Intelligent dynamic threshold early warning and diagnosis system and method based on nuclear power plant operation

Abstract

The application belongs to the field of equipment operation state monitoring, and relates to an intelligent dynamic threshold early warning and diagnosis system and method based on nuclear power equipment operation, comprising: a data acquisition and processing module, which acquires operation load, rotating speed and environmental temperature in real time and performs pretreatment; a feature extraction module, which extracts equipment operation related features from the pretreated load and rotating speed; load evaluation data and rotating speed evaluation data are extracted by using a machine learning feature selection algorithm; a dynamic threshold calculation module, which builds a first early warning model and a second early warning model, and calculates dynamic load early warning thresholds and dynamic rotating speed early warning thresholds; an early warning judgment module, which compares real-time load and rotating speed with corresponding early warning thresholds, and issues an early warning signal when real-time load or rotating speed exceeds the corresponding early warning threshold; an abnormal diagnosis and adjustment module, which executes an adjustment strategy and corresponding actions under the condition of receiving an early warning signal. The application improves the accuracy and reliability of nuclear power equipment operation monitoring.

Description

Technical Field

[0001] This invention belongs to the field of equipment operation status monitoring technology, specifically relating to an intelligent dynamic threshold early warning and diagnosis system and method based on the operation of nuclear power equipment. Background Technology

[0002] Equipment operation status monitoring refers to the real-time tracking and analysis of equipment operation status through a series of technical means to ensure normal equipment operation and timely detection of potential faults. The existing process mainly includes the following steps: First, real-time collection of equipment operation data, such as temperature, pressure, and vibration, is achieved through hardware devices like sensors. Second, the collected data is transmitted to a data processing center via wired or wireless communication. Next, the data processing center processes and analyzes the received data to assess the equipment's operating status. Finally, the processed data is displayed graphically, and an alarm signal is issued when equipment malfunctions, notifying relevant personnel for handling. Typically, the equipment's status indicators are compared with fixed thresholds to assess its operating status. However, comparing calculated status indicators with dynamic thresholds can significantly increase the error in the comparison results, as both sets of data are dynamic and subject to certain errors, raising concerns about the accuracy of the status assessment.

[0003] In existing technologies, equipment operation status monitoring methods include: acquiring historical and current status data of the equipment; calculating a predicted value for the current status data based on the historical status data; determining the equipment's operation status by comparing the current status data with the predicted value; and sending an early warning signal to the equipment based on its operation status. However, determining the equipment's operation status based on predicted and status values ​​has poor real-time performance, and the factors considered when deriving the predicted value are not comprehensive enough, leading to low accuracy in subsequent assessments of the equipment's status.

[0004] In summary, existing equipment condition monitoring and early warning systems primarily focus on data acquisition and preprocessing, but their in-depth data mining is insufficient. Early warning thresholds are typically fixed or based on historical data, lacking dynamic adjustment capabilities. Furthermore, they consider fewer factors during prediction, impacting the accuracy of fault prediction results. For example, in nuclear power equipment, components such as main pumps and charging pumps are susceptible to various factors during operation, such as load changes and speed fluctuations. While current technologies can acquire, preprocess, and set thresholds for these parameters in real time, practical applications may reveal that the failure rate of these components is related to specific parameter combinations that are not adequately considered in current solutions, thus affecting the accuracy of fault prediction results. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent dynamic threshold early warning and diagnosis system and method based on the operation of nuclear power equipment. This system not only improves the accuracy and reliability of nuclear power equipment operation monitoring, but also realizes the intelligent setting of early warning thresholds. This helps to detect potential problems in equipment operation in a timely manner, prevent the occurrence of faults, and thus ensure the safe and stable operation of nuclear power equipment. It can effectively solve the problems existing in the prior art.

[0006] Technical solution to achieve the purpose of this invention:

[0007] An intelligent dynamic threshold early warning and diagnostic system based on the operation of nuclear power equipment includes:

[0008] The data acquisition and processing module, under normal operating conditions, collects real-time data on the nuclear power equipment's operating load and rotational speed as operational data, and real-time data on ambient temperature as auxiliary data. It then preprocesses the operational and auxiliary data to obtain preprocessed operational and auxiliary data.

[0009] The feature extraction module extracts equipment operation-related features from preprocessed operating data within a preset time period M before scheduled maintenance shutdown; it then uses a machine learning feature selection algorithm to extract key features from these equipment operation-related features. The key features include load assessment data and rotational speed assessment data.

[0010] The dynamic threshold calculation module, within the current preset time period M, builds a first early warning model based on the load assessment data, the obtained equipment failure rate and auxiliary data, and calculates the dynamic load early warning threshold; based on the speed assessment data, the obtained equipment failure rate and auxiliary data, it builds a second early warning model and calculates the dynamic speed early warning threshold.

[0011] The early warning judgment module compares the real-time load and speed of the current nuclear power equipment with the corresponding early warning threshold within the current preset time period M. If the real-time load or speed exceeds the corresponding early warning threshold, an early warning signal is issued.

[0012] The anomaly diagnosis and adjustment module, upon receiving a warning signal, triggers an analysis and diagnosis mechanism to diagnose the cause of the fault and execute an adjustment strategy. It reduces the load or speed according to the set adjustment gradient and monitors the changes in the corresponding load or speed after each execution of the adjustment strategy, and then performs the corresponding action.

[0013] Furthermore, in the data acquisition and processing module, the real-time acquired nuclear power equipment operating load and speed are obtained through sensors arranged on the nuclear power equipment, including load sensors, speed sensors, and temperature sensors; the preprocessing process for operating data and auxiliary data is as follows: data filtering algorithms and outlier detection algorithms are applied to the operating data and auxiliary data to perform cleaning, noise reduction, and normalization processing.

[0014] Furthermore, in the feature extraction module, the load evaluation data includes the mean and variance of the load; the rotational speed evaluation data includes the mean and variance of the rotational speed.

[0015] Furthermore, in the feature extraction module, the machine learning feature selection algorithm includes either Lasso regression or random forest.

[0016] Furthermore, in the feature extraction module, the operating data of the nuclear power equipment under normal operating conditions collected within the previous preset time period T during the shutdown maintenance period is used as the basis, and the extracted key features are used as the load assessment data and speed assessment data for the next preset time period T; M > 0 and is a positive integer.

[0017] Furthermore, in the dynamic threshold calculation module, the calculation formula for the load warning threshold in the first warning model is as follows:

[0018]

[0019] In the formula, Lr represents the load warning threshold, μ represents the mean load within the previous preset time period M, σ represents the variance of the load within the previous preset time period M, r represents the equipment failure rate before this, Pt represents auxiliary data, k1 and k2 represent adjustment coefficients, and the values ​​of k1 and k2 are both in the range of [0, 1].

[0020] The formula for calculating the speed warning threshold in the second early warning model is:

[0021]

[0022] In the formula, Sr represents the speed warning threshold, α represents the mean speed within the previous preset time period M, β represents the variance of the speed within the previous preset time period M, k3 and k4 represent adjustment coefficients, and the values ​​of k3 and k4 are both in the range of [0, 1].

[0023] Furthermore, in the warning judgment module, if the real-time load or rotational speed exceeds the corresponding warning threshold, a warning signal is issued. Specifically, when the load exceeds the load warning threshold, a first-level warning signal is issued; when the rotational speed exceeds the rotational speed warning threshold, a second-level warning signal is issued; and when both the load and rotational speed exceed the corresponding warning thresholds, a third-level warning signal is issued.

[0024] Furthermore, the anomaly diagnosis and adjustment module monitors changes in the corresponding load or rotational speed and executes corresponding actions, specifically:

[0025] If the difference between the load or speed and the corresponding warning threshold shows a decreasing trend, it indicates that the adjustment strategy is effective. Continue to implement the adjustment strategy and continue to monitor the changes in the corresponding load or speed until the difference between the current load or speed and the corresponding warning threshold exceeds the previous difference, then stop.

[0026] Stop when the difference between the load or speed and the corresponding warning threshold shows an increasing trend;

[0027] If the load or speed exceeds the corresponding warning threshold when the machine stops, maintenance will be performed.

[0028] If the load or rotational speed does not exceed the corresponding warning threshold, maintain the normal operation of the nuclear power equipment and record it;

[0029] The difference between the load or speed and the corresponding warning threshold is: load or speed - corresponding warning threshold.

[0030] A method for intelligent dynamic threshold early warning and diagnosis based on the operation of nuclear power equipment, comprising:

[0031] S1. Under normal operating conditions of nuclear power equipment, load and rotation speed are collected in real time as operating data, and ambient temperature is collected in real time as auxiliary data.

[0032] S2. Preprocess the operational data and auxiliary data;

[0033] S3. Within a preset time period M before the scheduled shutdown maintenance, extract equipment operation-related features from the preprocessed operation data, and use a machine learning feature selection algorithm to extract key features from the equipment operation-related features; key features include: load assessment data and speed assessment data.

[0034] S4. Within the current preset time period M, based on the load assessment data, the obtained equipment failure rate and auxiliary data, build a first early warning model and calculate the dynamic load early warning threshold; based on the speed assessment data, the obtained equipment failure rate and auxiliary data, build a second early warning model and calculate the dynamic speed early warning threshold.

[0035] S5. Within the current preset time period M, compare the real-time load and speed of the current nuclear power equipment with the corresponding warning threshold. If the real-time load or speed exceeds the corresponding warning threshold, issue a warning signal.

[0036] S6. Upon receiving a warning signal, trigger the analysis and diagnosis mechanism to diagnose the cause of the fault and execute the adjustment strategy. Reduce the load or speed according to the set adjustment gradient. After each execution of the adjustment strategy, monitor the change in the corresponding load or speed and execute the corresponding action.

[0037] A computer-readable storage medium storing computer-readable instructions that, when executed by one or more processors, cause the one or more processors to perform the steps in the intelligent dynamic threshold early warning and diagnosis method based on device operation.

[0038] The beneficial technical effects of this invention are as follows:

[0039] 1. This invention effectively extracts key features of load and rotational speed, including mean and variance, from preprocessed operational data. These features accurately reflect the operating status of the equipment. Then, a machine learning feature selection algorithm is used to further optimize the feature set, improving the input quality of subsequent early warning models. Based on the mean and variance of load and rotational speed, combined with equipment failure rate and auxiliary data, first and second early warning models are built. These two models can calculate dynamic load and rotational speed early warning thresholds, enabling real-time updating and adjustment of the thresholds to better adapt to changes in equipment operating status. This invention not only improves the accuracy and reliability of nuclear power equipment operation monitoring but also achieves intelligent setting of early warning thresholds. This helps to promptly identify potential problems in equipment operation, prevent failures, and thus ensure the safe and stable operation of nuclear power equipment. Simultaneously, this solution also provides strong technical support for the maintenance and management of nuclear power equipment, improving overall operating efficiency and safety.

[0040] 2. The system of this invention can compare the load and speed of nuclear power equipment with dynamic early warning thresholds in real time. Once an anomaly is detected, an early warning signal is issued in a timely manner, which helps managers or the system to respond quickly and handle potential faults. At the same time, after receiving the early warning signal, the anomaly diagnosis and adjustment module can automatically trigger the analysis and diagnosis mechanism to accurately determine the cause of the fault and execute the corresponding adjustment strategy to reduce the load or speed. By monitoring the changes in load or speed after adjustment, the system can intelligently decide whether to continue to implement the adjustment strategy or to conduct shutdown maintenance, thereby ensuring the safe and stable operation of nuclear power equipment. This invention not only improves the intelligence level of nuclear power equipment operation monitoring, but also realizes the timely detection and handling of potential faults, effectively preventing the occurrence of serious faults. At the same time, through the support of automatic adjustment and shutdown maintenance decisions, the risks and costs of equipment operation are reduced, and the overall operating efficiency and safety are improved. In summary, this technical solution provides strong technical support and guarantee for the operation monitoring and safety management of nuclear power equipment. Attached Figure Description

[0041] Figure 1 This is a modular schematic diagram of an intelligent dynamic threshold early warning and diagnosis system based on the operation of nuclear power equipment provided by the present invention;

[0042] Figure 2 The flowchart illustrates an intelligent dynamic threshold early warning and diagnosis method based on the operation of nuclear power equipment, as provided by this invention. Detailed Implementation

[0043] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0044] like Figure 1 As shown, this invention provides an intelligent dynamic threshold early warning and diagnosis system based on nuclear power equipment operation, comprising: a data acquisition and processing module, a feature extraction module, a dynamic threshold calculation module, an early warning judgment module, and an anomaly diagnosis and adjustment module connected in sequence. The data acquisition and processing module, under normal operating conditions of the nuclear power equipment, deploys corresponding sensors on the equipment to collect real-time data on the operating load and rotational speed as operational data, and real-time data on ambient temperature as auxiliary data; it then preprocesses the operational data and auxiliary data to obtain preprocessed operational data and auxiliary data.

[0045] The sensors installed on nuclear power equipment include load sensors, speed sensors, and temperature sensors.

[0046] Load sensors are used to measure the load on equipment, such as 200T load force sensors and 50t load sensors. These sensors are typically designed with high-precision force standards and can accurately measure and output load data. Load sensors output measurement data through electrical signals, which can be captured by data acquisition systems and converted into digital signals for processing and analysis. On nuclear power equipment, load sensors may be connected to existing data acquisition systems via wired or wireless means to achieve real-time data transmission.

[0047] Speed ​​sensors, such as the EM3305N nuclear-grade speed sensor, are designed using the principle of electromagnetic induction to monitor the speed of the main coolant pump equipment in nuclear power plants. The speed sensor changes the magnetic flux of the sensor coil by a signal gear mounted on the pump shaft, inducing an AC voltage with a frequency proportional to the pump shaft speed. This voltage signal can be captured by the data acquisition system. On nuclear power equipment, speed sensors are usually connected to the data acquisition system via wired connection to ensure the stability and reliability of data transmission.

[0048] Temperature sensors are used to monitor temperature changes in the area where nuclear power equipment is located in real time. The sensors need to have high radiation resistance, high temperature performance and high accuracy to adapt to the special working environment of nuclear power plants. Temperature sensors sense temperature changes through thermistor elements and convert this change into an electrical signal output. On nuclear power equipment, temperature sensors may be connected to the data acquisition system through wired or wireless means to realize real-time data transmission.

[0049] Based on the structure of the nuclear power equipment and the monitoring requirements, appropriate sensors can be placed in suitable locations.

[0050] The preprocessing of operational and auxiliary data involves applying data filtering and outlier detection algorithms to clean, denoise, and normalize the operational and auxiliary data. This ensures data quality while improving data accuracy and reliability, providing a reliable foundation for subsequent analysis.

[0051] Data filtering algorithm: Median filtering is used to sort continuous sampled values ​​and take the median, which can effectively suppress random fluctuation interference. Combined with Kalman filtering, parameters are adjusted in real time based on the state-space model to deal with dynamic noise. The combination of the two removes impulse interference and adapts to noise changes, improving data purity. Outlier detection algorithm: The local outlier factor algorithm is used to analyze the local density of data points and identify outliers with large differences from their surroundings. At the same time, Z-Score detection is used to convert the data into a standard normal distribution. Data exceeding the threshold is judged as anomalies, and outlier data is removed in time to avoid misleading analysis. Normalization method: Min-max normalization is used to scale the data to the range of 0 to 1, eliminating differences in different feature dimensions and enhancing data comparability. By using the above algorithms, data filtering removes noise, outlier detection eliminates anomalies, and normalization unifies dimensions, thereby achieving the effects mentioned above.

[0052] By adopting the above technical solutions, the main problems of data acquisition, transmission, and processing in the operation monitoring of nuclear power equipment are solved. Specifically, by deploying high-precision, high-reliability load sensors, speed sensors, and temperature sensors on the nuclear power equipment, real-time acquisition and transmission of equipment load, speed, and ambient temperature are achieved. These sensors can accurately measure and output corresponding data, providing a reliable foundation for the operation monitoring of nuclear power equipment. At the same time, the collected operation data and auxiliary data are preprocessed, including data filtering, outlier detection, and normalization, which effectively cleans up noise and outliers in the data, improving the accuracy and reliability of the data. This provides a strong guarantee for subsequent data analysis, fault early warning, and decision support.

[0053] In summary, this technical solution enables real-time acquisition and efficient transmission of nuclear power equipment operation data. Furthermore, data preprocessing improves data quality, providing strong support for the safe and stable operation of nuclear power equipment. It helps to promptly identify and address potential problems in equipment operation, reduce the risk of failure, and improve the overall operating efficiency and safety of nuclear power plants.

[0054] The feature extraction module extracts equipment operation-related features from preprocessed operational data within a preset time period M before scheduled maintenance shutdown. It then uses a machine learning feature selection algorithm to extract key features from these operation-related features, which are used as input to the subsequent early warning model. Operation-related features include the range of load and speed variations, while key features include load assessment data and speed assessment data.

[0055] Load assessment data includes the mean and variance of the load; speed assessment data includes the mean and variance of the speed.

[0056] Machine learning feature selection algorithms represent automated feature extraction methods that select the most useful features for classification or regression tasks by training a model.

[0057] Machine learning feature selection algorithms optimize load and speed feature sets from multiple dimensions. First, based on correlation analysis, the Pearson correlation coefficient is used to measure the linear correlation between load / speed features and equipment fault labels, eliminating weakly correlated features and retaining key features such as mean and variance that are highly relevant to early warning, thus reducing redundant information. Second, model-based feature importance assessment is employed, such as using a random forest algorithm. The importance of each load / speed feature is ranked according to its contribution to the splitting of decision tree nodes, prioritizing the retention of highly important features and removing inefficient ones. Third, a recursive feature elimination method is used. Starting from all initial load and speed features, the model is continuously trained, and features that have the least impact on model performance are eliminated, gradually optimizing the feature subset so that the final retained features most accurately reflect the equipment's operating status, providing high-quality input to the early warning model and improving the accuracy and timeliness of early warnings.

[0058] In one specific implementation, the machine learning feature selection algorithm includes either Lasso regression or random forest; Lasso regression is used to select the feature most relevant to the device operating status from multiple candidate features, namely load assessment data and rotation speed assessment data;

[0059] It should be noted that: during the shutdown maintenance period, the key features extracted from the operating data of the nuclear power equipment under normal operating conditions collected within the previous preset time period M are used as the load assessment data and speed assessment data for the next preset time period M; M > 0 and is a positive integer.

[0060] The dynamic threshold calculation module, within the current preset time period M, builds a first early warning model based on load assessment data, as well as the acquired equipment failure rate and auxiliary data, and calculates the dynamic load early warning threshold according to the algorithm formula within the model; based on the rotational speed assessment data, as well as the acquired equipment failure rate and auxiliary data, it builds a second early warning model and calculates the dynamic rotational speed early warning threshold according to the algorithm formula within the model.

[0061] The formula for calculating the load warning threshold in the first early warning model is as follows:

[0062]

[0063] In the formula, Lr represents the load warning threshold, μ represents the mean load within the previous preset time period M, σ represents the variance of the load within the previous preset time period M, r represents the equipment failure rate before this period, Pt represents auxiliary data, i.e., the real-time ambient temperature, and k1 and k2 represent adjustment coefficients used to adjust the variance, equipment failure rate, and the influence of auxiliary data on the warning threshold. Their values ​​range from [0, 1]. If k1 and k2 are both 1, the above formula can be simplified as follows:

[0064]

[0065] In practical applications, these two adjustment coefficients can be determined based on experience or experimental data. The determination needs to be combined with the specific scenario. Taking the adjustment coefficient k1 in the data filtering algorithm as an example: In real-time monitoring scenarios where data fluctuates frequently and with large amplitude, and rapid response to changes is required, k1 takes a value close to 0, such as 0.1 to 0.3. At this time, the new sampled value has a large impact on the filtering result and can capture data changes in a timely manner, but the anti-interference ability is slightly weaker. In industrial control scenarios where the data is relatively stable and there is a lot of noise interference, k1 takes a value close to 1, such as 0.7 to 0.9. In this way, the filtering result is more dependent on historical data, which can effectively suppress noise and make the output smooth and stable. If it is between the above two scenarios, k1 takes an intermediate value, such as 0.4 to 0.6. The determination is mainly based on the actual data characteristics, system response requirements, and the trade-off between noise and sensitivity to changes. The filtering effect under different values ​​is tested through experiments, and the optimal value is selected (the value range of k1 under all conditions covers [0, 1]).

[0066] To calculate the dynamic load warning threshold, we need to consider the mean and variance of the load, which are two statistics that reflect the distribution characteristics of the load data. At the same time, the equipment failure rate is also an important reference factor, as it reflects the reliability of the equipment under a specific load. In addition, auxiliary data (such as ambient temperature) may also affect the load warning threshold.

[0067] The first early warning model takes into account the above influencing factors to calculate the load early warning threshold. The formula in the model combines the load mean, variance, equipment failure rate and auxiliary data, and obtains a dynamic early warning threshold through certain mathematical operations.

[0068] The design logic of the calculation formula for the first early warning model is:

[0069] Based on the load mean, it reflects the average load level; variance is introduced to adjust the warning threshold to reflect load fluctuations; considering the equipment failure rate, the warning threshold is corrected to reflect the equipment reliability under specific loads; auxiliary data is used to fine-tune the warning threshold to reflect the impact of environmental factors on the load.

[0070] In one specific embodiment, it is assumed that the average load of a certain nuclear power equipment is 50 (unit: tons), the variance is sqrt10, the equipment failure rate is 0.01 (i.e. 1%), the ambient temperature is 30°C (assuming that the effect of temperature on the load warning threshold is linear, and the threshold increases by 0.1 tons for every 1°C increase), k1 = 1, k2 = 0.1;

[0071] Substituting into the formula, we get:

[0072] Lr=50+sqrt10*sqrt1.01+30*0.1≈50+3.09+3=56.19;

[0073] The formula for calculating the speed warning threshold in the second early warning model is as follows:

[0074]

[0075] In the formula, Sr represents the speed warning threshold, α represents the mean speed within the previous preset time period M, β represents the variance of the speed within the previous preset time period M, and k3 and k4 represent adjustment coefficients, both of which have a value range of [0, 1]. The principle of the formula used to calculate the speed warning threshold is the same as that of the load warning threshold, so it will not be elaborated here.

[0076] By adopting the above technical solution, the problem of dynamically setting early warning thresholds in the operation monitoring of nuclear power equipment is mainly solved. Through the feature extraction module, key features of load and speed, including mean and variance, are effectively extracted from the preprocessed operation data. These features can accurately reflect the operating status of the equipment. Then, the feature set is further optimized by using a machine learning feature selection algorithm, which improves the input quality of the subsequent early warning model. In the dynamic threshold calculation module, based on the mean and variance of load and speed, combined with equipment failure rate and auxiliary data, first and second early warning models are built respectively. These two models can calculate dynamic load and speed early warning thresholds, realize the real-time updating and adjustment of thresholds, and better adapt to changes in the operating status of the equipment.

[0077] This technical solution not only improves the accuracy and reliability of nuclear power equipment operation monitoring, but also enables the intelligent setting of early warning thresholds. This helps to promptly identify potential problems in equipment operation, prevent malfunctions, and thus ensure the safe and stable operation of nuclear power equipment. At the same time, this solution also provides strong technical support for the maintenance and management of nuclear power equipment, improving overall operational efficiency and safety.

[0078] The early warning judgment module compares the real-time load and speed of the current nuclear power equipment with the corresponding early warning threshold within the current preset time period M. If the real-time load or speed exceeds the corresponding early warning threshold, it indicates an anomaly and issues an early warning signal; otherwise, it indicates normal operation and no response is taken.

[0079] Specifically, a Level 1 warning signal is issued when the load exceeds the load warning threshold; a Level 2 warning signal is issued when the speed exceeds the speed warning threshold; and a Level 3 warning signal is issued when both the load and speed exceed their respective warning thresholds. These three different warning signals help managers or the system to make distinctions.

[0080] By comparing real-time load with dynamic load warning thresholds, and real-time rotational speed with dynamic rotational speed warning thresholds, abnormal states in the operation of nuclear power equipment can be detected and addressed in a timely manner, ensuring the safe, stable, and efficient operation of nuclear power equipment. At the same time, this also demonstrates the important role of warning thresholds in equipment operation monitoring and safety management.

[0081] The anomaly diagnosis and adjustment module, upon receiving a warning signal, triggers an analysis and diagnosis mechanism to diagnose the cause of the fault and execute an adjustment strategy. It reduces the load or speed according to the set adjustment gradient and monitors the changes in the corresponding load or speed after each execution of the adjustment strategy, and then performs the corresponding action.

[0082] If the difference between the load or speed and the corresponding warning threshold shows a decreasing trend, it indicates that the adjustment strategy is effective. Continue to implement the adjustment strategy and continue to monitor the changes in the corresponding load or speed until the difference between the current load or speed and the corresponding warning threshold exceeds the previous difference, then stop.

[0083] Stop when the difference between the load or speed and the corresponding warning threshold shows an increasing trend;

[0084] If the load or speed exceeds the corresponding warning threshold when the machine stops, maintenance will be performed.

[0085] If the load or rotational speed does not exceed the corresponding warning threshold, the nuclear power equipment shall continue to operate normally and the data shall be recorded.

[0086] The difference between the load or speed and the corresponding warning threshold is represented as: load or speed - corresponding warning threshold.

[0087] It should be noted that the causes of the fault include load, speed, or both, and the set adjustment gradient represents a fixed value for each reduction in load or speed. The magnitude of each reduction is positively correlated with the difference between the load or speed and the corresponding warning threshold.

[0088] For the load:

[0089] ΔL=G L *(L—Lr); where ΔL represents the fixed value of load reduction each time, L represents the real-time load, Lr represents the load warning threshold, and G L G represents the proportionality coefficient for load adjustment, which is a fixed reduction value corresponding to each ton difference, and G L >0.

[0090] Regarding rotational speed:

[0091] ΔS=G S *(S—Sr); where ΔS represents the fixed value of each speed reduction, S represents the real-time speed, Sr represents the speed warning threshold, and G S The proportional coefficient representing the speed adjustment is a fixed decrease corresponding to each revolution / minute difference, and G S >0.

[0092] Proportional relationship: In the above formula, ΔL and ΔS are proportional to (L—Lr) and (S—Sr), respectively.

[0093] Adjusting the gradient: via the scaling factor G L and G S This allows control over the fixed value of each reduction; for example, if the fixed value for each load reduction is determined to be 0.5 tons, then when L - Lr = 1 ton, then G L =0.5; Similarly, for rotational speed, if the fixed value for each speed reduction is 500 r / min, and S - Sr = 1 r / min (this is just an example; the actual difference may be much greater than 1), then G S =500; but it should be noted that G here L and G S Adjustments should be made based on the actual situation to ensure that the reduced fixed value is within a reasonable range.

[0094] Flexibility: The proportionality coefficient G in the formula L and G S It offers flexibility, allowing adjustments to be made based on different equipment, operating conditions, and monitoring needs.

[0095] In summary, this proportional formula can accurately describe the relationship between the fixed value of each load or speed reduction and the difference between the load or speed and the corresponding warning threshold, and provides flexibility and adjustability.

[0096] By adopting the above technical solution, the problems of abnormal early warning, fault diagnosis and automatic adjustment in the operation of nuclear power equipment are mainly solved. The system can compare the load and speed of nuclear power equipment with the dynamic early warning threshold in real time. Once an abnormality is detected, an early warning signal will be issued in time, which helps managers or the system to respond quickly and deal with potential faults.

[0097] Meanwhile, upon receiving an early warning signal, the anomaly diagnosis and adjustment module can automatically trigger the analysis and diagnosis mechanism to accurately determine the cause of the fault and execute corresponding adjustment strategies to reduce the load or speed. By monitoring the changes in the adjusted load or speed, the system can intelligently decide whether to continue implementing the adjustment strategy or to conduct shutdown maintenance, thereby ensuring the safe and stable operation of nuclear power equipment.

[0098] This technical solution not only improves the intelligence level of nuclear power equipment operation monitoring, but also enables the timely detection and handling of potential faults, effectively preventing the occurrence of serious faults. At the same time, through automatic adjustment and shutdown maintenance decision support, it reduces the risk and cost of equipment operation and improves overall operating efficiency and safety. In summary, this technical solution provides strong technical support and guarantee for the operation monitoring and safety management of nuclear power equipment.

[0099] like Figure 2 As shown, the present invention also provides an intelligent dynamic threshold early warning and diagnosis method based on the operation of nuclear power equipment, including:

[0100] S1. Under normal operating conditions of nuclear power equipment, corresponding sensors are installed on the nuclear power equipment to collect load and rotation speed in real time as operating data, and ambient temperature in real time as auxiliary data.

[0101] S2. Preprocess the operational data and auxiliary data;

[0102] S3. Within a preset time period M before the scheduled shutdown maintenance, extract equipment operation-related features from the preprocessed operation data, and use a machine learning feature selection algorithm to extract key features from the equipment operation-related features. The operation-related features include the range of load and speed variation, and the key features include load assessment data and speed assessment data.

[0103] S4. Within the current preset time period M, based on the load assessment data and the obtained equipment failure rate and auxiliary data, a first early warning model is built, and the dynamic load early warning threshold is calculated according to the algorithm formula in the model; based on the speed assessment data and the obtained equipment failure rate and auxiliary data, a second early warning model is built, and the dynamic speed early warning threshold is calculated according to the algorithm formula in the model.

[0104] S5. Within the current preset time period M, compare the real-time load and speed of the current nuclear power equipment with the corresponding warning threshold. If the real-time load or speed exceeds the corresponding warning threshold, issue a warning signal.

[0105] S6. Upon receiving a warning signal, trigger the analysis and diagnosis mechanism to diagnose the cause of the fault and execute the adjustment strategy. Reduce the load or speed according to the set adjustment gradient. After each execution of the adjustment strategy, monitor the change in the corresponding load or speed and execute the corresponding action.

[0106] As an implementation of the above method, the present invention provides a computer-readable storage medium that corresponds to the above-described intelligent dynamic threshold early warning and diagnosis method based on the operation of nuclear power equipment.

[0107] The computer-readable storage medium of the present invention stores computer-readable instructions that can be executed by at least one processor to cause the at least one processor to perform the steps of the intelligent dynamic threshold early warning and diagnosis method based on the operation of nuclear power equipment as described above.

[0108] In this invention, the various formulas involved are all calculated by removing dimensions and taking their numerical values. The formulas are established by collecting a large amount of data and simulating the most recent real situation. Some coefficients or weights in the formulas are set by those skilled in the art according to the actual situation, so they will not be elaborated here.

[0109] The specific embodiments described above can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the specific embodiments described above can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the specific embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0110] The present invention has been described in detail above with reference to the accompanying drawings and embodiments. However, the present invention is not limited to the above embodiments, and various changes can be made within the scope of knowledge possessed by those skilled in the art without departing from the spirit of the present invention. All contents not described in detail in the present invention can be derived from existing technologies.

Claims

1. An intelligent dynamic threshold early warning and diagnostic system based on the operation of nuclear power equipment, characterized in that, include: The data acquisition and processing module collects the operating load and speed of the nuclear power equipment in real time as operating data, and collects the ambient temperature in real time as auxiliary data, under the normal operating conditions of the nuclear power equipment. Preprocess the operational data and auxiliary data to obtain preprocessed operational data and auxiliary data; The feature extraction module extracts equipment operation-related features from preprocessed operating data within a preset time period M before scheduled shutdown and maintenance. Machine learning feature selection algorithms are used to extract key features from device operation-related features. Key features include: load assessment data and rotational speed assessment data; The dynamic threshold calculation module, within the current preset time period M, builds a first early warning model based on the load assessment data, the obtained equipment failure rate and auxiliary data, and calculates the dynamic load early warning threshold; based on the speed assessment data, the obtained equipment failure rate and auxiliary data, it builds a second early warning model and calculates the dynamic speed early warning threshold. The early warning judgment module compares the real-time load and speed of the current nuclear power equipment with the corresponding early warning threshold within the current preset time period M. If the real-time load or speed exceeds the corresponding early warning threshold, an early warning signal is issued. The anomaly diagnosis and adjustment module, upon receiving a warning signal, triggers an analysis and diagnosis mechanism to diagnose the cause of the fault and execute an adjustment strategy. It reduces the load or speed according to the set adjustment gradient and monitors the changes in the corresponding load or speed after each execution of the adjustment strategy, and then performs the corresponding action.

2. The intelligent dynamic threshold early warning and diagnosis system based on nuclear power equipment operation according to claim 1, characterized in that, In the data acquisition and processing module, the real-time operating load and speed of the nuclear power equipment are acquired through sensors arranged on the nuclear power equipment. The sensors include load sensors, speed sensors, and temperature sensors. The preprocessing process for the operating data and auxiliary data is as follows: data filtering algorithms and outlier detection algorithms are applied to the operating data and auxiliary data to perform cleaning, noise reduction, and normalization.

3. The intelligent dynamic threshold early warning and diagnostic system based on the operation of nuclear power equipment according to claim 1, characterized in that, In the feature extraction module, the load evaluation data includes the mean and variance of the load; the rotational speed evaluation data includes the mean and variance of the rotational speed.

4. The intelligent dynamic threshold early warning and diagnosis system based on nuclear power equipment operation according to claim 1, characterized in that, In the feature extraction module, the machine learning feature selection algorithm includes either Lasso regression or random forest.

5. The intelligent dynamic threshold early warning and diagnosis system based on the operation of nuclear power equipment according to claim 1, characterized in that, In the feature extraction module, the operating data of the nuclear power equipment under normal operating conditions collected within the previous preset time period T during the shutdown maintenance period is used as the basis, and the extracted key features are used as the load assessment data and speed assessment data for the next preset time period T; M>0 and is a positive integer.

6. The intelligent dynamic threshold early warning and diagnosis system based on the operation of nuclear power equipment according to claim 1, characterized in that, In the dynamic threshold calculation module, the calculation formula for the load warning threshold in the first warning model is as follows: In the formula, Lr represents the load warning threshold, μ represents the mean load within the previous preset time period M, σ represents the variance of the load within the previous preset time period M, r represents the equipment failure rate before this, Pt represents auxiliary data, k1 and k2 represent adjustment coefficients, and the values ​​of k1 and k2 are both in the range of [0, 1]. The formula for calculating the speed warning threshold in the second early warning model is: In the formula, Sr represents the speed warning threshold, α represents the mean speed within the previous preset time period M, β represents the variance of the speed within the previous preset time period M, and k3 and k4 represent adjustment coefficients, with k3 and k4 both ranging from [0, 1].

7. The intelligent dynamic threshold early warning and diagnosis system based on nuclear power equipment operation according to claim 1, characterized in that, In the warning judgment module, if the real-time load or rotational speed exceeds the corresponding warning threshold, a warning signal is issued. Specifically, when the load exceeds the load warning threshold, a first-level warning signal is issued; when the rotational speed exceeds the rotational speed warning threshold, a second-level warning signal is issued; and when both the load and rotational speed exceed the corresponding warning thresholds, a third-level warning signal is issued.

8. The intelligent dynamic threshold early warning and diagnosis system based on nuclear power equipment operation according to claim 1, characterized in that, The anomaly diagnosis and adjustment module monitors changes in the corresponding load or rotational speed and executes corresponding actions, specifically: If the difference between the load or speed and the corresponding warning threshold shows a decreasing trend, it indicates that the adjustment strategy is effective. Continue to implement the adjustment strategy and continue to monitor the changes in the corresponding load or speed until the difference between the current load or speed and the corresponding warning threshold exceeds the previous difference, then stop. Stop when the difference between the load or speed and the corresponding warning threshold shows an increasing trend; If the load or speed exceeds the corresponding warning threshold when the machine stops, maintenance will be performed. If the load or rotational speed does not exceed the corresponding warning threshold, maintain the normal operation of the nuclear power equipment and record it; The difference between the load or speed and the corresponding warning threshold is: load or speed - corresponding warning threshold.

9. A method for intelligent dynamic threshold early warning and diagnosis based on the operation of nuclear power equipment, employing the intelligent dynamic threshold early warning and diagnosis system based on the operation of nuclear power equipment as described in any one of claims 1-8, characterized in that, include: S1. Under normal operating conditions of nuclear power equipment, load and rotation speed are collected in real time as operating data, and ambient temperature is collected in real time as auxiliary data. S2. Preprocess the operational data and auxiliary data; S3. Within a preset time period M before the scheduled shutdown maintenance, extract equipment operation-related features from the preprocessed operation data, and use machine learning feature selection algorithms to extract key features from the equipment operation-related features. Key features include: load assessment data and rotational speed assessment data; S4. Within the current preset time period M, based on the load assessment data, the obtained equipment failure rate and auxiliary data, build a first early warning model and calculate the dynamic load early warning threshold; based on the speed assessment data, the obtained equipment failure rate and auxiliary data, build a second early warning model and calculate the dynamic speed early warning threshold. S5. Within the current preset time period M, compare the real-time load and speed of the current nuclear power equipment with the corresponding warning threshold. If the real-time load or speed exceeds the corresponding warning threshold, issue a warning signal. S6. Upon receiving a warning signal, trigger the analysis and diagnosis mechanism to diagnose the cause of the fault and execute the adjustment strategy. Reduce the load or speed according to the set adjustment gradient. After each execution of the adjustment strategy, monitor the change in the corresponding load or speed and execute the corresponding action.

10. A computer-readable storage medium storing computer-readable instructions thereon, characterized in that, When computer instructions are executed by one or more processors, the one or more processors are caused to perform the steps in the intelligent dynamic threshold early warning and diagnosis method based on device operation as described in claim 9.

Images