown, according to the preset monitoring indicator group, the K device window operation log sequences are associated and iteratively identified in a time-ordered order, and K iterative device window operation status indicator groups are determined, including:
[0023] Taking the preset monitoring indicator group as the index, the device window operation logs in the K device window operation log sequences are respectively subjected to indicator extraction and sentence extraction of the indicators in the logs, so as to obtain K device window monitoring indicator group sequences and K log key sentence set sequences, wherein each log key sentence set corresponds to a set of key sentences of a device window operation log; the K device window monitoring indicator group sequences are subjected to associated iterative identification in a time-ordered order, so as to obtain K initial iterative device window operation status indicator groups; indicator trend analysis is performed based on the K log key sentence set sequences, so as to obtain K indicator trend sets; the indicators in the K initial iterative device window operation status indicator groups are correspondingly identified using the K indicator trend sets, so as to obtain the K iterative device window operation status indicator groups.
[0024] Specifically, with the preset detection indicator group (including active power generation, reactive power generation, grid voltage, grid frequency, and reactive compensation equipment operation indicators) as the index, for each device window operation log sequence, read the log data one by one; extract specific data that meets the monitoring indicator group from each log, for example, extract reactive power: 30kVar, voltage: 220V and other specific indicator values from a log; extract the original log sentences containing monitoring indicators, and construct a key sentence set for each log. For example, the original text of a log "The reactive power of device A at 10:00 is 30kVar, and the grid voltage is 220V" will be extracted as a key sentence to form K device window monitoring indicator group sequences and K log key sentence set sequences. Each log key sentence set consists of key sentences containing monitoring indicators in the corresponding device window operation log; associate and iterate the K device window monitoring indicator group sequences in chronological order from front to back, such as analyzing reactive power and grid voltage. The change relationship between the indicators is detected, the fluctuation of the grid frequency is detected, and the switching moment of the equipment load from high load to low load is marked, the correlation and change trend between the indicators are extracted, and K initial iterative equipment window operation status indicator groups are obtained; based on the K log key sentence set sequence, the time series trend analysis of the equipment monitoring indicators is further performed, such as calculating the change rate of reactive power, analyzing the periodic fluctuation or sudden rise and fall trend of the grid voltage, and forming K indicator trend sets; K indicator trend sets are used to identify the trends of each indicator in the K initial iterative equipment window operation status indicator groups, such as marking the reactive power fluctuation range as "stable" or "fluctuation", marking the voltage deviation as "normal" or "overlimit", and marking the frequency stability as "stable" or "abnormal", so as to improve the equipment operation status information, and finally determine the K iterative equipment window operation status indicator groups, which comprehensively reflect the equipment operation status and its dynamic changes, and provide comprehensive and accurate basic data for subsequent reactive power optimization analysis.
[0025] Furthermore, the K device window monitoring indicator group sequences are associated and iterated in order from the beginning to the end of time to obtain K initial iterative device window operation status indicator groups, including:
[0026] Extract K first device window monitoring indicator groups and K second device window monitoring indicator groups from the K device window monitoring indicator group sequence, perform associated iterative identification on the K first device window monitoring indicator groups and the K second device window monitoring indicator groups respectively, and obtain K first associated iterative device window monitoring indicator groups; extract K third device window monitoring indicator groups from the K device window monitoring indicator group sequence, perform associated iterative identification on the K third device window monitoring indicator groups and the K first associated iterative device window monitoring indicator groups, and obtain K second associated iterative device window monitoring indicator groups; and perform associated iterative identification in a similar manner, and use the K associated iterative device window monitoring indicator groups obtained from the last associated iterative identification as the K iterative device window operating status indicator groups.
[0027] Specifically, extract the K first device window monitoring indicator groups and K second device window monitoring indicator groups from the K device window monitoring indicator group sequence, and perform correlation iterative identification on the two groups of device window monitoring indicator groups, for example, analyze the relationship between the change of active power and reactive power, the fluctuation of grid voltage, the operating status of reactive compensation equipment, etc., and establish the correlation between the two groups of data; obtain K first correlation iterative device window monitoring indicator groups through iterative calculation, which preliminarily reflects the operating characteristics and status of the equipment. Extract the K third device window monitoring indicator groups from the K device window monitoring indicator group sequence, and perform correlation iterative identification with the K first correlation iterative device window monitoring indicator groups obtained previously, for example, analyze whether the reactive power fluctuation trend of the third group of equipment is affected by the operating status of the first two groups of equipment, or check whether the grid voltage has changed significantly due to the operating status of multiple devices. Through this correlation iterative processing, generate K second correlation iterative device window monitoring indicator groups to further improve the equipment operating status indicators. According to the above process, the subsequent device window monitoring indicator groups are sequentially associated and iteratively identified. In each iteration, the K associated iterative device window monitoring indicator groups obtained in the previous round are used as input to perform association analysis with the next group of device window monitoring indicator groups. After the last round of associated iterative identification is completed, the obtained K associated iterative device window monitoring indicator groups are used as the final K iterative device window operating status indicator groups. Each indicator group comprehensively reflects the operating characteristics of the equipment, including dynamic indicators (such as reactive power change trend, grid voltage fluctuation range), state identification (such as "stable", "fluctuation", "abnormal"), and association relationships (such as the impact intensity and trend between multiple devices).
[0028] Further, including:
[0029] Perform the same indicator mapping similarity analysis on the K first device window monitoring indicator groups and the K second device window monitoring indicator groups to obtain K first mapping similarity sets; perform standardization on the K first mapping similarity sets, and perform convolution calculation on the processing results with the K second device window monitoring indicator groups to obtain the K first associated iterative device window monitoring indicator groups.
[0030] Preferably, for the same monitoring indicator (such as reactive power, grid voltage, etc.) in the K first device window monitoring indicator groups and the K second device window monitoring indicator groups, the mapping similarities between them are calculated in turn to obtain K first mapping similarity sets, each first mapping similarity set corresponds to the similarity values between all monitoring indicators of a device and other devices. Specifically, the similarity between the two indicator value sequences can be calculated by methods such as Euclidean distance, cosine similarity or dynamic time warping (DTW). For example, the similarity between the reactive power sequence of the first device and the reactive power sequence of the second device is calculated to reflect the correlation between the operating states of the two devices. The similarity values of the K first mapping similarity sets are standardized to ensure consistent data dimensions and facilitate subsequent calculations. Specifically, minimum-maximum standardization or Z-score standardization can be used. The standardized first mapping similarity set is convolved with the indicator sequence of the K second device window monitoring indicator groups, that is, the weighted sum of the first mapping similarity set and each monitoring indicator sequence in the second device window monitoring indicator group is calculated, and then K first associated iterative device window monitoring indicator groups are generated to reflect the state correlation between the first device window monitoring indicator group and the second device window monitoring indicator group.
[0031] Furthermore, an indicator trend analysis is performed based on the K log key sentence set sequences to obtain K indicator trend sets, including:
[0032] Extracting log key sentence sequences corresponding to the same type of monitoring indicators from the K log key sentence set sequences to obtain K log key sentence sequence sets, wherein each log key sentence sequence corresponds to a monitoring indicator of a device and is a sequence formed by trend description key sentences arranged from front to back in time in the operation log of a preset monitoring window; performing indicator trend analysis on the K log key sentence sequence sets respectively using a semantic identifier to obtain K indicator trend sets, wherein each indicator trend corresponds to a monitoring indicator.
[0033] Preferably, for K log key sentence set sequences, the log key sentences are classified according to the same type of monitoring indicators (such as reactive power, grid voltage, etc.), and the log key sentence sequence corresponding to each monitoring indicator is extracted; from the K log key sentence sets, the key sentences containing a certain monitoring indicator are screened, for example, key sentences related to "reactive power" are extracted, such as "the reactive power of device A at 10:00 is 30kVar"; the key sentences of the same device are arranged in chronological order from front to back to form a sequence of trend description key sentences, and then K log key sentence sequence sets are obtained, each log key sentence sequence corresponds to a certain monitoring indicator of a device in a preset A sequence of trend-describing key sentences arranged in chronological order from front to back in the operation log within the monitoring window; using a semantic identifier to perform semantic analysis on each key sentence in the K log key sentence sequence sets, extract the changing trend of the monitoring indicators in the time series, and identify the trend characteristics of the indicators, including rising, falling, fluctuating, and stable, etc. For example, through semantic analysis of the key sentence sequence of the reactive power log, it is identified that the reactive power shows an "increasing trend" from 30kVar to 35kVar; combining the semantic analysis results of all monitoring indicators, K indicator trend sets are generated, where each indicator trend set corresponds to the changing law of a certain monitoring indicator of a device within a preset monitoring window.
[0034] A pre-built device reactive power demand analyzer is called to analyze the K iterative device window operation status indicator groups to obtain reactive power demand information of the K devices.
[0035] According to the key indicators such as reactive power, grid voltage, grid frequency, load fluctuation, etc. contained in the iterative device window operating status indicator group of each device, the device reactive power demand analyzer is used to comprehensively calculate and analyze them; the device reactive power demand analyzer is based on the historical data of the device operating status, real-time operating indicators and device characteristic parameters (such as maximum output capacity, reactive power adjustment range, etc.), combined with the grid operating conditions (such as the allowable range of grid voltage deviation, grid frequency fluctuation amplitude, etc.), to accurately evaluate the current reactive power demand of each device; the analysis result is output as reactive power demand information of K devices, including the reactive power value required by each device in the current operating state, the adjustment direction (such as the need to increase or decrease reactive power) and its priority, for example, "the reactive power demand of device 1 is +20kVar, and reactive power needs to be increased", "the reactive power demand of device 2 is -15kVar, and reactive power needs to be reduced".
[0036] Furthermore, the pre-built device reactive power demand analyzer is called to analyze the K iterative device window operation status indicator groups to obtain reactive power demand information of K devices, including:
[0037] Acquire multiple sample iterative device window operating status indicator groups and multiple sample device reactive power demand information as training data; use the training data to perform supervised training on a framework built based on a feedforward neural network, learn a one-to-one mapping relationship between the iterative device window operating status indicator group and the device reactive power demand information, until the training converges, and obtain the trained device reactive power demand analyzer.
[0038] Specifically, multiple sample iterative device window operating status indicator groups and corresponding sample device reactive power demand information are obtained as training data, wherein the sample data include operating status indicators such as reactive power, grid voltage, grid frequency, load fluctuation, etc. of the equipment under various operating scenarios, as well as reactive power demand output values corresponding to these indicators; then, the analysis framework constructed based on the feedforward neural network is supervised and trained using these training data, and the sample iterative device window operating status indicator group is input, and the nonlinear mapping relationship between it and the device reactive power demand information is learned through the feedforward calculation layer, the activation function layer, and the error back propagation optimization layer; during the training process, the difference between the model prediction value and the sample true value is evaluated by minimizing the loss function (such as mean square error, MSE), and the network weights and bias parameters are adjusted until the training process converges and the output error of the model is within the target range; finally, a trained device reactive power demand analyzer is obtained, which can output reactive power demand information of each device based on the input K iterative device window operating status indicator groups, including the required reactive power value, adjustment direction (increase or decrease), priority, etc.
[0039] A reactive power demand distribution diagram of the target global wind farm is constructed based on the reactive power demand information of the K devices.
[0040] The reactive power demand information of K devices is sorted out. The reactive power demand information of each device includes the reactive power demand value (a positive value indicates that reactive power needs to be increased, and a negative value indicates that reactive power needs to be reduced), the device's geographical location, the operating status label and the priority. Then, according to the geographical location of the device, the reactive power demand information of the K devices is associated with the spatial layout of the wind farm to form the spatial distribution characteristics of the device, for example, by region (such as the east area and the west area) for zoning and classification, so as to intuitively display the regional differences in demand. Then, the drawing parameters of the reactive power demand distribution map are defined, including the color gradient (red indicates an increase in reactive power, blue indicates a decrease in reactive power), the size of the circle or the height of the column (corresponding to the reactive power demand value) and the geographical location coordinates of the device. These parameters are used to draw a distribution map on a two-dimensional coordinate system or a geographic information system (GIS) platform, in which each device is drawn with a visual symbol (such as a color circle or a bar chart) corresponding to its location to indicate the direction and intensity of its reactive power demand. For example, at the device location (x 1 ,y 1) draws a red circle to indicate that the reactive power needs to be increased by +30kVar. 2 ,y 2 ) draws a blue circle to indicate that the reactive power needs to be reduced by -20kVar; finally, an overall statistical analysis is performed on the distribution map to extract the distribution characteristics of the global reactive power demand, such as the total regional demand, the demand distribution of high-priority equipment, and the equipment locations of the maximum and minimum demand points, and finally a reactive power demand distribution map of the target global wind farm is generated, providing intuitive and clear data support for subsequent reactive power optimization control.
[0041] The dynamic adjustment range of the SVG equipment of the target global wind farm is obtained and used as the first optimization constraint.
[0042] Traverse all SVG (static VAR generator) devices in the target global wind farm, and extract the dynamic adjustment range information of its reactive power from the operating parameters of the devices, including key parameters such as the minimum output capacity, maximum output capacity and adjustment speed of each SVG device. For example, the dynamic adjustment range of reactive power of a certain SVG device is -50kVar to +50kVar; then, integrate the dynamic adjustment ranges of all SVG devices to form the dynamic adjustment range of SVG devices in the target global wind farm, and use it as the first optimization constraint to constrain the dynamic adjustment boundary in the reactive power optimization allocation scheme, ensuring that the optimization result meets the actual dynamic capacity range of each SVG device, and providing equipment-level safety and rationality guarantees for the reactive power optimization control of the global wind farm.
[0043] K reactive power output tolerance ranges of the K devices are obtained to generate a second optimization constraint.
[0044] Traverse the K devices (including wind turbines, SVG devices, compensation devices, etc.) in the target global wind farm, extract the operating parameters and technical specifications of each device, and obtain its reactive power output range information, including the maximum output capacity, minimum output capacity, adjustment sensitivity and operating safety restrictions of the device. For example, the reactive power output range of a wind turbine is -30kVar to +40kVar; secondly, combine the operating status indicators of the equipment (such as voltage deviation, load fluctuation amplitude and operating temperature, etc.), dynamically adjust the output range of the equipment, and determine its actual reactive power output tolerance range. For example, if a device is limited by its operating status, its dynamic tolerance range may be reduced to -20kVar to +30kVar; then, integrate the reactive power output tolerance range of all devices to form a reactive power output capacity set of K devices in the entire wind farm, and use these output tolerance ranges as the second optimization constraint to clarify the output restriction conditions of each device in the optimization allocation plan, ensure that the reactive power optimization control results operate within the actual capacity range of each device, and provide constraint guarantees for the feasibility of the optimization plan and the safety of the equipment.
[0045] Construct a reactive power optimization control function, combine the first optimization constraint and the second optimization constraint, perform reactive power optimization control analysis on the reactive power demand distribution diagram, and when the reactive power optimization control function reaches the minimum value, obtain a target reactive power optimization control scheme.
[0046] A reactive power optimization control function is constructed to minimize the reactive power loss of the entire wind farm and improve the grid stability and equipment operation efficiency as the optimization goals. Then, the first optimization constraint is introduced, that is, the dynamic adjustment range of all SVG devices in the target wind farm is used as a constraint condition to ensure that the reactive power output of each SVG device is within its minimum and maximum adjustment range. At the same time, the second optimization constraint is introduced, and the reactive power output tolerance range of all devices is used as an optimization limit to ensure that the reactive power output value of the device is within its safe operating range. On this basis, the reactive power demand distribution map is input, the dynamic adjustment range and output tolerance range of all devices are integrated, and the objective function is solved using optimization algorithms such as linear programming or quadratic programming, so that the reactive power optimization control function reaches the minimum value under the constraint conditions. When the optimization converges, the target reactive power optimization control scheme is output, including the reactive power output value of each device, the adjustment output value of the SVG device, and the reactive power distribution result of the entire wind farm, to ensure that the reactive power demand is met while reducing the equipment operation loss, improving the operation efficiency of the wind farm and the stability of the grid, and providing optimization strategy support for subsequent operation and scheduling.
[0047] Furthermore, the reactive power optimization control function is:
[0048] Among them, L min is the output value when the reactive power optimization control function is minimized, is the reactive transmission loss of the ith device, ΔV i is the voltage deviation at the access point of the ith device, and W is the weight coefficient used to balance the loss and voltage deviation.
[0049] In the reactive power optimization control function, L min is the output value when the reactive power optimization control function is minimized, is the reactive power transmission loss of the ith device, indicating the loss of the device during reactive power transmission due to grid impedance and other reasons; ΔV iis the access point voltage deviation of the target global wind farm, indicating the deviation between the actual voltage at the device access point and the nominal voltage; W is the weight coefficient, which is used to balance the optimization target of reactive transmission loss and voltage deviation, and the specific value is set according to the operation requirements. Through the reactive power optimization control function, the impact of equipment reactive transmission loss and access point voltage deviation on the operation of the global wind farm can be considered at the same time, and a trade-off can be achieved between the two. When the objective function L min When the minimum value is reached, the optimal allocation scheme of reactive power for the entire wind farm is obtained, ensuring that the voltage stability of the access point is maintained while reducing reactive transmission losses. This function combines the first optimization constraint (dynamic adjustment range of SVG equipment) and the second optimization constraint (reactive power output tolerance range of equipment) to ensure that the optimization scheme meets the equipment capacity limit and operation safety, further improving the operation efficiency of the wind farm and the stability of the power grid.
[0050] In summary, the embodiments of the present application have at least the following technical effects:
[0051] First, traverse the K devices of the interactive target global wind farm, and obtain the K device window operation log sequences in the monitoring logs within the preset monitoring window, where K is an integer greater than or equal to 1. Then, according to the preset monitoring indicator group, the K device window operation log sequences are associated and iteratively identified in the order from front to back in time, and the K iterative device window operation status indicator groups are determined. Further, call the pre-built device reactive demand analyzer to analyze the K iterative device window operation status indicator groups to obtain the reactive demand information of the K devices. Then, based on the reactive demand information of the K devices, construct the reactive power demand distribution map of the target global wind farm. At the same time, obtain the dynamic adjustment range of the SVG equipment of the target global wind farm and use it as the first optimization constraint; obtain the K reactive power output tolerance ranges of the K devices and generate the second optimization constraint. Finally, construct the reactive power optimization control function, combine the first optimization constraint and the second optimization constraint, perform reactive power optimization control analysis on the reactive power demand distribution map, and when the reactive power optimization control function reaches the minimum value, obtain the target reactive power optimization control scheme. The technical problem that the reactive power distribution in the prior art is not accurate enough, resulting in low wind farm operating efficiency and grid stability, is solved, and the technical effect of improving the wind farm operating efficiency and grid stability is achieved.
[0052] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. The processes depicted in the accompanying drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.
[0053] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.
[0054] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.
Claims
1. A reactive power optimization control method for a global wind farm, characterized in that: The method comprises: Traverse the monitoring logs of K devices in the interactive target global wind farm and obtain the K device window operation log sequences in the preset monitoring window, where K is an integer greater than or equal to 1; According to the preset monitoring indicator group, the K device window operation log sequences are iteratively identified in chronological order, and K iterative device window operation status indicator groups are determined; Calling a pre-built device reactive power demand analyzer to analyze the K iterative device window operation status indicator groups to obtain reactive power demand information of the K devices; Constructing a reactive power demand distribution map of the target global wind farm based on the reactive power demand information of the K devices; Obtaining a dynamic adjustment range of the SVG equipment of the target global wind farm, and using it as a first optimization constraint; Obtaining K reactive power output tolerance ranges of the K devices to generate a second optimization constraint; Construct a reactive power optimization control function, combine the first optimization constraint and the second optimization constraint, perform reactive power optimization control analysis on the reactive power demand distribution diagram, and when the reactive power optimization control function reaches the minimum value, obtain a target reactive power optimization control scheme.
2. The reactive power optimization control method for the global wind farm according to claim 1, characterized in that: The preset monitoring index group includes active power generation, reactive power generation, grid voltage, grid frequency and reactive compensation equipment operation index.
3. The reactive power optimization control method for the global wind farm according to claim 1, characterized in that: According to the preset monitoring indicator group, the K device window operation log sequences are associated and iterated in a time-ordered order to determine K iterative device window operation status indicator groups, including: Taking the preset monitoring indicator group as an index, respectively extracting indicators and extracting sentences containing the indicators in the device window operation logs in the K device window operation log sequences, and obtaining K device window monitoring indicator group sequences and K log key sentence set sequences, wherein each log key sentence set corresponds to a set of key sentences in a device window operation log; Performing association iterative identification on the K device window monitoring indicator group sequences in chronological order, and obtaining K initial iterative device window operation status indicator groups; Performing indicator trend analysis based on the K log key sentence set sequences to obtain K indicator trend sets; The K indicator trend sets are used to correspondingly identify the indicators in the K initial iterative device window operating status indicator groups to obtain the K iterative device window operating status indicator groups.
4. The reactive power optimization control method for the global wind farm according to claim 3, characterized in that: The K device window monitoring indicator group sequences are associated and iterated in order from the beginning to the end of time to obtain K initial iterative device window operation status indicator groups, including: Extracting K first device window monitoring indicator groups and K second device window monitoring indicator groups from the K device window monitoring indicator group sequences, and performing associated iterative identification on the K first device window monitoring indicator groups and the K second device window monitoring indicator groups, respectively, to obtain K first associated iterative device window monitoring indicator groups; Extracting K third device window monitoring indicator groups from the K device window monitoring indicator group sequences, and performing associative iterative identification on the K third device window monitoring indicator groups and the K first associated iterative device window monitoring indicator groups to obtain K second associated iterative device window monitoring indicator groups; The associated iterative identification is performed in a similar manner, and the K associated iterative device window monitoring indicator groups obtained by the last associated iterative identification are respectively used as the K iterative device window operation status indicator groups.
5. The reactive power optimization control method for the global wind farm according to claim 4, characterized in that: include: Performing a same-indicator mapping similarity analysis on the K first device window monitoring indicator groups and the K second device window monitoring indicator groups to obtain K first mapping similarity sets; The K first mapping similarity sets are standardized, and a convolution calculation is performed on the processing result and the K second device window monitoring indicator groups to obtain the K first associated iterative device window monitoring indicator groups.
6. The reactive power optimization control method for the global wind farm according to claim 3, characterized in that: Based on the K log key sentence set sequences, an indicator trend analysis is performed to obtain K indicator trend sets, including: Extracting log key sentence sequences corresponding to the same type of monitoring indicators from the K log key sentence set sequences to obtain K log key sentence sequence sets, wherein each log key sentence sequence corresponds to a sequence of trend description key sentences arranged from front to back in time in the operation log of a monitoring indicator of a device in a preset monitoring window; The semantic identifier is used to perform indicator trend analysis on the K log key sentence sequence sets respectively to obtain K indicator trend sets, wherein each indicator trend corresponds to a monitoring indicator.
7. The reactive power optimization control method for the global wind farm according to claim 1, characterized in that: The pre-built device reactive power demand analyzer is called to analyze the K iterative device window operation status indicator groups to obtain reactive power demand information of the K devices, including: Acquire multiple sample iteration device window operation status indicator groups and multiple sample device reactive power demand information as training data; The training data is used to perform supervised training on a framework built based on a feedforward neural network, and the one-to-one mapping relationship between the device window operation status indicator group and the device reactive power demand information is learned iteratively until the training converges, thereby obtaining the trained device reactive power demand analyzer.
8. The reactive power optimization control method for the global wind farm according to claim 1, characterized in that: The reactive power optimization control function is: Among them, L min is the output value when the reactive power optimization control function is minimized, is the reactive transmission loss of the ith device, ΔV i is the voltage deviation at the access point of the ith device, and W is the weight coefficient used to balance the loss and voltage deviation.