Wind turbine generator safe operation regulation and control system

Abstract

The invention relates to the technical field of operation regulation and control, in particular to a wind turbine generator safe operation regulation and control system which comprises a state sensing module, a fault coupling recognition module, a state transition control module, a regulation and control strategy screening module and an interface execution isolation module. According to the method, the migration path sequence is formed through historical operation state labels and state migration characteristic extraction, fine control over a state transition mechanism is achieved, matching and comparison are conducted on target parameter intervals in the control instruction screening process, regulation and control actions incompatible with the current working condition are avoided, and the control accuracy is improved. Before the instruction is executed, a risk event is judged and an interface permission isolation strategy is set in combination with a transient electric parameter change trend, so that the fault identification precision and the response pertinence can be improved, and the execution abnormity of the regulation and control instruction caused by an emergent working condition can be effectively avoided; and the safety robustness of full-period operation of the wind turbine generator and the real-time closed-loop consistency of the regulation and control effect are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of operation regulation, and particularly relates to a wind turbine safety operation regulation system. BACKGROUND

[0002] The operation regulation technical field is a core link in the wind power technology system, and involves control and management of state monitoring, power output management, energy conversion efficiency regulation, load allocation and abnormal condition identification of wind turbines during operation. This technical field includes real-time state monitoring, operation parameter optimization, power output stability regulation, grid-connected operation coordination, fault detection and diagnosis strategy of wind turbines, aiming to ensure efficient, safe and stable operation of wind turbines under complex environmental conditions such as wind speed fluctuations and power grid disturbances. Operation regulation technology usually relies on wind farm SCADA systems, wind resource assessment models, power prediction models and power grid interaction control algorithms, emphasizing data interconnection and joint decision mechanism among multiple systems to improve the economy and reliability of the overall wind power system.

[0003] Among them, the wind turbine safety operation regulation system is an integrated system for wind turbine safety control and dynamic regulation, which is mainly used to realize real-time identification, risk assessment and regulation response of wind turbine operation state through collection and analysis of wind speed, current, voltage, speed, mechanical load, vibration and other parameters. The system aims to prevent wind turbines from running inefficiently or causing safety accidents due to abnormal conditions, thereby ensuring the stability and life of wind power equipment and improving the grid quality and power generation efficiency of wind farms.

[0004] The traditional regulation system lacks periodic trend structure extraction and combined change judgment mechanism in parameter fluctuation trend identification and cross-parameter coupling state identification, and is difficult to cope with the multi-parameter linkage effect caused by sudden changes in working conditions. In the fault identification stage, it only relies on single-point threshold judgment or static index early warning, which leads to insufficient identification accuracy of complex fault characteristics. In the state control aspect, no clear state transition path and response sequence is established, and only relies on the current state to execute control instructions, which is easy to cause state jump out of control or time lag of regulation instructions, causing power fluctuations, electrical load overruns and other problems. Especially in the fast switching of grid-connected conditions or extreme weather conditions, such control defects are more likely to be exposed. SUMMARY

[0005] The purpose of the present application is to solve the shortcomings in the prior art and to provide a wind turbine safety operation regulation system.

[0006] In order to achieve the above purpose, the present application adopts the following technical scheme: a wind turbine safety operation regulation system, the system comprises: The state perception module obtains operation parameters collected by a SCADA system in a wind turbine, performs single-period difference on a joint trend sequence of current, temperature rise and output power, judges whether the fluctuation direction appears continuous reversal in three periods and is accompanied by a maximum difference value exceeding a single-period change threshold, and generates a running state parameter fluctuation feature set; The fault coupling identification module judges whether there is a combined coupling difference exceeding a set fault coupling critical threshold based on the running state parameter fluctuation feature set, marks a corresponding operation condition node as a fault high coupling point, and obtains fault coupling node positioning information; The state migration control module extracts a historical instruction response period, a state switching rate and an average value of running state residence time between state pairs according to the fault coupling node positioning information, calculates whether a migration correlation value exceeds a state transition trigger reference value, sets a state pair as a to-be-migrated path, and generates a state transition path identification sequence; The regulation strategy screening module obtains an instruction scheme set bound to a terminal state based on the state transition path identification sequence, judges whether there is a conflict item between current operation parameters and instruction requirements, removes a control action with parameter mismatch, and generates a regulation scheme execution list.

[0007] The application improves that the running state parameter fluctuation feature set includes wind wheel rotation period difference statistics, output power variation rate samples and temperature rise time sequence fluctuation indicators, the fault coupling node positioning information includes high-risk state combination point identification, parameter coupling strength sorting labels and state label indexes in a combination interval, the state transition path identification sequence specifically is a migration state chain number set, a state correlation probability sequence set and a running state transition path pointing index table, and the regulation scheme execution list includes an effective instruction code set, a parameter matching successful instruction identification and a state target corresponding instruction path number.

[0008] The application improves that the state perception module includes: A data acquisition submodule obtains operation parameters collected by a SCADA system in a wind turbine, including wind wheel speed, current value, voltage value, temperature rise value and output power value, monitors speed data of a wind wheel blade root sensor, voltage data of a generator bus voltage node and power output data of a variable pitch system power channel, calls a measurement time sequence, establishes a basic data matrix, and obtains operation parameter basic information; A slope extraction submodule extracts three groups of sequences of current, temperature rise and output power based on the operation parameter basic information, calculates average slope change values of the sequences in adjacent time periods, extracts difference value samples of the current period and the previous period from single-period difference, combines a standard time interval of the current measurement period, calculates power disturbance response degree, screens samples with disturbance response degree exceeding a threshold interval, performs slope anomaly marking, and obtains slope mutation screening results; The fluctuation identification submodule judges whether a direction reverse signal appears in the current, temperature rise and output power sequence in three continuous periods according to the slope mutation screening result, identifies a sample sequence number corresponding to a maximum fluctuation value, screens a time window range, calculates a fluctuation range value interval in combination with a difference between the maximum value in the difference sample and a period average fluctuation threshold value, and generates a running state parameter fluctuation feature set.

[0009] The application improves that the fault coupling identification module comprises: The parameter pairing submodule is based on the running state parameter fluctuation feature set, combines three groups of parameter pairs according to the speed change rate, electromagnetic torque change rate, voltage drop gradient, current response slope, temperature rise increment and main shaft load change amount, divides the parameter pairs into three types of parameter pairing forms of speed-torque, voltage-current and temperature rise-load according to a structural coupling relationship, establishes a pairing matrix, calls time synchronization interval values in a current measurement period, generates parameter combination pairing information, and the like. The difference calculation submodule constructs a slope change vector based on time step difference values in a measurement period according to the parameter combination pairing information, obtains a combination coupling deviation coefficient through operation, sorts each coupling deviation coefficient, extracts combinations higher than a set coupling critical threshold value, and generates a high deviation combination sequence. The coupling positioning submodule calls the high deviation combination sequence, identifies corresponding running condition node numbers, judges whether there is historical trend data of parameter slope mutation in continuous periods in the nodes, marks the nodes as coupling high-risk points and adds them to a positioning list if there is, establishes arrangement order information of coupling combinations contained in each node, and generates fault coupling node positioning information.

[0010] The application improves that the state migration control module comprises: The state pair identification submodule extracts running state labels in a high coupling node in a current wind turbine according to the fault coupling node positioning information, cross-compare the labels with label values in a state label mapping set, calls an adjacency matrix path set between the labels, screens a data pair set with associated edge weights, and obtains a label adjacency state pair set. The migration determination submodule extracts an instruction response period, a state switching rate and an average value of running state residence time corresponding to the state pair according to the label adjacency state pair set, obtains a current wind speed fluctuation amplitude and a wind wheel torque response time as superposition factors, obtains a migration strength value of the state pair through operation, judges whether the migration strength value exceeds a state conversion trigger reference value, marks the state pair as a to-be-migrated path if the migration strength value exceeds the state conversion trigger reference value, and generates a migration strength screening result. The path construction submodule calls the existing target state number sequence in the state label path dictionary according to the migration strength screening result, rearranges the node index order in the target state sequence, connects the state label to the path identifier set between the target state nodes, establishes the label mapping information of the target state path, and generates the state transition path identifier sequence.

[0011] The application improves that the regulation strategy screening module comprises: The instruction extraction submodule obtains the state transition path identifier sequence, calls the strategy library instruction index table corresponding to the terminal state number, screens the regulation instruction items bound to the current node, extracts the action type, applicable state number and regulation target parameter set corresponding to each instruction, and generates a target instruction index set; The parameter verification submodule, based on the target instruction index set, according to the required current peak value range, wind wheel deceleration time and winding temperature rise limit value of each instruction, calls the running current, wind wheel speed change duration and winding temperature acquisition sequence in the current measurement period, judges whether each operation value is within the instruction required threshold interval, screens the effective instruction items meeting the constraint requirements, and obtains the parameter adaptation screening result; The channel screening submodule, based on the parameter adaptation screening result, according to the effective instruction items, matches the executable channel number in the regulation path mapping table, judges whether the channel is in the available state and has not triggered the protection limit logic, eliminates the channel items with action conflict or execution path conflict, and generates a regulation scheme execution list.

[0012] The application improves that the system further comprises: The interface execution isolation module judges whether there is an abnormal event of continuous voltage drop in the instruction period and accompanied by current negative jump exceeding the short-time drop threshold according to the regulation scheme execution list, and if it is satisfied, the interface issuing authority of the corresponding control instruction is frozen and the interface port state is recorded to obtain a regulation command execution isolation mark; The regulation command execution isolation mark specifically refers to the interface channel frozen number, control instruction waiting list position and fault state identification label.

[0013] The application improves that the interface execution isolation module comprises: The signal extraction submodule, based on the regulation scheme execution list, according to the execution channel number of the control instruction, calls the bus voltage jump value, current negative slope time slice distribution and wind wheel speed instantaneous response state corresponding to each channel in the controller, extracts the sampling sequence and constructs the parameter matrix according to the channel period, and generates control signal monitoring information; The abnormality identification submodule monitors information based on the control signal, judges whether there is a voltage value descending trend of more than two time points in each record according to a period sequence of voltage and current, and whether there is a record entry accompanied by a current jump amplitude greater than a short-time drop threshold at the same time, screens an abnormal channel number set, and obtains an abnormal channel index sequence; The permission freezing submodule calls the abnormal channel index sequence, locates a current state in an interface management unit according to a control instruction number corresponding to a marked channel, judges whether the state is in an execution front queue, if yes, stops instruction issuing and records a freezing time point state label and a channel state, establishes a control channel freezing registration table, and generates a control command execution isolation mark.

[0014] Compared with the prior art, the advantages and positive effects of the present application are that: In the present application, periodic difference and trend slope inversion identification are implemented on time sequence operation parameters collected by the SCADA system, a fluctuation set containing combined variation characteristics of wind wheels, electrical and thermal loads is constructed, a slope difference normalization sorting method is introduced based on the characteristic set to identify the combined coupling relationship between parameters, and a potential fault high coupling node in the working condition is determined accordingly, a migration path sequence is extracted in combination with historical operation state labels and state transition characteristics, fine control of the state transition mechanism is realized, matching comparison of target parameter intervals is performed in the process of screening control instructions, incompatible control actions with the current working condition are avoided, risk events are determined in combination with transient electrical parameter variation trends before instruction execution, and an interface permission isolation strategy is set, which can improve fault identification accuracy and response specificity, effectively avoid execution abnormalities of control instructions caused by sudden working conditions, and improve the safety robustness of the wind turbine during the whole cycle and the real-time closed-loop consistency of the control effect. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 The system flowchart of the present application is shown in the figure; Figure 2 The flowchart of the state perception module of the present application is shown in the figure; Figure 3 The flowchart of the fault coupling identification module of the present application is shown in the figure; Figure 4 The flowchart of the state transition control module of the present application is shown in the figure; Figure 5 The flowchart of the control strategy screening module of the present application is shown in the figure; Figure 6 The flowchart of the interface execution isolation module of the present application is shown in the figure. DETAILED DESCRIPTION

[0016] In order to make the objects, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not to limit the present application.

[0017] In the description of the present application, it should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation on the present application. In addition, in the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise explicitly and specifically limited.

[0018] Please refer to Figure 1 The present application provides a technical solution: a wind turbine safe operation control system, the system comprises a state sensing module, a fault coupling identification module, a state transition control module, a control strategy screening module and an interface execution isolation module; The state sensing module obtains the operating parameters collected by the SCADA system in the wind turbine, including the wind wheel speed, current value, voltage value, temperature rise value and output power value, extracts the fluctuation slope change of adjacent time periods according to the time series measurement results of the wind wheel blade root sensor, generator bus voltage node and variable pitch system power channel, performs single period difference on the joint trend sequence of current, temperature rise and output power, judges whether the fluctuation direction appears continuous reversal in three periods and is accompanied by maximum difference exceeding the single period change threshold, and generates the operating state parameter fluctuation feature set; The fault coupling identification module calls the rotation speed change rate and electromagnetic torque change rate, voltage drop gradient and current response slope, temperature rise increment and main shaft load change amount three groups of parameters based on the operating state parameter fluctuation feature set, sorts the slope difference of the double parameter combinations, judges whether there is a combination coupling difference exceeding the set fault coupling critical threshold, if there is, marks the corresponding operating condition node as a fault high coupling point, establishes the combination coupling sequence of the corresponding point in the parameter space, and obtains the fault coupling node positioning information; The state transition control module calls the operating state label of the current wind turbine and the adjacent state path set according to the fault coupling node positioning information, extracts the state pair identified by the corresponding fault high coupling node, performs state label interactive query, judges whether it constitutes a transition critical pair, extracts the historical instruction response period, state switching rate and average value of operating state residence time between the state pairs, calculates whether the transition correlation value exceeds the state transition trigger reference value, if it exceeds, sets the state pair as a to-be-transitioned path, constructs a path identification chain pointing to the target state, and generates a state transition path identification sequence; The regulation strategy screening module identifies a state transition path sequence, acquires a set of instruction schemes bound to the terminal state, and matches the regulation instruction to the required current peak value range, wind wheel deceleration time, and winding temperature rise limit value, judges whether the current operating parameters conflict with the instruction requirements, if there is no conflict, performs execution channel matching screening on the scheme set, removes control actions with parameter mismatches, and generates a regulation scheme execution list; The interface execution isolation module calls the bus voltage jump value, current negative slope time slice distribution, and wind wheel speed instantaneous response state within the execution cycle issued by the controller according to the regulation scheme execution list, judges whether there is an abnormal event of continuous voltage drop within the instruction cycle accompanied by current negative jump exceeding the short-time drop threshold, if it is satisfied, freezes the interface issuing authority of the corresponding control instruction and records the interface port state, and obtains a regulation command execution isolation marker; The operating state parameter fluctuation feature set includes wind wheel rotation period difference statistics, output power variation rate samples, and temperature rise time sequence fluctuation indicators, the fault coupling node positioning information includes high-risk state combination point identification, parameter coupling strength sorting label, and state label index in the combination interval, the state transition path identification sequence is specifically a set of state chain number, a set of state association probability order, and an index table of operating state transition path, the regulation scheme execution list includes a set of valid instruction codes, parameter matching successful instruction identification, and state target corresponding instruction path number, and the regulation command execution isolation marker specifically refers to interface channel freezing number, control instruction waiting list position, and fault state identification label.

[0019] Please refer to Figure 2 , the state perception module includes: The data acquisition submodule acquires operating parameters collected by the SCADA system in the wind turbine, including wind wheel speed, current value, voltage value, temperature rise value, and output power value, monitors the speed data of the wind wheel blade root sensor, the voltage data of the generator bus voltage node, and the power output data of the variable pitch system power channel, calls the measurement time sequence, and establishes a basic data matrix to obtain the operating parameter basic information; The wind turbine speed, current value, voltage value, temperature rise value and output power value collected by the SCADA system in the wind turbine are acquired, and the acquisition period is set to 10 seconds for a complete sampling period. Relying on the wind wheel root sensor, 60 groups of speed data points are acquired, and the speed data recording unit is r / min. In each sampling period, a single group of data is extracted per second to form a speed sequence. The generator bus voltage is detected by the bus measurement point, and the range is set to 0V to 1200V. In actual sampling, the detection range is commonly between 620V and 740V. 60 data points are collected in a single period, and the voltage value is collected and paired according to the 10ms synchronous timestamp. The power channel output power of the pitch system is measured in kW, and the maximum value is set to 2100kW in combination with the wind speed monitoring value. In this example, 54 power output points are collected, and the missing data segment under the condition of partial wind speed sudden drop needs to be marked for processing. In the data alignment processing stage, the data collected by the three channels are aligned based on the master control system timestamp to construct a unified 60x5-dimensional basic data matrix. Each column represents the speed, current, voltage, temperature rise and power. In order to illustrate the effectiveness of the example, the data of one measurement period is selected as follows: Table 1 Partial monitoring parameter table of wind turbine As shown in Table 1, the speed shows a steady upward trend, the voltage remains above 700V, and the power output increases synchronously. After synchronizing the data of the above three channels, the basic data matrix is constructed, and the running parameter basic information is obtained.

[0020] The slope extraction submodule extracts three groups of sequences of current, temperature rise and output power based on the running parameter basic information, calculates the average slope change value of the sequences in the adjacent time period, extracts the difference sample of the current and the previous cycle from the single cycle difference, and combines the standard time interval of the current measurement period. The formula is as follows: ; The operation obtains the power disturbance response degree, selects the samples whose disturbance response degree exceeds the threshold interval, marks the slope anomaly, and obtains the slope mutation screening result; Among them, represents the power disturbance response degree, represents the temperature rise normalization value in the current cycle, represents the temperature rise normalization value in the previous cycle, represents the output power normalization value in the current measurement cycle of current, represents the current average change duration normalization value, represents the wind speed variation intensity coefficient, which is used to reflect the disturbance background meteorological excitation level, and the value is taken from the ratio of the maximum wind speed to the minimum wind speed in the measurement period, represents the voltage normalization value of the current cycle, This represents the normalized voltage value of the previous cycle. This is a meteorological background excitation correction factor used to improve the sensitivity of the calculation to sudden wind speed disturbances; Based on the basic operating parameters, three sequences are extracted: current, temperature rise, and output power. Each sequence is 60 units in length. First, the length of the sequence is calculated. Item and the The difference between the terms is divided by the 1-second interval between the two points to obtain the average slope change value sequence. After obtaining the slope difference value sequence for the current period, it is subtracted from the corresponding term of the slope sequence of the previous period to calculate the difference sample. Taking output power as an example, the output power at the 20th second and the 19th second are 1542kW and 1534kW respectively, with a slope of 8kW / s. This point is substituted into the difference sequence, and then combined with the standard time interval of 10 seconds for measurement period to perform unit conversion, so that the slope difference value has a unified time base. Then, the wind speed variation intensity parameter is called. The value is obtained from the ratio of the maximum wind speed of 16.2 m / s to the minimum wind speed of 13.4 m / s: ; Substitute into the formula to calculate the disturbance response: ; The calculation logic of this formula is as follows: temperature rise change value This indicates the intensity of thermal disturbance to the equipment, and is a power-current-time composite factor. The combined factor is used to construct a dynamic comprehensive factor of load, combined with the wind speed disturbance excitation factor. The amplitude is adjusted, and finally the square root of the voltage difference is used. As the denominator for disturbance suppression, it reflects the stability characteristics of the system itself. The formula comprehensively reflects the system's state sensitivity under the influence of multiple sources of disturbance.

[0021] The parameters are explained below: Power disturbance response; , The normalized temperature rise values ​​are derived from the actual temperature rise values ​​of 63.5℃ and 54.6℃, respectively, and normalized to the rated upper limit of 120℃. Normalized output power, the original value is 1480kW, normalized to the maximum of 2100kW; : Normalized value of average current change time, the original value is 6.8s, normalized to a period of 10s; Wind speed fluctuation intensity, calculated based on the ratio of the maximum to the minimum wind speed change; , Voltage normalized value, the original value is 744V and 780V, normalized by 1200V; The determination threshold of the slope mutation screening is set to 0.065, which is derived from the lower limit of the 95% confidence interval of the slope disturbance response degree distribution in 2000 groups of wind power operation samples. The test group data shows that most of the non- abnormal sample values are concentrated between 0.02 and 0.06. Therefore, the upper limit is set to 0.065 as the dynamic discrimination standard to avoid missing edge state samples. The sample disturbance result is 0.0775, which has obviously exceeded the threshold, so it is judged as an abnormal sample, and the slope mutation screening result is obtained.

[0022] The fluctuation identification submodule judges whether the direction reversal signal appears in the current, temperature rise and output power sequence within three continuous periods according to the slope mutation screening result, identifies the sample sequence number corresponding to the maximum fluctuation value, and screens the time window range. Combined with the difference between the maximum value in the difference sample and the period average fluctuation threshold, the fluctuation range value interval is calculated, and the operating state parameter fluctuation feature set is generated; According to the sample sequence number marked as abnormal in the slope mutation screening result, the original data segment in the current, temperature rise and output power sequence is called. A 5-point window is constructed with each abnormal point as the center and each extending 2 time points forward and backward. It is judged whether the sign change sequence of the corresponding slope in the window appears in the positive-negative alternating reversal structure in three continuous periods. The structure that meets the condition is recorded as a direction reversal signal. At the same time, the fluctuation range is judged with the reversal position point as the center. The difference between the maximum value and the minimum value of each parameter is extracted. Taking the power sequence as an example, the maximum power in a window is 1635kW, and the minimum value is 1550kW. Therefore, the range is 85kW. The period average fluctuation threshold is set to 75kW, which is set in reference to the average value of the effective power fluctuation amplitude in 20 operating periods. The average value is 73.4kW, which is rounded to 75kW as the threshold value. The sample range 85kW has exceeded the threshold value, which meets the fluctuation mutation condition. Therefore, it is confirmed that the point is the fluctuation main trigger point. Its index, timestamp and parameter combination are marked into the feature set. Finally, all sampling paragraphs that meet the three-period reversal condition and the fluctuation amplitude condition are counted to generate the operating state parameter fluctuation feature set.

[0023] Please refer to Figure 3 The fault coupling identification module includes: The parameter pairing submodule is based on the operating state parameter fluctuation feature set. According to the speed change rate, electromagnetic torque change rate, voltage drop gradient, current response slope, temperature rise increment and main shaft load change, three groups of parameter pairs are combined and divided into three types of parameter pairing forms of speed-torque, voltage-current and temperature rise-load according to the structure coupling relationship. The pairing matrix is established, and the time synchronization interval value in the current measurement period is called to generate the parameter combination pairing information. Six types of parameter items in the operating state parameter fluctuation feature set, i.e., the rotational speed change rate, the electromagnetic torque change rate, the voltage drop gradient, the current response slope, the temperature rise increment and the main shaft load change amount, are obtained, and 60 groups of data are extracted in each 10-second sampling period. After normalization processing, the normalized value interval is all mapped to [0, 1] so as to facilitate subsequent combination operation. According to the pre-defined structure coupling mode, the six types of parameter items are divided into three types of combinations, i.e., the rotational speed and the electromagnetic torque, the voltage and the current, and the temperature rise and the main shaft load, which respectively correspond to the rotating power chain, the electrical driving chain and the thermal-mechanical coupling chain. The combination logic is derived from the unit power and stress transmission path. From the main shaft rotational speed to the generated torque, the electrical voltage drives the current response, and the temperature rise process reflects the interactive energy evolution of the load and the structure. Two-dimensional arrays are respectively established for the three groups of combinations, and two parameter normalized values at each time point are arranged side by side to construct a 60-row-by-2-column data pair array. Due to the signal acquisition delay and synchronization error of different systems of the wind turbine, a system synchronization mechanism needs to be introduced. The synchronization window of the system is set to ±0.5s, i.e., taking the median time point as the reference, the average value of the two parameter data pairs in the front and rear 0.5s is calculated to filter out transient interference and time sequence deviation. For example, the voltage normalized value is 0.62 and the current is 0.57 at the 12th second. The front and rear data are 0.63, 0.58 and 0.61, 0.56. The average paired values are 0.62 and 0.57, respectively, which constitute the voltage-current combination data point at this time point. The combination sequence is constructed, and the parameter combination pairing information is generated.

[0024] The difference calculation sub-module constructs the slope change vector based on the time step difference in the measurement period according to the parameter combination pairing information, and uses the formula: ; The combination coupling deviation coefficient is obtained by operation. Each coupling deviation coefficient is sorted, and the combination with a coupling critical threshold higher than the set coupling critical threshold is extracted to generate a high deviation combination sequence. wherein, represents the combination coupling deviation coefficient, represents the slope change value of the first item in the parameter pair per unit time, represents the slope change value of the second item in the parameter pair per unit time, represents the wind speed fluctuation amplitude normalized value, represents the wind wheel acceleration normalized value, represents the maximum vibration frequency normalized value in the current period, represents the temperature rise change periodicity amplitude normalized value; According to the parameter combination pairing information, the normalized parameter sequence of each item in each paired data is extracted, and the difference operation is performed on adjacent time points to obtain the parameter item and That is, the slope change value in unit time, and then calculate the absolute value of the difference to construct the coupling difference index, select the representative period sample as follows: Table 2 Parameter on Slope Difference Example Table As shown in Table 2, in the combination of 20 seconds and 30 seconds, the difference reaches 0.016, which belongs to the higher deviation range, and then it is multiplied by the wind speed fluctuation range And the wind wheel acceleration The sum of the current period , normalized by the maximum wind speed 15.8 m / s and the minimum 13.1 m / s, Normalized by the wind wheel acceleration calculation, the maximum vibration frequency , normalized from the current period main shaft vibration frequency, the temperature change period amplitude is , into the formula: ; The calculation logic of the formula is as follows: the numerator Reflects the dynamic deviation between parameters, and the disturbance excitation factor After multiplication, the excitation amplification term is obtained, and the denominator part Is the system damping correction term, reflecting the system's own oscillation characteristics and heat buffering capacity, the whole structure reacts to the coupling distortion degree under the condition of strong excitation and weak response, the larger the calculation result is, the more serious the coupling structure imbalance is; The coupling critical threshold is set to 0.0062, which is obtained according to the coupling index statistics of the 80th percentile in 500 samples, the result 0.0076 is greater than the threshold, which is marked as a coupling high deviation combination, and added to the high deviation combination sequence, to obtain the high deviation combination sequence.

[0025] The coupling positioning sub-module calls the high deviation combination sequence, identifies the corresponding operating condition node number, judges whether there is historical trend data of parameter slope mutation in continuous period in the node, if there is, mark the node as a coupling high-risk point and add it to the positioning list, establish the arrangement order information of each node containing coupling combination, and generate fault coupling node positioning information; Call the parameter index information in the high bias combination sequence, corresponding to the node number of the running state in the structure working condition record table in the system, such as the node number corresponding to the speed-torque combination is N45, and the slope history mutation trend sequence is consulted in the node. The sequence identifies whether there is a continuous 3-period slope change direction reversal record in the current period to the previous 5 periods. The judgment standard is the same as in the slope mutation, which needs to meet the polarity reversal of the slope difference value more than twice, and at least one slope difference value is greater than 0.02. For example, the records in node N45 are positive-negative-positive, and the corresponding difference values are 0.015, 0.027, and 0.022, respectively. The reverse + intensity condition is met, the node anomaly is confirmed, and it is marked as a coupled high-risk point. The three parameter combinations associated with the structure are sorted according to the deviation size, and the sorting information is recorded as voltage-current, temperature rise-load, and speed-torque in order. The sorting information is written into the positioning list to form the node corresponding structure sequence mapping table, and finally the index binding relationship between the working condition node and the bias source is established to generate the fault coupling node positioning information.

[0026] Please refer to Figure 4 , the state migration control module comprises: The state pair identification submodule extracts the running state label of the current wind turbine under the high coupling node according to the fault coupling node positioning information, and cross-comparisons the label value in the state label mapping set. The adjacency matrix path set between the labels is called to filter the data pair set with associated edge weight to obtain the label adjacency state pair set. Get the high coupling node index information recorded in the fault coupling node positioning information, for example, the node number is set to , the node is marked as an abnormal high-risk working condition area in the current measurement period. Based on the state monitoring system of the current wind turbine, the real-time running state label recorded under the node is extracted, which represents the state identification number corresponding to the wind turbine under multiple operating indicators (current, wind speed, speed, etc.). The pre-stored state label mapping set is called, and the mapping set records a total of state labels. The current label is compared with the state labels of the remaining to form state pairs, respectively, wherein . All possible label combinations are traversed and compared, and the state label adjacency matrix in the system is further called. The matrix is dimensional, and any element represents the adjacency weight value between state labels and . If exists, it is considered that there is a reachability path between the state pair, which has relevance. is the source state label, query the state labels with non-zero corresponding weight in the adjacent path, such as , , , whose adjacent weights are , , , are all higher than the typical lower limit value 0.10 of non-associated state. The above state pair set with associated weight is screened as the label adjacent state pair set, and its corresponding label serial number and adjacent matrix weight are recorded. Finally, this set will be used as the input data source for subsequent state migration evaluation.

[0027] The migration determination submodule extracts the instruction response period, state switching rate and average running state residence time corresponding to the state pair from the label adjacent state pair set, and obtains the current wind speed fluctuation amplitude and wind turbine torque response time as superimposed factors, using the formula: ; to obtain the migration intensity value of the state pair, and determine whether the migration intensity value exceeds the state transition trigger reference value. If it exceeds, mark the state pair as a to-be-migrated path, and generate a migration intensity screening result; wherein, represents the migration intensity value, represents the instruction response period normalized value, represents the state switching rate normalized value, represents the average running state residence time normalized value, represents the wind speed fluctuation amplitude normalized value, represents the wind turbine torque slope normalized value, represents the normalized difference value of the speed peak value, represents the normalized value of the standard deviation of the axial vibration frequency of the unit main shaft, represents the weighted adjustment term of the state migration uncertainty of the vibration amplitude change; According to the label adjacent state pair set, extract the main indicators in the historical state migration process for each state pair in the set, including: the instruction response period normalized value, denoted as , representing the average time required for the system to respond after the control instruction is issued; the state switching rate normalized value, denoted as , representing the frequency of state label switching per unit time; the normalized average value of the current state residence time, denoted as , representing the average length of time that the unit remains in the same state under the current state.

[0028] In addition, the following four disturbance factors are extracted from the wind turbine monitoring data in the current period: Wind speed fluctuation amplitude normalized value , i.e. the difference between the maximum and minimum wind speed in the measurement period divided by the rated wind speed; Wind wheel torque slope normalized value , used to reflect the load disturbance when the wind wheel rotates; Rotational speed peak normalized difference value , i.e. the difference between the maximum value and the period average value in the rotational speed data; Main shaft axial vibration frequency standard deviation normalized value , used to measure the instability degree of the dynamic response of the unit structure.

[0029] Substitute the above participation items into the formula: ; The first part of the formula is the state reactivity ratio, the numerator part reflects the state activity (i.e. the sum of the control response and the state switching double indicators), and the denominator part introduces the current state stability factor, 1 is the balance term to avoid zero denominator; the root after the multiplication sign represents the disturbance accumulation term, which is the sum of the wind speed disturbance , the rotational speed peak and the vibration frequency adjustment term for uncertainty .

[0030] Actual calculation is carried out by substituting sample data (from real-time recording of the system): , , ; , , , ; The calculation is as follows: ; The result value represents the migration intensity value between the state pair as 0.2645, representing the tendency of the pair to migrate from the previous state to the subsequent state under the current disturbance background. According to the state transition trigger reference value set in the wind turbine operation data, which is 0.21 (derived from the average migration intensity of the top 20% samples with the highest migration rate in all state pairs, to ensure sufficient sensitivity and stability), since , the state pair is marked as a to-be-migrated path and is included in the migration intensity screening result.

[0031] The path construction submodule, based on the migration intensity filtering results, calls the existing target state number sequence in the state label path dictionary, rearranges the node index order in the target state sequence, connects the path identifier set between the state label and the target state node, establishes the label mapping information of the target state path, and generates the state transition path identifier sequence. Based on the marked status in the migration strength screening results Enter the wind turbine status label path dictionary and search for the target label. All corresponding subsequent target state nodes. Based on the historical switching path, the target sequence is: The system internally stores the priority scores for state jumps between each path, such as transition stability scores. They are respectively , , Based on this, the path priorities are reordered as follows: .

[0032] The set of path identifiers from the call status label to the target node, corresponding to the following path numbers: , , , , connect to main tag Generate a new path chain from the above path sequence: And record the corresponding path sequence as Simultaneously, a mapping relationship is established between the nodes contained in the path and the path number, and the final output is a sequence of state transition path identifiers. This sequence serves as the valid path chain for the controller state machine to execute transitions and is read by the system control instruction module at the execution layer.

[0033] Please see Figure 5 The regulation strategy screening module includes: The instruction extraction submodule obtains the state transition path identifier sequence, calls the policy library instruction index table corresponding to the endpoint state number, filters the control instruction entries bound to the current node, extracts the action type, applicable state number and control target parameter set corresponding to each instruction, and generates a target instruction index set. The terminal state number in the state transition path identification sequence is obtained, and the number is parsed to locate the current wind turbine at the end of the operation cycle state. The corresponding number such as "S17" indicates that it has entered the high-load steady-state interval. The instruction entry recorded in the strategy library instruction index table is called, the action type bound in the control instruction is extracted through the structure traversal method, for example, the action is "power reduction control" or "winding cooling control", and then the applicable state number such as "S17, S18" and the corresponding control target parameter set bound in each instruction are extracted, for example, the current peak limit is 480A, the wind wheel deceleration time is 6.2 seconds, and the winding temperature rise should not exceed 65℃. In this process, a dictionary access strategy should be adopted for the binding structure in the strategy library to avoid redundant extraction caused by instruction rearrangement. The instruction tag index sequence is established by comparison and extraction, and is saved to the target instruction index set. As shown in Table 3, the strategy library instruction item data under the current terminal state number "17" is listed.

[0034] Table 3 Strategy library instruction item data table As shown in Table 3, the current state "S17" corresponds to three strategy instructions, each instruction has a unique instruction number, a control action type, an adaptive state number and a control target parameter, forming a target instruction index set.

[0035] The parameter verification submodule calls the running current, wind wheel speed change duration and winding temperature acquisition sequence in the current measurement period based on the target instruction index set according to the current peak range, wind wheel deceleration time and winding temperature limit required by each instruction, judges whether each operation value is within the instruction required threshold interval, selects the effective instruction item meeting the constraint requirement, and obtains the parameter adaptation screening result. According to the target instruction index set, the regulation target parameter item of each regulation instruction is analyzed in turn, the current peak range, the wind wheel deceleration time and the winding temperature limit value are extracted, and the threshold is set as the value in the instruction ± 3% floating interval, that is, the threshold corresponding to the D_021 instruction is the current peak range 466A to 494A, the wind wheel deceleration time 6.016.39s, and the winding temperature limit value 63.05~66.95℃, the current measurement sequence in the current measurement period is called, such as [488, 489, 492]A, the speed change duration sequence is such as [6.15, 6.22, 6.28]s, and the winding temperature sequence is such as [64.8, 65.2, 64.9]℃, the three parameters at each time step are judged with the corresponding threshold interval, and if all parameters at a certain time fall into the interval, it is marked as an effective instruction adaptation point. The number of valid samples is counted, and the instruction item with more than three is screened out to form the parameter adaptation screening result. The above judgment process needs to clearly define the limiting method of the regulation target parameter floating space. The floating space ± 3% value is based on the redundant setting standard reserved for the element bearing capacity in the regulation strategy setting process. The setting is derived from the preset redundant tolerance band of the unit electrical load capacity measured data comparison, which ensures that the result has quantitative and reproducible characteristics, and forms a reusable screening logic.

[0036] The channel screening submodule is based on the parameter adaptation screening result, matches the executable channel number in the regulation path mapping table according to the effective instruction item, judges whether the channel is in the available state and has not triggered the protection limit logic, eliminates the channel item with action conflict or execution path conflict, and generates the regulation scheme execution list; The effective instruction item in the parameter adaptation screening result, such as D_021 and D_023, is called according to the mapping rule to enter the regulation path mapping table, the corresponding executable channel number is queried, such as D_021 mapping channel T3 and D_023 mapping channel T4, the current state marker information of the channel is extracted, it is judged whether it is in the "running state", and the dispatching control log is queried to confirm that it has not triggered the protection logic in the recent period. If the channel state is "disabled" or there is an action occupation record in the last execution period, the channel is marked as an action conflict, and the available channel set is eliminated. Finally, all effective instruction items and their channel execution states are paired one by one to remove the conflict items and generate the regulation scheme execution list. In this step, the channel available state judgment is based on the dispatching control identification field "status". The value of the status field is "1" indicating executable, and "0" indicating disabled. In this round of state judgment, only D_021 is reserved among the three available instructions to form a unique effective regulation instruction scheme, and the channel execution structure index is established.

[0037] Please refer to Figure 6 , the interface execution isolation module comprises: The signal extraction submodule executes the regulation scheme execution list, calls the bus voltage jump value, current negative slope time slice distribution and wind turbine speed instantaneous response state corresponding to each channel in the controller according to the execution channel number of the control instruction, extracts the sampling sequence according to the channel period and constructs the parameter matrix to generate the control signal monitoring information; The execution channel numbers of all control instructions in the regulation scheme execution list are obtained, the numbers such as T5, T7 and T11 are extracted, the signal acquisition unit bound with these channel numbers in the controller is called, the corresponding bus voltage jump value record, current negative slope time slice sequence and instantaneous response state of the wind turbine speed at each time point are obtained from the data storage area, the parameter three-element sequence set based on the time point is constructed respectively, the sampling frequency is unified to 10 Hz, the extraction period range covers the latest two complete control periods, the period length is set to 12 s, the sequence length obtained by each channel is 120 data, the 120*3-dimensional data matrix is constructed, the channel numbers are numbered as M_5, M_7 and M_11 respectively, the sampling time point and the corresponding parameter value sequence are recorded in the form of a table, and the following is the T5 channel record data: Table 4 Parameter sampling record table of control channel T5 As shown in Table 4, after extracting the period sequence of each channel, the voltage jump value, current change rate and wind turbine speed at different time points are combined into control signal monitoring information.

[0038] The abnormality identification submodule judges whether there is a voltage value descending trend at two or more consecutive time points in each record and whether there is a record item accompanied by a current jump amplitude greater than a short-time drop threshold value according to the period sequence of the voltage and the current based on the control signal monitoring information, screens the abnormal channel number set, and obtains the abnormal channel index sequence. Based on the control signal monitoring information constructed in paragraph 1, the data matrix corresponding to each channel is traversed in turn, the voltage jump sequence is directionally differentiated, the voltage difference between each sampling time point and the previous time point is obtained, whether there are two or more consecutive difference values less than 0 in the sequence is recorded, indicating that the voltage is continuously decreasing, and the current negative slope sequence is extracted, the absolute value of each sample value is taken, and is compared with the short-time drop threshold 0.12, if any sampling point satisfies the voltage continuous decrease and the corresponding current jump amplitude is greater than 0.12A / s, the channel is marked as an abnormal channel, and the channel number is added to the abnormal marking list; The threshold setting process is as follows: according to the data accuracy set by the controller and the normal noise jitter range (about ±0.05A / s) of the wind power electronic control system, reserve twice the disturbance redundancy, calculate 0.05*2+0.02=0.12A / s as the abnormal recognition critical threshold, and confirm that the threshold can cover more than 85% of the instantaneous jump behavior through random sampling of 8 groups of samples under different wind conditions, which has sufficient engineering reproducibility, and on this basis, an abnormal channel index sequence is formed.

[0039] The permission freezing sub-module calls the abnormal channel index sequence, locates the current state in the interface management unit according to the control instruction number corresponding to the marked channel, judges whether the state is in the execution of the front queue, if yes, stops the instruction issuing and records the freezing time point state label and the channel state, establishes a control channel freezing registration table, and generates a control command execution isolation mark; Call the abnormal channel index sequence marked in paragraph 2, extract the control instruction number corresponding to it, for example, the instruction number corresponding to T5 channel is D_021, enter the state management table of the interface management module, call the current execution state mark of D_021 instruction, if the state field is “to be issued”, it means that it has not been issued to the execution register queue, immediately rewrite the instruction execution state as “frozen”, and record the current freezing time label such as “2025-07-1610:28:00” and the channel number state as “abnormal blocking”, combine the freezing event information into a freezing registration structure, add it to the unified freezing registration list, and at the same time, generate a unique isolation mark field “ISO_D021” for the instruction according to the controller module marking rule, which represents that the instruction D_021 is isolated and blocked due to the abnormality of channel T5, forming a control command execution isolation mark.

[0040] The above is only a preferred embodiment of the present application, and does not limit the present application in other forms. Any skilled person in the art can modify or change the above disclosed technical content to equivalent embodiments applied to other fields, but any simple modification, equivalent change and modification made on the basis of the technical essence of the present application to the above embodiments still belongs to the protection scope of the technical solution of the present application.

Claims

1. A wind turbine safe operation control system, characterized in that, The system includes: The state perception module acquires the operating parameters collected by the SCADA system in the wind turbine, performs single-cycle difference on the joint trend sequence of current, temperature rise and output power, determines whether the fluctuation direction has continuously reversed within three cycles and is accompanied by the maximum difference exceeding the single-cycle change threshold, and generates a set of operating state parameter fluctuation characteristics. Based on the set of operating parameters fluctuation features, the fault coupling identification module determines whether there is a combined coupling difference that exceeds the set fault coupling critical threshold, marks the corresponding operating condition node as a high fault coupling point, and obtains the fault coupling node location information. Based on the fault coupling node location information, the state transition control module extracts the historical command response cycle, state switching rate and average running state dwell time between state pairs, calculates whether the migration association value exceeds the state transition trigger benchmark value, sets the state pair as the path to be transferred, and generates a state transition path identifier sequence. The control strategy filtering module obtains the instruction scheme set bound to the endpoint state based on the state transition path identifier sequence, determines whether there are any conflicts between the current running parameters and the instruction requirements, removes control actions with mismatched parameters, and generates a control scheme execution list.

2. The wind turbine safe operation control system according to claim 1, characterized in that, The set of operational parameter fluctuation features includes statistics on the difference in wind turbine rotation period, samples of output power change rate, and temperature rise time series fluctuation indicators. The fault coupling node location information includes high-risk state combination point identifiers, parameter coupling strength sorting labels, and state label indexes within the combination interval. The state transition path identifier sequence specifically includes a set of migration state chain numbers, a set of state association probability order sets, and an index table pointing to operational state transition paths. The control scheme execution list includes a set of valid instruction codes, parameter matching success instruction identifiers, and instruction path numbers corresponding to state targets.

3. The wind turbine safe operation control system according to claim 2, characterized in that, The state awareness module includes: The data acquisition submodule acquires the operating parameters collected by the SCADA system in the wind turbine, including rotor speed, current value, voltage value, temperature rise value and output power value. It monitors the speed data of the rotor blade root sensor, the voltage data of the generator bus voltage node, and the power output data of the pitch system power channel. It calls the measurement time series and establishes a basic data matrix to obtain basic information on operating parameters. Based on the aforementioned basic information of operating parameters, the slope extraction submodule extracts three sets of sequences: current, temperature rise, and output power. It calculates the average slope change value of the sequences in adjacent time periods, extracts the difference sample between the current and previous cycles from the single-cycle difference, and calculates the power disturbance response degree by combining it with the standard time interval of the current measurement cycle. It then filters samples whose disturbance response degree exceeds the threshold range, marks slope anomalies, and obtains the slope mutation screening results. Based on the slope change filtering results, the fluctuation identification submodule determines whether there is a direction reversal signal within three consecutive cycles in the current, temperature rise and output power sequences, identifies the sample number corresponding to the maximum fluctuation value, filters the time window range, and calculates the fluctuation range by combining the difference between the maximum value in the difference sample and the average fluctuation threshold of the cycle, and generates the operating state parameter fluctuation feature set.

4. The wind turbine safe operation control system according to claim 3, characterized in that, The fault coupling identification module includes: Based on the set of operating parameters fluctuation characteristics, the parameter pairing submodule combines three sets of parameter pairs according to the rate of change of rotational speed, rate of change of electromagnetic torque, voltage drop gradient, current response slope, temperature rise increment, and spindle load change. These pairs are then classified into three types of parameter pairing forms according to structural coupling relationships: speed-torque, voltage-current, and temperature rise-load. A pairing matrix is ​​established, and the time synchronization interval value within the current measurement cycle is called to generate parameter combination pairing information. The difference calculation submodule constructs a slope change vector based on the time step difference within the measurement period according to the parameter combination pairing information, calculates and obtains the coupling deviation coefficient between combinations, sorts each group of coupling deviation coefficients, extracts combinations that are higher than the set coupling critical threshold, and generates a high deviation combination sequence. The coupling positioning submodule calls the high deviation combination sequence, identifies the corresponding operating condition node number, determines whether the node has historical trend data of parameter slope change within a continuous period, if so, marks the node as a high-risk coupling point and adds it to the positioning list, establishes the arrangement order information of the coupling combination contained in each node, and generates fault coupling node positioning information.

5. The wind turbine safe operation control system according to claim 4, characterized in that, The state transition control module includes: The state pair identification submodule extracts the operating state label of the current wind turbine under the high coupling node based on the fault coupling node location information, cross-compares it with the label value in the state label mapping set, calls the adjacency matrix path set between labels, filters the data pair set with associated edge weights, and obtains the label adjacency state pair set. The migration determination submodule extracts the average value of the instruction response cycle, state switching rate and running state dwell time corresponding to the state pair based on the set of adjacent state pairs of the label, and obtains the current wind speed fluctuation amplitude and wind turbine torque response time as superposition factors to calculate the migration intensity value of the state pair. It then determines whether the migration intensity value exceeds the state transition trigger benchmark value. If it does, the state pair is marked as a path to be transferred, and a migration intensity screening result is generated. The path construction submodule, based on the migration intensity filtering results, calls the existing target state number sequence in the state label path dictionary, rearranges the node index order in the target state sequence, connects the path identifier set between the state label and the target state node, establishes the label mapping information of the target state path, and generates the state transition path identifier sequence.

6. The wind turbine safe operation control system according to claim 5, characterized in that, The regulation strategy screening module includes: The instruction extraction submodule obtains the state transition path identifier sequence, calls the policy library instruction index table corresponding to the endpoint state number, filters the control instruction entries bound to the current node, extracts the action type, applicable state number and control target parameter set corresponding to each instruction, and generates a target instruction index set. The parameter verification submodule, based on the target instruction index set, calls the current, wind turbine speed change duration and winding temperature rise limit required by each instruction, and determines whether each operating value is within the threshold range required by the instruction. It then filters out valid instruction items that meet the constraints and obtains the parameter adaptation filtering results. The channel filtering submodule adapts the filtering results based on the parameters, matches the executable channel number in the control path mapping table according to the valid instruction item, determines whether the channel is in an available state and has not triggered the protection restriction logic, removes channel items with action conflicts or execution path conflicts, and generates a control scheme execution list.

7. The wind turbine safe operation control system according to claim 6, characterized in that, The system also includes: The interface execution isolation module determines whether there is an abnormal event where the voltage drop continuously decreases within the instruction cycle and is accompanied by a negative current jump exceeding the short-term drop threshold, based on the control scheme execution list. If the event is met, the interface issuing permission of the corresponding control instruction is frozen and the interface port status is recorded to obtain the control command execution isolation flag. The control command execution isolation marker specifically refers to the interface channel freeze number, the position in the control command waiting list, and the fault status identification label.

8. The wind turbine safe operation control system according to claim 7, characterized in that, The interface execution isolation module includes: The signal extraction submodule, based on the control scheme execution list, calls the bus voltage jump value, current negative slope time slice distribution, and wind turbine speed instantaneous response state corresponding to each channel in the controller according to the execution channel number of the control command, extracts the sampling sequence according to the channel period, constructs the parameter matrix, and generates control signal monitoring information; Based on the control signal monitoring information, the anomaly identification submodule determines whether there is a voltage value decreasing trend at two or more consecutive time points in each record, and whether it is accompanied by a current jump amplitude greater than the short-term drop threshold. It then filters the set of anomaly channel numbers to obtain the anomaly channel index sequence. The permission freeze submodule calls the abnormal channel index sequence, locates the current status in the interface management unit according to the control instruction number corresponding to the marked channel, determines whether the status is in the execution pre-queue, and if so, stops the instruction issuance and records the status label and channel status at the time of freeze, establishes a control channel freeze registration table, and generates a control command execution isolation mark.

Images