A method, system, terminal and medium for adaptive humidity early warning in wind tunnels of hydro-generator units.
By dividing the wind tunnel of the hydro-generator unit into multiple operating states and constructing dynamic alarm thresholds, the false alarm and lag problems of humidity monitoring and early warning in the existing technology are solved, realizing adaptive and highly accurate early warning, reducing the false alarm rate and capturing humidity anomalies in advance, and ensuring the safe operation of the unit.
Patent Information
- Application Number
- CN202611277148.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-21
- Publication Date
- 2026-11-17
AI Technical Summary
In the existing technology, the humidity monitoring and early warning method for wind tunnels of hydro-generator units cannot distinguish between normal transient fluctuations and abnormal defects in humidity under different operating conditions, resulting in frequent false alarms and delayed early warnings. It is unable to capture abnormal change trends in the early stage of rapid humidity rise, and the single humidity sensor has measurement reliability issues.
By acquiring real-time operating data of the hydro-generator unit, the unit's operating status is divided into multiple independent states based on active power and monitoring system process signals. Dynamic alarm thresholds are constructed, including absolute humidity value and rate of increase alarm thresholds. The system automatically switches to the corresponding dynamic alarm threshold according to the real-time status and judges the real-time comprehensive humidity value. The system combines data from multiple humidity sensors for validity verification and fusion processing.
It enables adaptive adjustment of alarm logic under different operating conditions, reduces false alarm rate, detects abnormal humidity trends in advance, improves early warning accuracy, reduces maintenance personnel fatigue, and ensures timely alarm issuance when humidity is abnormal.
Smart Images

Figure CN123112331A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of hydro-generator unit operation monitoring technology, and more specifically, to a method, system, terminal, and medium for adaptive early warning of humidity in a wind tunnel for hydro-generator units. Background Technology
[0002] Humidity inside the wind tunnel of a hydroelectric generator set is a key parameter affecting the insulation performance of the generator stator windings and core. Excessive humidity can cause the insulation materials to absorb moisture and condensation on their surface, which in severe cases can lead to insulation breakdown or short circuit faults, threatening the safe operation of the unit.
[0003] Currently, humidity monitoring and early warning in wind tunnels for hydro-generator units mainly employs a fixed threshold alarm method. This involves deploying one or more humidity sensors within the wind tunnel and setting a fixed humidity alarm threshold. When the humidity data collected by the sensors exceeds this threshold, an alarm signal is triggered. This method is simple in structure and easy to implement, and it played a certain role in early applications. However, hydro-generator units operate under complex and varied conditions in actual operation, encompassing stable operation, load adjustment, and start-up and shutdown. The airflow organization and heat exchange characteristics within the wind tunnel differ significantly under each state, resulting in varying humidity levels and their variation patterns. For example, changes in cooling water system switching and ventilation volume during unit start-up and shutdown can cause significant fluctuations in humidity, while humidity remains relatively stable during stable operation. The fixed threshold alarm method cannot distinguish between normal transient humidity fluctuations and genuine abnormalities under these different operating conditions. This leads to frequent false alarms during start-up, shutdown, and load changes. Maintenance personnel experience alarm fatigue from dealing with a large number of invalid alarms over a long period, making it easier to overlook genuine humidity anomalies. Meanwhile, fixed thresholds only focus on judging the absolute value of humidity exceeding the limit, failing to capture abnormal trends in the early stages of a rapid increase in humidity. This results in a significant lag in early warnings, often issuing alarms only after humidity has already severely exceeded the limit, missing the optimal time for intervention. Furthermore, single humidity sensors suffer from measurement reliability issues; sensor drift, localized airflow dead zones, or temporary condensation can all lead to distorted monitoring data, further increasing the risk of false alarms and missed alarms.
[0004] Therefore, researching and designing an adaptive early warning method, system, terminal, and medium for wind tunnel humidity of hydro-generator units that can overcome the above-mentioned defects is an urgent problem to be solved. Summary of the Invention
[0005] To address the shortcomings of existing technologies, the purpose of this invention is to provide a method, system, terminal, and medium for adaptive humidity early warning in wind tunnels of hydro-generator units, thereby improving early warning accuracy, reducing alarm fatigue among maintenance personnel, achieving advanced early warning of abnormal humidity trends, and ensuring continuous and stable early warning accuracy during long-term operation.
[0006] The above-mentioned technical objective of the present invention is achieved through the following technical solution: Firstly, a humidity adaptive early warning method for wind tunnels of hydro-generator units is provided, including: The real-time operating data of the hydro-generator unit is obtained, including active power, process signals of the monitoring system, and wind tunnel humidity data collected by multiple humidity sensors. Based on the active power and the process signals of the monitoring system, the operating state of the hydro-generator unit is divided into multiple independent states, including at least a stable state, a load increase state, a load decrease state, a start-up process state, and a shutdown process state. Historical operating data of the hydro-generator unit is retrieved, and historical humidity data is classified according to various states. Dynamic alarm thresholds are independently constructed for each state, including absolute humidity alarm thresholds and humidity rise rate alarm thresholds. Based on the real-time determination of the current operating status, automatically switch to the dynamic alarm threshold corresponding to the current operating status; The real-time comprehensive humidity value is judged based on the dynamic alarm threshold, and an early warning signal is issued when the real-time comprehensive humidity value meets the alarm conditions.
[0007] Furthermore, based on the active power and the monitoring system process signals, the operating state of the hydro-generator unit is divided into several independent states, including: When the change in active power per minute is less than a preset threshold value, it is determined to be in a steady state; When the positive change in active power per minute exceeds a preset threshold, it is determined to be the load increase state; When the negative change in active power per minute exceeds a preset threshold, it is determined to be a load reduction state; When the monitoring system process signal indicates that the power-on process has started, it is determined to be in the power-on process state; When the monitoring system process signal indicates that the shutdown process has started, it is determined to be in the shutdown process state.
[0008] Furthermore, the independent construction of dynamic alarm thresholds for each state includes: For each state, the historical humidity data are fitted with a Gaussian distribution to obtain the mean and standard deviation of humidity. Construct a safe humidity range based on the average humidity value; The absolute humidity alarm threshold is set according to the upper limit of the safe humidity range, and the humidity rise rate alarm threshold is set according to the statistical distribution of the humidity change rate within the safe humidity range.
[0009] Furthermore, the determination of the real-time comprehensive humidity value based on the dynamic alarm threshold includes: When the current operating state is the stable state, if the real-time comprehensive humidity value exceeds the absolute humidity alarm threshold and the duration exceeds the first preset duration, a humidity over-limit alarm is issued; if the rate of increase of the real-time comprehensive humidity value exceeds the rate of increase of humidity alarm threshold, a rapid increase in humidity warning is issued. When the current operating state is the increased load state or the decreased load state, if the real-time comprehensive humidity value exceeds the absolute humidity value alarm threshold or the rate of increase of the real-time comprehensive humidity value exceeds the rate of increase of humidity alarm threshold, an alarm will be issued. When the current operating state is the power-on process state or the power-off process state, if the rate of increase of the real-time comprehensive humidity value exceeds the humidity increase rate alarm threshold and the real-time comprehensive humidity value also exceeds the relaxed humidity absolute value alarm threshold, an alarm is triggered.
[0010] Furthermore, the method also includes: The validity of the wind tunnel humidity data collected by multiple humidity sensors is verified, and abnormal measurement points with communication interruptions, data jumps, and no change in values for a long time are eliminated. The remaining valid measuring points are divided into layers according to vertical elevation, and the median value of the valid measuring point values in each layer is taken. The median values of each layer are weighted and summed according to preset weights to obtain the real-time comprehensive humidity value.
[0011] Furthermore, the method also includes: At the end of each month, acquire the newly added valid humidity data for that month, incorporate the newly added valid humidity data for that month into the historical humidity dataset for the corresponding status, and remove the humidity data for the oldest month. Based on the updated historical humidity dataset, Gaussian distribution fitting is re-performed to update the dynamic alarm thresholds for various states.
[0012] Furthermore, the method also includes: When determining the current operating state, a de-jitter delay is introduced; At the moment of state transition, a switching interlock interval is set, and the alarm logic before the transition remains unchanged within the switching interlock interval.
[0013] Secondly, a humidity adaptive early warning system for a hydro-generator wind tunnel is provided, comprising: The data acquisition unit is used to acquire real-time operating data of the hydro-generator unit. The real-time operating data includes active power, monitoring system process signals, and wind tunnel humidity data collected by multiple humidity sensors. The state determination unit is used to divide the operating state of the hydro-generator unit into multiple independent states based on the active power and the process signal of the monitoring system. The multiple states include at least a stable state, a load increase state, a load decrease state, a start-up process state, and a shutdown process state. The threshold construction unit is used to retrieve historical operating data of the hydro-generator unit, classify the historical humidity data according to various states, and independently construct dynamic alarm thresholds for each state. The dynamic alarm thresholds include humidity absolute value alarm thresholds and humidity rise rate alarm thresholds. The logic switching unit is used to automatically switch to the dynamic alarm threshold corresponding to the current operating state based on the real-time determination of the current operating state. The early warning execution unit is used to judge the real-time comprehensive humidity value based on the dynamic alarm threshold, and to issue an early warning signal when the real-time comprehensive humidity value meets the alarm conditions.
[0014] Thirdly, a computer terminal is provided, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement a wind tunnel humidity adaptive early warning method for a hydro-generator set as described in any one of the first aspects.
[0015] Fourthly, a computer-readable medium is provided having a computer program stored thereon, the computer program being executed by a processor to implement a wind tunnel humidity adaptive early warning method for a hydro-generator set as described in any one of the first aspects.
[0016] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention provides a wind tunnel humidity adaptive early warning method for hydro-generator units. It acquires real-time operating data of the hydro-generator unit, including active power, monitoring system process signals, and wind tunnel humidity data collected by multiple humidity sensors. Based on active power and monitoring system process signals, the unit's operating state is divided into several independent states, such as stable state, increased load state, decreased load state, start-up process state, and shutdown process state. Historical operating data is retrieved and categorized according to different states. For each state, a dynamic alarm threshold is independently constructed, including an absolute humidity alarm threshold and a humidity rise rate alarm threshold. The method automatically switches to the corresponding dynamic alarm threshold based on the real-time determined operating state. Finally, the real-time comprehensive humidity value is judged based on the dynamic alarm threshold, and an early warning signal is issued. The state classification results of active power and process signals in this invention provide accurate labeling basis for the classification and statistics of historical data, while the state-based statistics of historical data provide an objective statistical basis for the construction of dynamic thresholds. The real-time state judgment and the automatic switching of dynamic thresholds form a closed-loop adaptive alarm mechanism. More importantly, the absolute humidity alarm threshold and the humidity rise rate alarm threshold work together. The absolute value threshold is used to determine whether the humidity has exceeded the safe range, while the rate threshold is used to detect rapid increases in humidity before it exceeds the limit. Together, they form a three-dimensional early warning system. Since the normal fluctuation range and rate of change of humidity vary significantly under different operating conditions, this invention uses state-specific dynamic thresholds to enable the alarm logic to adaptively adjust to changes in the group's operating state. This effectively distinguishes normal physical transients from genuine abnormal defects during start-up, shutdown, and load changes, significantly reducing the false alarm rate. Simultaneously, the advanced warning through the rate threshold can issue an alarm before the absolute humidity value exceeds the limit, buying valuable time for maintenance personnel to respond.
[0017] 2. This invention determines steady-state, load-increasing, and load-reducing states by comparing the change in active power per minute with a preset threshold value. It directly determines the start-up and shutdown process states through monitoring system process signals. Simultaneously, it introduces anti-jitter delay during state determination and sets a switching interlocking interval at the moment of state transition. The active power change rate threshold value in this invention is set according to the unit's rated power, adapting to the speed regulation characteristics of units with different capacities. Zero-delay recognition of process signals ensures the immediacy of start-up and shutdown transient state determination; the two complement each other, covering all operating conditions. The anti-jitter delay effectively filters out instantaneous state jumps caused by small grid fluctuations and power sampling noise by waiting for a period of time before confirming the state transition. The switching interlocking interval maintains the alarm logic of the previous state unchanged for a short period after the formal state transition, avoiding false alarms caused by excessive differences between the old and new thresholds at the moment of transition.
[0018] 3. This invention performs Gaussian distribution fitting on historical humidity data for each state to obtain the mean and standard deviation of humidity. The interval formed by adding or subtracting three times the standard deviation from the mean humidity is used as the safe humidity interval. An absolute humidity alarm threshold is set based on the upper limit of this interval, and a humidity rise rate alarm threshold is set based on the statistical distribution of humidity change rate. Simultaneously, at the end of each month, newly added valid humidity data is added to the historical dataset for the corresponding state, the oldest month's data is removed, and Gaussian distribution fitting is re-executed based on the updated dataset to update the dynamic alarm thresholds for each state. Since the humidity distribution characteristics differ under different states—for example, the humidity distribution is concentrated and the standard deviation is small in the steady state, while the humidity distribution is dispersed and the standard deviation is large in the startup process state—the Gaussian fitting method can automatically adapt to the inherent fluctuation characteristics of each state. The iterative update mechanism ensures that the thresholds do not gradually become invalid due to environmental changes. This invention achieves objectivity, adaptability, and long-term stability in threshold setting, eliminating the need for manual parameter modification by maintenance personnel.
[0019] 4. This invention verifies the validity of wind tunnel humidity data collected by multiple humidity sensors, eliminating abnormal measurement points with communication interruptions, data jumps, and prolonged periods of unchanged values. The median value is taken by vertical elevation layer, and then the medians of each layer are weighted and summed to obtain the real-time comprehensive humidity value. Based on this, differentiated alarm criteria are designed for different operating states: a two-dimensional strategy is used for steady state, where an alarm is only triggered when the absolute value exceeds the limit for a certain period, and an immediate warning is given when the rate exceeds the limit; an OR gate strategy is used for load increase and load decrease states, where an alarm is triggered when either the absolute value or the rate exceeds the limit; and an AND gate strategy is used for start-up and shutdown states, where an alarm is only triggered when both the rate and the absolute value exceed the limit. This invention enables the early warning system to maintain a sensitive response to real anomalies while minimizing false alarms caused by normal fluctuations in each state. Attached Figure Description
[0020] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings: Figure 1 This is a flowchart from Embodiment 1 of the present invention; Figure 2 This is a system block diagram in Embodiment 2 of the present invention. Detailed Implementation
[0021] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.
[0022] Example 1: An adaptive humidity early warning method for wind tunnels of hydro-generator units, such as... Figure 1 As shown, it includes: S1: Acquire real-time operating data of the hydro-generator unit. The real-time operating data includes active power, process signals of the monitoring system, and wind tunnel humidity data collected by multiple humidity sensors. S2: Based on active power and monitoring system process signals, the operating state of the hydro-generator unit is divided into multiple independent states, including at least the steady state, load increase state, load decrease state, start-up process state and shutdown process state; S3: Retrieve historical operating data of the hydro-generator unit, classify historical humidity data according to multiple states, and independently construct dynamic alarm thresholds for each state. Dynamic alarm thresholds include absolute humidity alarm thresholds and humidity rise rate alarm thresholds. S4: Based on the real-time determination of the current operating status, automatically switch to the dynamic alarm threshold corresponding to the current operating status; S5: Based on the dynamic alarm threshold, the real-time comprehensive humidity value is judged, and an early warning signal is issued when the real-time comprehensive humidity value meets the alarm conditions.
[0023] In step S1, humidity data in the wind tunnel is first collected in real time through a distributed humidity sensor network deployed within the turbine generator unit's wind tunnel. This distributed humidity sensor network consists of multiple humidity sensors evenly distributed along the circumference of the wind tunnel, and measurement points are arranged in layers along the vertical elevation direction, forming a multi-source humidity acquisition array covering the entire wind tunnel space. Each humidity sensor uses an industrial-grade capacitive humidity probe with a measurement accuracy of 2% relative humidity and a sampling frequency of 10Hz to ensure the capture of transient characteristics of rapidly changing humidity. The collected raw humidity data is transmitted to a data acquisition card via shielded cables and then stored in a real-time database after analog-to-digital conversion.
[0024] Simultaneously, real-time active power data of the hydro-generator unit is acquired through the unit's LCU cabinet. The active power data is output from the unit's power transmitter, converted into a digital signal by the analog acquisition module of the LCU cabinet, and the sampling frequency is kept consistent with the humidity data, set to 10Hz, to facilitate synchronous analysis in subsequent status determination.
[0025] In addition, the unit's start-up and shutdown process signals are acquired from the monitoring system's process bus. Start-up process signals are generated at key nodes such as the issuance of the unit start-up command, guide vane opening, and excitation activation. Shutdown process signals are generated at nodes such as unit disconnection, demagnetization, and brake activation. Both types of process signals are switch signals, connected to the data acquisition system via hardwiring. The signal level is DC 24V; a high level indicates the process is in progress, and a low level indicates the process is not activated.
[0026] In summary, the real-time operational data includes: multiple wind tunnel humidity data sequences from a distributed humidity sensor network, denoted as... Where the subscript i represents the sensor number, j represents the vertical elevation layer number, and t is the sampling time; the active power data sequence from the unit's LCU cabinet is denoted as... ; and process signals from the monitoring system, including power-on process flags. and shutdown procedure signs All data are Boolean variables. All data are aligned on the timeline to form a unified real-time data frame with a frame period of 0.1s.
[0027] In step S2, the five states are defined according to the mutual exclusion principle, that is, the unit can only be in one of the states at any given time, and the state switch must meet the triggering conditions.
[0028] First, calculate the change in active power per minute. The calculation formula is as follows: ; in, For the current moment Active power, measured in megawatts; This represents the active power value in the 60 seconds preceding the current moment.
[0029] Steady state determination: When the change in active power per minute... The absolute value is less than the preset threshold value If this condition is met continuously for 3 seconds (i.e., a 3-second anti-shake delay is introduced), then the unit is determined to be in a stable state. Preset threshold value. Take the rated power of the unit 15%. This threshold value is set based on the unit capacity and speed control system response characteristics, and can effectively filter out misjudgments caused by small fluctuations in the power grid. For example, for a unit with a rated power of 300MW, .
[0030] Determination of increased load state: When the active power changes positively per minute If this condition is met continuously for 3 seconds, it is determined to be a load increase state. A positive change refers to the direction in which the active power at the current moment increases compared to the previous minute; a positive value indicates that the unit is increasing its output. The 3-second anti-jitter delay is used to avoid instantaneous misjudgments caused by power sampling noise or brief disturbances.
[0031] Determination of load reduction state: When the active power changes negatively per minute If this condition is met continuously for 3 seconds, the system is considered to be in a load reduction state. A negative change indicates that the unit is reducing its output. The judgment logic is symmetrical to that of the load increase state, and the threshold values are the same.
[0032] Determining the power-on process state: When the monitoring system process signal... When the signal changes from low to high, it is directly identified as the power-on process state, without introducing anti-jitter delay. This is because the power-on process signal is generated by the monitoring system based on key nodes such as guide vane opening and excitation activation, requiring no additional filtering. This determination achieves zero-delay identification of the power-on transient state, ensuring that the corresponding alarm logic can be switched at the initial stage of unit startup.
[0033] Determining the shutdown process state: When the monitoring system process signal... When the signal changes from low to high, it is directly identified as the shutdown process state, and no anti-jitter delay is introduced. The shutdown process signal is generated at nodes such as unit disconnection, demagnetization, and brake activation, and its rising edge marks the start of the shutdown process. Zero-delay recognition ensures accurate capture of rapid humidity changes during the shutdown process.
[0034] The priority order of the five states is as follows: start-up and shutdown processes take precedence over load increase, load decrease, and steady state. That is, when a start-up or shutdown process signal is valid, regardless of the change in active power, it is preferentially determined to be in the start-up or shutdown process state. This priority design ensures that special alarm logic during the unit's start-up and shutdown transients is not mistakenly overwritten by power changes. When there is no process signal and the power change does not exceed the threshold, the steady state is maintained; when the power change exceeds the threshold and there is no process signal, it is determined to be in the load increase or load decrease state according to the direction of the power change. To avoid logic jitter during state transitions, a switching interlocking interval is set.
[0035] In step S3, historical operating data of the hydro-generator unit for at least 12 consecutive months is retrieved. This historical operating data includes historical active power, historical monitoring system process signals, and historical wind tunnel humidity data. First, according to a five-state classification rule, the historical humidity data is categorized according to the corresponding operating state, forming five independent historical humidity datasets. Each dataset contains only the comprehensive humidity value sequence of all valid humidity sensors under that state, excluding data from periods outside that state and abnormal data.
[0036] For each historical humidity dataset under each condition, it is assumed that the humidity values follow a Gaussian distribution, i.e., a normal distribution. A Gaussian distribution is fitted to this dataset, and the average value of the historical humidity data under that condition is calculated. and standard deviation , where the subscript s indicates the state type.
[0037] The average value of state s The calculation formula is: ; in, This represents the total number of historical humidity data points corresponding to state s. is the comprehensive humidity value of the kth sample point in the historical humidity data corresponding to state s.
[0038] Standard deviation The calculation formula is: .
[0039] Based on average and three standard deviations Construct a safe humidity range for this state. The lower limit of the safe humidity range is... The upper limit is Based on the properties of the Gaussian distribution, this range theoretically covers approximately 99.7% of normal operating humidity data. Humidity values outside this range are considered low-probability events, corresponding to potential abnormal operating conditions. The safe humidity range is expressed as: .
[0040] Set the absolute humidity alarm threshold for this state based on the upper limit of the safe humidity range. The absolute humidity alarm threshold is equal to the upper limit of the safe humidity range, that is: When the real-time comprehensive humidity value exceeds this threshold, it indicates that the humidity has deviated from the normal statistical range and an alarm needs to be triggered. For the start-up and shutdown processes, considering the physical characteristic that humidity will inevitably fluctuate significantly during start-up and shutdown, the absolute humidity alarm threshold can be appropriately relaxed. The relaxation factor is set to 1.2 times based on actual unit operating experience, meaning the relaxed threshold is... .
[0041] Based on the statistical distribution of humidity change rate within the safe humidity range, set the humidity rise rate alarm threshold for this state. The specific method is as follows: calculate the humidity change rate sequence for adjacent sampling points in the historical humidity dataset. The calculation formula is: ; in, The sampling time interval is defined. Then, the change rate sequence is statistically analyzed, and its average value is calculated. and standard deviation .
[0042] The alarm threshold for the rate of increase in humidity is set as follows: ; This threshold represents the normal upper limit of the humidity rise rate under this condition; exceeding this value indicates that the humidity is rising abnormally rapidly. For the start-up and shutdown processes, since the humidity change rate is naturally larger during start-up and shutdown, a relaxation factor of 1.2 is also used for adjustment.
[0043] Through the above steps, a set of dynamic alarm thresholds were independently constructed for steady state, load increase state, load decrease state, start-up process state, and shutdown process state. Each set of thresholds includes an absolute humidity alarm threshold and a humidity rise rate alarm threshold. These thresholds are derived from long-term operating data of the unit and can adapt to the normal humidity fluctuation range under different conditions, fundamentally solving the problem that fixed thresholds cannot distinguish between normal transients and abnormal defects.
[0044] In step S4, firstly, the system maintains a status register. This is used to store the current valid running state. After each state determination, a temporary state value is output. .Compare and Are they consistent? If they are consistent, the current alarm threshold group remains unchanged; if they are inconsistent, a state switching procedure is triggered.
[0045] Anti-jitter delay mechanism: To prevent frequent state jumps caused by active power sampling noise or instantaneous grid disturbances, an anti-jitter delay is introduced in the state determination stage. Specifically, when... First time with At the same time, start a timer with a duration of 3 seconds. Within 3 seconds, if... If it remains constant, it will reach its destination in 3 seconds. Write The state switch is officially confirmed; if it occurs within 3 seconds... If the situation changes again, the timer is reset and restarted until a certain state remains stable for a full 3 seconds. This anti-jitter delay effectively filters out instantaneous state fluctuations lasting less than 3 seconds, such as the recovery of active power after a brief overshoot caused by grid frequency adjustments.
[0046] Switching Lockout Interval Mechanism: To avoid momentary false alarms due to excessive differences between old and new thresholds at the instant of formal state transition, a switching lockout interval is set. Specifically, when... When a change occurs, a timer is immediately started with a duration of 2 seconds. During this 2-second switching interlocking period, the alarm logic continues to use the dynamic alarm threshold group corresponding to the previous state, maintaining the absolute humidity alarm threshold and humidity rise rate alarm threshold unchanged. Only after the timer expires is the new dynamic alarm threshold group corresponding to the new state officially activated for subsequent alarm judgments. This mechanism ensures a smooth transition of alarm logic during state switching. For example, when switching from a stable state to an increased load state, the absolute humidity alarm threshold in the stable state is usually low. An immediate switch might cause the normal humidity rise in the initial stage of the increased load state to be misjudged as abnormal.
[0047] In step S5, the wind tunnel humidity data collected by multiple humidity sensors are first fused and processed to calculate the real-time comprehensive humidity value. Then, a warning judgment is made based on the dynamic alarm threshold corresponding to the current operating state. The dynamic alarm threshold of each state is automatically iterated and updated at the end of each month.
[0048] The validity of the raw humidity values collected by each humidity sensor at time t is verified. Three types of abnormal measurement points are eliminated during verification: first, null or timed-out values due to communication interruptions; second, abnormal jump data where the difference between the current and adjacent sampling times exceeds a preset jump threshold (set to 10%RH based on the sensor's range); and third, data with no change for an extended period, i.e., suspected dead values where the reading change is less than 0.1%RH over 300 seconds. After verification, the set of valid humidity values from each layer is retained.
[0049] Then, the median value of the effective measuring points is taken for each layer according to vertical elevation. Let the set of effective measuring point humidity values for the j-th layer at time t be . ,in, The number of effective measurement points in layer j. The inter-layer median of this layer. The calculation formula is: ; Number of valid measurement points within the middle filtering window When the number is odd, take the first number after sorting. A value; when When the value is even, the arithmetic mean of the two middle values after sorting is taken. This median filter effectively suppresses impulse noise caused by individual sensor failures or local airflow disturbances within a single layer.
[0050] Finally, the medians of each layer are summed according to preset weights to obtain the real-time comprehensive humidity value at time t. The weighted summation formula is: ; in, This represents the total number of vertical elevation layers. Let be the weight coefficient of the j-th layer, satisfying The weighting coefficients are preset based on the degree of influence of each layer on the sensitivity of the stator end insulation: the weight of the upper layer is 0.4, the weight of the middle layer is 0.3, and the weight of the lower layer is 0.3. When all measuring points in a certain layer are marked as invalid, the weight of that layer is proportionally distributed to the remaining valid layers to ensure that the total weight is 1.
[0051] The aforementioned composite algorithm transforms raw data from multiple humidity sensors distributed along the circumference and vertical elevation of the wind tunnel into a real-time comprehensive humidity value that accurately reflects the overall humidity level of the wind tunnel. .
[0052] When the current operating state is determined to be stable, the system simultaneously monitors both the absolute level and the rate of increase of the real-time comprehensive humidity value. If the real-time comprehensive humidity value exceeds the humidity absolute value alarm threshold corresponding to this stable state... If the excessive humidity condition persists for more than a preset duration of 5 seconds, it is determined to be a humidity exceeding the limit, and a humidity exceeding the limit alarm signal is issued. If the rate of increase of the real-time comprehensive humidity value exceeds the humidity increase rate alarm threshold corresponding to the steady state, further action will be taken. If the humidity is abnormally high, an early warning signal for rapid humidity increase will be issued immediately without waiting for the duration.
[0053] When the current operating state is either load increase or load decrease, an OR gate is used for judgment. The alarm will be triggered if the real-time comprehensive humidity value exceeds the humidity absolute value alarm threshold corresponding to that state. or Or, the rate of increase of the real-time comprehensive humidity value exceeds the humidity increase rate alarm threshold corresponding to this state. or An alarm will be triggered if either of the two conditions is met. Under varying load conditions, the requirement that the absolute value exceedance last for 5 seconds is no longer required, in order to speed up the response.
[0054] When the current operating state is either the power-on or power-off state, an AND gate-based judgment strategy is employed. The requirement is that the rate of increase of the real-time comprehensive humidity value exceeds the humidity increase rate alarm threshold corresponding to that state. or Furthermore, the real-time comprehensive humidity value simultaneously exceeds the relaxed absolute humidity alarm threshold. The relaxed absolute humidity alarm threshold is obtained by multiplying the set base threshold by a relaxation factor of 1.2. An alarm is triggered only when both the rate and absolute humidity conditions are met simultaneously.
[0055] Automatic threshold updates are performed monthly at the end of each month. First, all newly added valid humidity data for the month is retrieved from the database; this data has already undergone validity verification. Then, according to the state classification rules, the newly added valid humidity data is assigned to the corresponding historical humidity dataset for each state. Simultaneously, the oldest month's humidity data is removed from each state's historical humidity dataset, maintaining a 12-month sliding window length for the historical dataset. Finally, based on the updated historical humidity dataset, the Gaussian distribution fitting process is re-executed to recalculate the average humidity and standard deviation of the historical humidity data for each state. This reconstructs the safe humidity range and updates the absolute humidity alarm threshold and the humidity rise rate alarm threshold. The updated thresholds automatically take effect at midnight on the first day of the following month.
[0056] Example 2: A humidity adaptive early warning system for a hydro-generator wind tunnel, which implements the humidity adaptive early warning method for a hydro-generator wind tunnel described in Example 1, such as... Figure 2As shown, it includes a data acquisition unit, a status determination unit, a threshold construction unit, a logic switching unit, and an early warning execution unit.
[0057] The system includes the following components: a data acquisition unit for acquiring real-time operating data of the hydro-generator unit, including active power, monitoring system process signals, and wind tunnel humidity data collected by multiple humidity sensors; a state determination unit for classifying the operating state of the hydro-generator unit into multiple independent states based on active power and monitoring system process signals, including at least a steady state, a load increase state, a load decrease state, a start-up process state, and a shutdown process state; a threshold construction unit for retrieving historical operating data of the hydro-generator unit, classifying historical humidity data according to multiple states, and independently constructing dynamic alarm thresholds for each state, including absolute humidity alarm thresholds and humidity rise rate alarm thresholds; a logic switching unit for automatically switching to the dynamic alarm threshold corresponding to the current operating state based on the real-time determination; and an early warning execution unit for judging the real-time comprehensive humidity value based on the dynamic alarm thresholds and issuing an early warning signal when the real-time comprehensive humidity value meets the alarm conditions.
[0058] Example 3: A computer terminal includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements a wind tunnel humidity adaptive early warning method for a hydro-generator set as described in Example 1.
[0059] Example 4: A computer-readable medium having a computer program stored thereon, the computer program being executed by a processor to implement a wind tunnel humidity adaptive early warning method for a hydro-generator set as described in Example 1.
[0060] Working Principle: This invention acquires real-time operating data of a hydro-generator unit, including active power, monitoring system process signals, and wind tunnel humidity data collected by multiple humidity sensors. Based on the active power and monitoring system process signals, the unit's operating state is divided into several independent states, such as steady state, increased load state, decreased load state, start-up process state, and shutdown process state. Historical operating data is retrieved and categorized according to different states. For each state, a dynamic alarm threshold is independently constructed, including an absolute humidity alarm threshold and a humidity rise rate alarm threshold. The current operating state is then automatically switched to the corresponding dynamic alarm threshold based on the real-time determination. Finally, the real-time comprehensive humidity value is judged based on the dynamic alarm threshold, and an early warning signal is issued. The state classification results of active power and process signals in this invention provide accurate labeling basis for the classification and statistics of historical data, while the state-based statistics of historical data provide an objective statistical basis for the construction of dynamic thresholds. The real-time state determination and the automatic switching of dynamic thresholds form a closed-loop adaptive alarm mechanism. More importantly, the absolute humidity alarm threshold and the humidity rise rate alarm threshold work together. The absolute value threshold is used to determine whether the humidity has exceeded the safe range, while the rate threshold is used to detect rapid increases in humidity before it exceeds the limit. Together, they form a three-dimensional early warning system. Since the normal fluctuation range and rate of change of humidity vary significantly under different operating conditions, this invention uses state-specific dynamic thresholds to enable the alarm logic to adaptively adjust to changes in the group's operating state. This effectively distinguishes normal physical transients from genuine abnormal defects during start-up, shutdown, and load changes, significantly reducing the false alarm rate. Simultaneously, the advanced warning through the rate threshold can issue an alarm before the absolute humidity value exceeds the limit, buying valuable time for maintenance personnel to respond.
[0061] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.
[0062] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.
[0063] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.
[0064] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.
[0065] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.
Claims
1. A humidity adaptive early warning method for wind tunnels of hydro-generator units, characterized in that, include: The real-time operating data of the hydro-generator unit is obtained, including active power, process signals of the monitoring system, and wind tunnel humidity data collected by multiple humidity sensors. Based on the active power and the process signals of the monitoring system, the operating state of the hydro-generator unit is divided into multiple independent states, including at least a stable state, a load increase state, a load decrease state, a start-up process state, and a shutdown process state. Historical operating data of the hydro-generator unit is retrieved, and historical humidity data is classified according to various states. Dynamic alarm thresholds are independently constructed for each state, including absolute humidity alarm thresholds and humidity rise rate alarm thresholds. Based on the real-time determination of the current operating status, automatically switch to the dynamic alarm threshold corresponding to the current operating status; The real-time comprehensive humidity value is judged based on the dynamic alarm threshold, and an early warning signal is issued when the real-time comprehensive humidity value meets the alarm conditions.
2. The adaptive humidity early warning method for a wind tunnel of a hydro-generator unit according to claim 1, characterized in that, Based on the active power and the monitoring system process signals, the operating state of the hydro-generator unit is divided into several independent states, including: When the change in active power per minute is less than a preset threshold value, it is determined to be in a steady state; When the positive change in active power per minute exceeds a preset threshold, it is determined to be the load increase state; When the negative change in active power per minute exceeds a preset threshold, it is determined to be a load reduction state; When the monitoring system process signal indicates that the power-on process has started, it is determined to be in the power-on process state; When the monitoring system process signal indicates that the shutdown process has started, it is determined to be in the shutdown process state.
3. The adaptive humidity early warning method for a wind tunnel of a hydro-generator unit according to claim 1, characterized in that, The method of independently constructing dynamic alarm thresholds for each state includes: For each state, the historical humidity data are fitted with a Gaussian distribution to obtain the mean and standard deviation of humidity. Construct a safe humidity range based on the average humidity value; The absolute humidity alarm threshold is set according to the upper limit of the safe humidity range, and the humidity rise rate alarm threshold is set according to the statistical distribution of the humidity change rate within the safe humidity range.
4. The adaptive humidity early warning method for a wind tunnel of a hydro-generator unit according to claim 1, characterized in that, The determination of the real-time comprehensive humidity value based on the dynamic alarm threshold includes: When the current operating state is the stable state, if the real-time comprehensive humidity value exceeds the absolute humidity alarm threshold and the duration exceeds the first preset duration, a humidity over-limit alarm is issued; if the rate of increase of the real-time comprehensive humidity value exceeds the rate of increase of humidity alarm threshold, a rapid increase in humidity warning is issued. When the current operating state is the increased load state or the decreased load state, if the real-time comprehensive humidity value exceeds the absolute humidity value alarm threshold or the rate of increase of the real-time comprehensive humidity value exceeds the rate of increase of humidity alarm threshold, an alarm will be issued. When the current operating state is the power-on process state or the power-off process state, if the rate of increase of the real-time comprehensive humidity value exceeds the humidity increase rate alarm threshold and the real-time comprehensive humidity value also exceeds the relaxed humidity absolute value alarm threshold, an alarm is triggered.
5. The adaptive humidity early warning method for a wind tunnel of a hydro-generator unit according to claim 1, characterized in that, The method also includes: The validity of the wind tunnel humidity data collected by multiple humidity sensors is verified, and abnormal measurement points with communication interruptions, data jumps, and no change in values for a long time are eliminated. The remaining valid measuring points are divided into layers according to vertical elevation, and the median value of the valid measuring point values in each layer is taken. The median values of each layer are weighted and summed according to preset weights to obtain the real-time comprehensive humidity value.
6. The adaptive humidity early warning method for a wind tunnel of a hydro-generator unit according to claim 1, characterized in that, The method also includes: At the end of each month, acquire the newly added valid humidity data for that month, incorporate the newly added valid humidity data for that month into the historical humidity dataset for the corresponding status, and remove the humidity data for the oldest month. Based on the updated historical humidity dataset, Gaussian distribution fitting is re-performed to update the dynamic alarm thresholds for various states.
7. The adaptive humidity early warning method for a wind tunnel of a hydro-generator unit according to claim 1, characterized in that, The method also includes: When determining the current operating state, a de-jitter delay is introduced; At the moment of state transition, a switching interlock interval is set, and the alarm logic before the transition remains unchanged within the switching interlock interval.
8. A humidity adaptive early warning system for a wind tunnel of a hydro-generator unit, characterized in that, include: The data acquisition unit is used to acquire real-time operating data of the hydro-generator unit. The real-time operating data includes active power, monitoring system process signals, and wind tunnel humidity data collected by multiple humidity sensors. The state determination unit is used to divide the operating state of the hydro-generator unit into multiple independent states based on the active power and the process signal of the monitoring system. The multiple states include at least a stable state, a load increase state, a load decrease state, a start-up process state, and a shutdown process state. The threshold construction unit is used to retrieve historical operating data of the hydro-generator unit, classify the historical humidity data according to various states, and independently construct dynamic alarm thresholds for each state. The dynamic alarm thresholds include humidity absolute value alarm thresholds and humidity rise rate alarm thresholds. The logic switching unit is used to automatically switch to the dynamic alarm threshold corresponding to the current operating state based on the real-time determination of the current operating state. The early warning execution unit is used to judge the real-time comprehensive humidity value based on the dynamic alarm threshold, and to issue an early warning signal when the real-time comprehensive humidity value meets the alarm conditions.
9. A computer terminal, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements a wind tunnel humidity adaptive early warning method for a hydro-generator set as described in any one of claims 1-7.
10. A computer-readable medium having a computer program stored thereon, characterized in that, The computer program, when executed by a processor, can implement a wind tunnel humidity adaptive early warning method for a hydro-generator set as described in any one of claims 1-7.
