Method for detecting failure of nuclear power plant ring crane and visual system

Abstract

The application relates to a nuclear power plant ring crane fault detection method and a visualization system. The method comprises the following steps: collecting operation data of a ring crane device, wherein the operation data comprises time series data of multiple monitoring dimensions; preprocessing the collected operation data to obtain standardized effective data, and extracting time trend features and space coupling features based on the effective data; inputting the time trend features and the space coupling features into a long short-term memory network, selectively memorizing and updating historical state data and current input feature data through a forgetting gate, an input gate and an output gate of the long short-term memory network, and outputting a fault probability of the device at the current time; mapping the fault probability to obtain a corresponding risk level, and correcting the risk level based on historical risk events. The method can realize early prediction and accurate positioning of device faults, and intuitively present the risk level and the operation state through a visualization interface, so that the fault troubleshooting time is shortened.

Description

Technical Field

[0001] This application relates to the field of ring crane equipment technology, and in particular to a method and visualization system for detecting ring crane faults in nuclear power plants. Background Technology

[0002] In nuclear power plants, especially the Taishan Nuclear Power Plant which uses French Generation III EPR units, the nuclear island hoist and airlock are critical imported French equipment. During engineering applications and major overhauls, their electrical control systems frequently malfunction, severely impacting operation and maintenance efficiency and safety. Currently, the hoist only has an HMI display screen on the control panel at the 20-meter platform. When a malfunction occurs, maintenance personnel must inquire about the fault code by phone with the operator; if the fault is accidentally reset, the code is often lost, forcing maintenance personnel to personally go to the control panel at the 20-meter platform to locate it—a cumbersome and time-consuming process. On the critical overhaul path, such delays directly hinder the overall progress and affect the safe and stable operation of the power plant.

[0003] Traditional single-dimensional data monitoring, without combining big data models for multi-dimensional time-series data fusion analysis, makes it difficult to accurately predict fault probabilities and quantitatively assess risk levels, thus failing to meet the high safety requirements of nuclear power plant operation and maintenance. Third, existing systems have a fixed structure, unsuitable for flexible on-site installation and mobile operation, and lack robust data storage, fault tracing, and multi-user access control functions.

[0004] Therefore, there is an urgent need for a fault detection method and visualization system for nuclear power plant hoists, which can enable early prediction and precise location of equipment faults, and intuitively present the risk level and operating status through a visualization interface, thereby shortening the fault investigation time and ensuring the safe and efficient operation and maintenance of nuclear power plants. Summary of the Invention

[0005] Therefore, it is necessary to provide a nuclear power plant hoist fault detection method and visualization system that can predict and accurately locate equipment faults in advance, and intuitively present the risk level and operating status through a visual interface, thereby shortening the fault investigation time and ensuring the safe and efficient operation and maintenance of nuclear power plants.

[0006] Firstly, this application provides a method for detecting faults in a nuclear power plant's ring crane, including:

[0007] Collect operational data of the ring crane equipment, which includes time-series data from multiple monitoring dimensions;

[0008] The collected operational data is preprocessed to obtain standardized effective data, and time trend features and spatial coupling features are extracted based on the effective data; wherein, the time trend features are used to characterize the data change pattern of a single monitoring dimension within a time window, and the spatial coupling features are used to characterize the correlation between different monitoring dimensions;

[0009] The time trend features and spatial coupling features are input into the Long Short-Term Memory (LSTM) network. Through the forget gate, input gate, and output gate of the LSM network, historical state data and currently input feature data are selectively memorized and updated, and the failure probability of the device at the current moment is output.

[0010] The corresponding risk level is obtained based on the failure probability mapping, and the risk level is corrected based on historical risk events.

[0011] In one embodiment, the preprocessing of the collected operational data includes:

[0012] The three-standard-deviation criterion is used to identify outliers in each monitoring dimension. Values ​​that deviate from the mean by more than three standard deviations are judged as outliers and corrected accordingly.

[0013] The data for each monitoring dimension after correction are standardized and mapped to a preset range to obtain the standardized effective data.

[0014] In one embodiment, the extraction of time trend features based on the effective data includes:

[0015] A sliding window is used to calculate the trend slope of a preset number of time steps prior to the current moment, and the trend slope is used as the time trend feature; the length of the sliding window corresponds to the number of time steps of a preset duration.

[0016] In one embodiment, the extraction of spatial coupling features based on the effective data includes:

[0017] Calculate the mutual information between different monitoring dimensions, whereby the mutual information is used to characterize the degree of interdependence between two monitoring dimensions;

[0018] Based on the preset weight coefficients of each monitoring dimension, the calculated mutual information from multiple sets is weighted and summed to obtain the spatial coupling feature.

[0019] In one embodiment, obtaining the corresponding risk level based on the failure probability mapping and correcting the risk level based on historical risk events includes:

[0020] According to a preset mapping relationship between multiple probability intervals and risk levels, the failure probability is converted into an initial risk level; in the mapping relationship, the failure probability and the corresponding risk level are directly proportional.

[0021] Obtain the historical risk levels for multiple consecutive days prior to the current moment, and count the number of days that reached the medium risk level or above;

[0022] Based on the number of days obtained from statistics and the preset historical risk impact coefficient, a risk correction factor is calculated, and the product of the initial risk level and the risk correction factor is taken as the final risk level; wherein, the more recent the historical risk event is, the greater its weight in the correction factor.

[0023] In one embodiment, the method further includes:

[0024] The final risk level is mapped to a visual color parameter;

[0025] Based on the weight coefficients of each monitoring dimension, the failure probability, and the standardized valid data, the visualization weight corresponding to each monitoring dimension is calculated, and the corresponding components of the hoisting equipment are highlighted differently according to the level of the visualization weight.

[0026] In one embodiment, the method further includes:

[0027] The system performs full lifecycle monitoring on at least one of the following components in the hoisting equipment: frequency converter, brake contactor, motor, and holding brake. It records the running time and number of actions, and outputs an early warning signal when the running time or number of actions reaches a preset threshold.

[0028] In addition, the on / off status of the power circuit circuit breaker and contactor is displayed in the form of a single-line diagram, and the on / off status of each network node in the DP bus and Ethernet is displayed in real time in the form of a node list.

[0029] Secondly, this application also provides a nuclear power plant ring crane fault detection visualization system, including:

[0030] The data acquisition module is used to establish communication with the PLC of the hoisting equipment and collect the operating data of the hoisting equipment, which includes time-series data of multiple monitoring dimensions.

[0031] The feature extraction module is used to preprocess the collected operational data to obtain standardized effective data, and extract time trend features and spatial coupling features based on the effective data; wherein, the time trend features are used to characterize the data change pattern of a single monitoring dimension within a time window, and the spatial coupling features are used to characterize the correlation between different monitoring dimensions;

[0032] The fault prediction and risk assessment module is used to input the time trend features and the spatial coupling features into the long short-term memory network, selectively memorize and update the historical state and the current input through the forget gate, input gate and output gate of the long short-term memory network, and output the fault probability of the device at the current moment; the corresponding risk level is obtained according to the fault probability mapping, and the risk level is corrected based on historical risk events, and the corrected final risk level is output.

[0033] The visualization module is used to generate and display a visualization interface containing equipment operating status, fault information and risk level based on the final risk level and the operating data.

[0034] In one embodiment, the data acquisition module is used to establish communication with the PLC of the hoisting equipment via the Ethernet TCP / IP protocol. The PLC includes OPLC and PTMS. The operating data collected by the data acquisition module includes: current, voltage, speed, torque of the drive device, limit switch status, encoder reading, frequency converter operating parameters, and brake operating frequency, cumulative working time, and number of actions.

[0035] In one embodiment, the visualization module includes:

[0036] The fault alarm display unit is used to display fault codes and fault alarm records in real time in the form of a text list;

[0037] The trend curve display unit is used to provide the display and export functions of real-time trend curves and historical trend curves of each monitored variable;

[0038] The status monitoring and display unit is used to display the on / off status of the power circuit circuit breaker and contactor in the form of a single-line diagram, and to display the on / off status of each network node in the DP bus and Ethernet in the form of node status.

[0039] The brake status display unit is used to display the real-time status, operating frequency, and cumulative operating time of the brake.

[0040] The parameter input unit provides an interface for users to input preset thresholds to support the setting of early warning conditions in the whole life cycle monitoring.

[0041] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0042] Collect operational data of the ring crane equipment, which includes time-series data from multiple monitoring dimensions;

[0043] The collected operational data is preprocessed to obtain standardized effective data, and time trend features and spatial coupling features are extracted based on the effective data; wherein, the time trend features are used to characterize the data change pattern of a single monitoring dimension within a time window, and the spatial coupling features are used to characterize the correlation between different monitoring dimensions;

[0044] The time trend features and spatial coupling features are input into the Long Short-Term Memory (LSTM) network. Through the forget gate, input gate, and output gate of the LSM network, historical state data and currently input feature data are selectively memorized and updated, and the failure probability of the device at the current moment is output.

[0045] The corresponding risk level is obtained based on the failure probability mapping, and the risk level is corrected based on historical risk events.

[0046] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0047] Collect operational data of the ring crane equipment, which includes time-series data from multiple monitoring dimensions;

[0048] The collected operational data is preprocessed to obtain standardized effective data, and time trend features and spatial coupling features are extracted based on the effective data; wherein, the time trend features are used to characterize the data change pattern of a single monitoring dimension within a time window, and the spatial coupling features are used to characterize the correlation between different monitoring dimensions;

[0049] The time trend features and spatial coupling features are input into the Long Short-Term Memory (LSTM) network. Through the forget gate, input gate, and output gate of the LSM network, historical state data and currently input feature data are selectively memorized and updated, and the failure probability of the device at the current moment is output.

[0050] The corresponding risk level is obtained based on the failure probability mapping, and the risk level is corrected based on historical risk events.

[0051] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0052] Collect operational data of the ring crane equipment, which includes time-series data from multiple monitoring dimensions;

[0053] The collected operational data is preprocessed to obtain standardized effective data, and time trend features and spatial coupling features are extracted based on the effective data; wherein, the time trend features are used to characterize the data change pattern of a single monitoring dimension within a time window, and the spatial coupling features are used to characterize the correlation between different monitoring dimensions;

[0054] The time trend features and spatial coupling features are input into the Long Short-Term Memory (LSTM) network. Through the forget gate, input gate, and output gate of the LSM network, historical state data and currently input feature data are selectively memorized and updated, and the failure probability of the device at the current moment is output.

[0055] The corresponding risk level is obtained based on the failure probability mapping, and the risk level is corrected based on historical risk events.

[0056] The aforementioned nuclear power plant hoist fault detection method and visualization system collects time-series data from multiple monitoring dimensions of the hoist equipment. By preprocessing and standardizing the raw data, it effectively eliminates interference from sensor noise, missing values, and dimensional differences, providing a high-quality data foundation for subsequent analysis. Based on this, it extracts temporal trend features reflecting equipment degradation and spatial coupling features revealing the relationships between various components, overcoming the limitations of traditional single-dimensional monitoring. These spatiotemporal features are input into a Long Short-Term Memory (LSTM) network, which uses its forget gate, input gate, and output gate to dynamically and selectively memorize and update historical states and current features. This accurately captures the long-term dependence and transient changes in equipment operating status, outputting high-confidence fault probabilities and enabling early fault prediction rather than post-event alarms. Furthermore, the fault probability is mapped to a quantified risk level, and historical risk events are introduced for correction. This ensures that the risk assessment reflects both the current situation and the cumulative impact of recent medium- to high-risk events, preventing the overlooking of recurring potential faults. This significantly shortens fault location and troubleshooting time, reduces delay risks on critical overhaul paths, and comprehensively improves the safety and efficiency of nuclear power plant hoist equipment operation and maintenance. Attached Figure Description

[0057] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0058] Figure 1 This is an application environment diagram of a nuclear power plant ring crane fault detection method in one embodiment;

[0059] Figure 2 This is a flowchart illustrating a method for detecting faults in a nuclear power plant ring crane in one embodiment;

[0060] Figure 3 This is a flowchart illustrating a nuclear power plant ring crane fault detection method in another embodiment;

[0061] Figure 4 This is a schematic diagram of the communication interaction between the PLC and the host system in one embodiment;

[0062] Figure 5 This is a structural block diagram of a nuclear power plant ring crane fault detection visualization system in one embodiment;

[0063] Figure 6 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0064] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0065] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0066] The nuclear power plant ring crane fault detection method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104, or it can be located in the cloud or on another network server.

[0067] The terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle systems, and projection devices. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted displays. Head-mounted displays can be virtual reality (VR) devices, augmented reality (AR) devices, and smart glasses. The server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0068] In one exemplary embodiment, such as Figure 2 As shown, a method for detecting faults in a nuclear power plant's ring crane is provided, which is then applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps S202 to S208. Wherein:

[0069] Step S202: Collect the operating data of the ring crane equipment. The operating data includes time-series data from multiple monitoring dimensions.

[0070] Specifically, it is necessary to collect operational data from the hoisting equipment, including time-series data from multiple monitoring dimensions. These multiple monitoring dimensions refer to monitoring points for different components or physical quantities of the hoisting equipment, such as the current, voltage, speed, and torque of the drive unit, the status signals of limit switches, the position readings of the encoder, the operating parameters of the frequency converter, and network status (such as DP bus, Ethernet) and power circuit status. Values ​​for each dimension are collected continuously or discretely along the time axis, forming time-series data with a chronological order. For example, for the hoisting mechanism motor of the hoisting equipment, the current value can be recorded every 100 milliseconds, thus obtaining a current curve that changes over time; similarly, the closing state of the brake contactor can be recorded as a series of discrete logic values ​​(0 or 1) over time.

[0071] Step S204: Preprocess the collected operational data to obtain standardized effective data, and extract time trend features and spatial coupling features based on the effective data; wherein, the time trend features are used to characterize the data change pattern of a single monitoring dimension within a time window, and the spatial coupling features are used to characterize the correlation between different monitoring dimensions.

[0072] Specifically, due to the complex operating environment of the hoisting equipment, sensor signals may be subject to noise interference, or missing values ​​or abnormal jumps may occur due to communication interruptions. Directly using this raw data for analysis would reduce the accuracy of subsequent detection. Therefore, the purpose of preprocessing is to eliminate these adverse factors. For example, for the monitoring dimension of motor current, if peak values ​​far exceeding the normal range occasionally occur due to electromagnetic interference, the preprocessing stage will identify them and perform reasonable correction or smoothing. For missing data points, interpolation or forward imputation can be used to supplement them. After outlier correction and missing value processing, the data for each dimension also needs to be standardized.

[0073] Building upon this foundation, two core features are further extracted from these effective data: temporal trend features and spatial coupling features. Temporal trend features focus on the changing patterns of a single monitoring dimension over time. From normal operation to malfunction, the hoisting equipment often experiences a gradual or abrupt change in a certain physical quantity along the time axis, such as a slow increase in motor current or a gradual rise in brake temperature. To capture this temporal change, data from the most recent time window for that dimension can be extracted at each moment. By calculating indicators such as the trend slope, amplitude of change, or degree of fluctuation within that window, the current trend can be quantitatively described.

[0074] Unlike temporal trend characteristics, spatial coupling characteristics focus on the correlation between different monitoring dimensions. As a complex electromechanical system, the components of a hoist do not operate independently but are tightly coupled. For example, when the inverter output frequency increases, the motor speed and current usually increase synchronously; when a limit switch is triggered, the corresponding brake will activate. If this normal coupling relationship becomes abnormal, such as an increase in speed but an abnormal decrease in current, it may indicate an encoder malfunction or a problem with the transmission system. Therefore, it is necessary to calculate the degree of interdependence between different monitoring dimensions, such as quantifying the consistency, synchronicity, or information sharing of data changes in two dimensions using statistical methods.

[0075] Step S206: Input the time trend features and spatial coupling features into the long short-term memory network. Through the forget gate, input gate and output gate of the long short-term memory network, selectively memorize and update the historical state data and the currently input feature data, and output the failure probability of the device at the current moment.

[0076] Specifically, the extracted temporal trend features and spatial coupling features are used as inputs and fed into a Long Short-Term Memory (LSTM) network. The reason for choosing LSTM instead of a regular recurrent neural network is that the operating data of the hoisting equipment has a long-term dependency, meaning that the current failure tendency may originate from state changes several minutes or even hours ago. Ordinary networks are prone to gradient vanishing or gradient exploding when processing long sequences, resulting in an inability to effectively remember early information.

[0077] In LSTM operations, the forget gate, input gate, and output gate work together to manage an internal memory unit. The forget gate reads the hidden state from the previous time step (i.e., the network's compressed representation of past information) and the temporal trend and spatial coupling features of the current input, outputting a value between 0 and 1 that determines how much information from the previous time step's memory unit should be retained (value close to 1) or discarded (value close to 0). The input gate controls how much new information from the current input features needs to be written into the memory unit. It also calculates a weight based on the previous time step's hidden state and the current features, and updates the memory unit's content using a candidate memory vector. The output gate, based on the updated memory unit and the current input, determines the final output of the hidden state, which is used both for the next time step's calculations and for generating the current time step's fault probability.

[0078] Among them, the Gate of Oblivion:

[0079] ;

[0080] Input Gate:

[0081] ;

[0082] Candidate state:

[0083] ;

[0084] Cell status:

[0085] ;

[0086] Output gate:

[0087] ;

[0088] Hidden state:

[0089] ;

[0090] Where: σ() is the Sigmoid activation function, tanh() is the hyperbolic tangent function, ⊙ is the Hadamard product, and Wf / W i / W C / W o Let b be the weight matrix. f / b i / b C / b o For bias terms;

[0091] Through iterative calculations using the aforementioned gating mechanism, the LSTM outputs a hidden state vector at each time step. This vector integrates historical sequence information and current spatiotemporal features. This hidden state is mapped to a value between 0 and 1, representing the probability of device failure at the current moment. This probability value reflects the likelihood of device failure: the closer the value is to 0, the more normal the operating state; the closer the value is to 1, the higher the risk of failure.

[0092] Among them, the fault probability output is:

[0093] ,in: For output layer weights, For output layer bias, The closer the failure probability is to 1, the greater the likelihood of equipment failure, thus enabling early prediction of failures.

[0094] Step S208: Obtain the corresponding risk level based on the failure probability mapping, and correct the risk level based on historical risk events.

[0095] Specifically, the probability of failure is first divided into corresponding risk levels according to a preset probability range. This classification method is compatible with the existing safety classification system of nuclear power plants, making it easier for operation and maintenance personnel to quickly determine the urgency of current equipment and allocate maintenance resources.

[0096] However, risk classification based solely on the current failure probability may have limitations, as equipment failures often have cumulative and recurring characteristics. To overcome this issue, historical risk events are introduced to correct the initial risk level. Specifically, the system stores risk level records for each moment over a past period (e.g., the last 7 days). When calculating the final risk level for the current moment, it considers not only the current failure probability mapping but also the frequency, severity, and time decay factors of recent medium- or high-risk events.

[0097] The aforementioned method for detecting faults in nuclear power plant hoisting equipment effectively eliminates interference from sensor noise, missing values, and dimensional differences by collecting time-series data from multiple monitoring dimensions of the hoisting equipment and preprocessing and standardizing the raw data. This provides a high-quality data foundation for subsequent analysis. Furthermore, it extracts temporal trend features reflecting equipment degradation and spatial coupling features revealing the relationships between various components, overcoming the limitations of traditional single-dimensional monitoring. These spatiotemporal features are input into a Long Short-Term Memory (LSTM) network, which uses its forget gate, input gate, and output gate to dynamically and selectively memorize and update historical states and current features. This accurately captures the long-term dependence and transient changes in equipment operating status, outputting a high-confidence fault probability and enabling early fault prediction rather than post-event alarms. Further, the fault probability is mapped to a quantified risk level, and historical risk events are introduced for correction. This ensures that the risk assessment reflects both the current situation and the cumulative impact of recent medium- to high-risk events, preventing recurring potential faults from being overlooked. This significantly shortens fault location and troubleshooting time, reduces delay risks on critical overhaul paths, and comprehensively improves the safety and efficiency of nuclear power plant hoisting equipment operation and maintenance.

[0098] In one exemplary embodiment, preprocessing the collected runtime data includes:

[0099] The three-standard-deviation criterion is used to identify outliers in each monitoring dimension. Values ​​that deviate from the mean by more than three standard deviations are judged as outliers and corrected accordingly.

[0100] The data for each monitoring dimension after correction are standardized and mapped to a preset range to obtain standardized effective data.

[0101] Specifically, the three-standard-deviation criterion is used in the preprocessing stage to identify outliers in each monitoring dimension. This criterion is based on the statistical principle of normal distribution: for data following a normal distribution, approximately 99.7% of the samples will fall within the range of the mean plus or minus three standard deviations. Therefore, values ​​exceeding this range are highly likely to be anomalies caused by abnormal factors such as sensor malfunction, communication interference, or accidental shocks, rather than reflecting the true operating status of the equipment. After identifying outliers, corrections are needed to avoid interfering with subsequent analysis. Correction methods can include replacing them with the upper limit (130A) or lower limit (70A), or interpolating using the mean of the preceding and following time points. In this way, peaks or troughs in the original data are effectively suppressed, making the data sequence smoother and more accurately reflecting the actual operating status of the equipment.

[0102] After correcting outliers, the corrected data for each monitoring dimension are standardized to eliminate incomparability caused by differences in units and numerical ranges between different monitoring dimensions. Standardization maps the data of each dimension to a unified interval [0, 1] through a linear transformation. The transformation formula is: the standardized value equals (original value minus the minimum value of the dimension) divided by (maximum value minus the minimum value of the dimension). Taking motor current as an example, assuming its minimum value is 50A and its maximum value is 150A, then a measured value of 100A becomes (100-50) / (150-50)=0.5 after standardization; for temperature signals, if the minimum value is 20℃ and the maximum value is 80℃, then the standardized value corresponding to 50℃ is (50-20) / (80-20)=0.5. Standardizing brings two dimensions with originally vastly different numerical ranges to the same order of magnitude (0 to 1), allowing subsequent feature extraction and model training to treat each monitoring dimension equally and avoiding unit bias.

[0103] Among them, the 3σ criterion is used to identify outliers, and the formula is as follows:

[0104] ;

[0105] in: (No. (dimensional mean) (No. (dimensional standard deviation) This is a sign function used to correct outliers to a reasonable range, preventing abnormal data from interfering with subsequent algorithms.

[0106] Data standardization: Eliminating the influence of different monitoring dimensions by mapping data to a preset interval, generally [0,1], as shown in the following formula:

[0107] ;

[0108] in: , The first Minimum and maximum values ​​of a dimension This represents the value of the i-th monitoring dimension at time t after outlier correction. The standardized dataset facilitates subsequent feature extraction and model training.

[0109] In this embodiment, outliers are identified and corrected using the three-standard-deviation criterion, effectively eliminating the interference of sensor noise and signal jumps on the authenticity of the data. Furthermore, standardization maps the data of each dimension to a preset range, eliminating the impact of differences in units and inconsistent numerical ranges. This allows subsequent feature extraction and model training to treat each monitoring dimension equally, significantly improving the accuracy of fault detection and the reliability of data processing.

[0110] In one exemplary embodiment, extracting time trend features based on valid data includes:

[0111] A sliding window is used to calculate the trend slope of a preset number of time steps before the current moment, and the trend slope is used as the time trend feature; the length of the sliding window corresponds to the number of time steps of a preset duration.

[0112] Specifically, a sliding window mechanism is first employed to capture the local variation patterns of a single monitoring dimension along the time axis. For each current time t, the value at that moment is not viewed in isolation; instead, data from several consecutive time steps preceding that moment is extracted as a window. The evolution direction of the equipment state within this local time interval is characterized by analyzing the changing trends of the data within the window. Specifically, the length w of the sliding window corresponds to a preset number of time steps. For example, based on the operating characteristics and sampling cycle of the hoisting equipment, the window length can be set to correspond to a 30-minute time span. Assuming a data sampling cycle of once per minute, then w = 30, meaning each window contains 30 consecutive sampling points; if the sampling cycle is 1 second, then the window contains 1800 sampling points.

[0113] The trend slope of the data over the previous w time steps is calculated using a sliding window to capture the temporal variation pattern of the data. The formula is as follows:

[0114] ;

[0115] Where w is the length of the sliding window. Combining the operating characteristics of the nuclear power plant's ring crane and airlock, w=30 (corresponding to 30 minutes of data) can accurately capture the short-term changing trend of the equipment's operating status and provide timing support for fault early warning.

[0116] Trend fitting of this data segment typically employs the least squares method to calculate the slope of a linear regression, which is then used as the time trend characteristic at the current moment. For example, regarding the current dimension of a hoisting motor, if the standardized current values ​​of the most recent 30 sampling points (corresponding to 30 minutes) show a gradual upward trend, the calculated trend slope will be positive, and a larger absolute value of the slope indicates a faster rate of increase; conversely, if the current gradually decreases, the slope is negative; if the current fluctuates steadily, the slope is close to zero. For instance, assuming the hoist frequently performs heavy-load lifting operations during a major overhaul, and the motor current slowly increases from 0.3 (after standardization) to 0.6 within 30 minutes, the calculated slope is approximately 0.01 (an increase of 0.01 per minute). This positive slope characteristic reflects a continuously increasing motor load, potentially indicating poor lubrication or increased mechanical friction. For example, regarding the brake temperature dimension, if the temperature within the window remains constant and the slope is close to zero, it indicates a good thermal equilibrium. If the temperature slope suddenly changes from positive to negative and the absolute value is large, it may mean that the brake has stopped working or the cooling system has abnormally intervened.

[0117] The time trend features extracted in this way are essentially a compressed representation of the original time-series data. They condense the overall direction and rate of change of multiple data points within a window into a single slope value. Compared to directly using the original data or simple differencing, the sliding window slope effectively suppresses short-term random fluctuations and highlights physically meaningful trend changes. Furthermore, since the sliding window moves forward one position at each time step, a corresponding time trend feature is obtained at each moment, forming a continuous sequence of trend features for further analysis by the Long Short-Term Memory (LSTM) network.

[0118] In this embodiment, the trend slope within a preset time period before the current moment is calculated using a sliding window as a time trend feature. This effectively suppresses short-term random fluctuations and highlights the true evolution direction and rate of equipment parameters on the time axis. This feature allows the system to detect early signs of faults based on the continuously rising or falling trend slope before values ​​such as current and temperature reach alarm thresholds. This provides key inputs rich in dynamic change patterns for the subsequent long short-term memory network, thereby improving the early warning capability of faults.

[0119] In one exemplary embodiment, spatial coupling features are extracted based on effective data, including:

[0120] Calculate the mutual information between different monitoring dimensions. Mutual information is used to characterize the degree of interdependence between two monitoring dimensions.

[0121] Based on the preset weight coefficients of each monitoring dimension, the calculated mutual information from multiple sets is weighted and summed to obtain the spatial coupling characteristics.

[0122] Specifically, mutual information measures the degree of interdependence between two random variables, that is, how much knowing the value of one variable reduces the uncertainty of the other. Unlike ordinary correlation coefficients, mutual information can not only capture linear relationships but also detect non-linear statistical dependencies. In its calculation, for any two monitored dimensions i and j, the system uses standardized valid data to statistically analyze the joint probability distribution and individual marginal probability distributions of the two dimensions at the sampling points, and then performs the calculation according to the definition of mutual information. The value of mutual information is usually non-negative; a larger value indicates a stronger interdependence between the two dimensions. If the mutual information is zero, it indicates that the two dimensions are statistically independent.

[0123] After obtaining the mutual information of all dimension pairs, the calculated mutual information is further weighted and summed according to the preset weight coefficients of each monitoring dimension to obtain a single spatial coupling feature value. The purpose of introducing weight coefficients is to reflect the differences in the degree of influence of different monitoring dimensions on equipment failure. The preset weight coefficients can be based on the experience and knowledge of domain experts, or they can be calibrated through principal component analysis or sensitivity analysis of historical failure data. The final calculated spatial coupling feature is a scalar value that comprehensively reflects the overall dependency structure between all monitoring dimensions of the hoisting equipment at the current moment. When the equipment is in normal operation, the coupling mode between the dimensions is maintained in a long-term stable manner, and the spatial coupling feature value usually fluctuates within a relatively fixed range; however, when a component fails or its performance degrades, the originally normal coupling relationship may be disrupted (e.g., the loss of encoder signal leads to a decrease in its mutual information with motor speed), or new abnormal coupling may occur (e.g., the correlation between vibration and current suddenly increases), and the spatial coupling feature value will change significantly.

[0124] The formula for calculating the spatial coupling characteristics is as follows:

[0125] ;

[0126] The formula for calculating mutual information is as follows:

[0127] .

[0128] P() represents the probability distribution; W i W j These are the weight coefficients for the i-th and j-th dimensions, respectively. The weights are calibrated based on the degree of influence of each dimension on the fault, ensuring the specificity of the coupling features.

[0129] In this embodiment, by calculating and weighting the mutual information between different monitoring dimensions, the linear and nonlinear coupling relationships between various components of the hoisting equipment can be effectively captured. Simultaneously, based on preset weights, the contribution of key dimensions to fault detection is highlighted. This spatial coupling feature reveals the overall collaborative state of the system. When a component fails or its performance degrades, disrupting the normal coupling relationship, the feature value changes significantly. This provides crucial information reflecting multi-dimensional correlation anomalies for subsequent fault prediction, improving the comprehensiveness and accuracy of fault detection.

[0130] In one exemplary embodiment, such as Figure 3 As shown, the corresponding risk level is obtained based on the failure probability mapping, and the risk level is corrected based on historical risk events, including:

[0131] Step S302: According to the preset mapping relationship between multiple probability intervals and risk levels, the failure probability is converted into an initial risk level; in the mapping relationship, the failure probability and the corresponding risk level are directly proportional.

[0132] Step S304: Obtain the historical risk levels for multiple consecutive days prior to the current moment, and count the number of days that reached the medium risk level or above;

[0133] Step S306: Calculate the risk correction factor based on the statistically obtained number of days and the preset historical risk impact coefficient, and use the product of the initial risk level and the risk correction factor as the final risk level; wherein, the closer the historical risk event is to the current time, the greater its weight in the correction factor.

[0134] Specifically, several consecutive and non-overlapping probability intervals are pre-defined, each corresponding to a risk level. The probability of failure is directly proportional to the risk level; that is, the higher the probability of failure, the higher the risk level. For example, five risk levels can be defined: level 1 (no risk) is mapped to a probability below 0.1; level 2 (low risk) is between 0.1 and 0.3; level 3 (medium risk) is between 0.3 and 0.5; level 4 (high risk) is between 0.5 and 0.8; and level 5 (extremely high risk) is mapped to a probability not lower than 0.8. This classification method is compatible with existing safety classification standards for nuclear power plants, facilitating maintenance personnel's quick understanding of the urgency of current equipment needs.

[0135] However, risk classification based solely on the current failure probability has limitations because equipment failures often have cumulative and recurring characteristics. For example, a brake contactor might have experienced multiple medium-risk alarms in the past week, but the failure probability would briefly drop after each alarm. Looking only at the current low probability value might misjudge a safe state, thus ignoring potential gradual degradation trends. To overcome this problem, we obtain historical risk level records for each day of the preceding consecutive days (e.g., 7 consecutive days) and count the number of days reaching medium risk (level 3) or higher. This statistic reflects the equipment's recent "poor record": the more frequent the historical medium-to-high risk events, the greater the likelihood of recurring anomalies or continuous degradation, and a positive correction should be applied to the current risk level. For example, if 3 out of the past 7 days had medium-to-high risk events, the count is 3 days; if there were no medium-to-high risk events in the past 7 days, the count is 0 days.

[0136] in, α=0.2 (historical risk impact coefficient), M=7 (the last 7 days, adapted to the overhaul cycle of the nuclear power plant), I() is the indicator function (takes 1 if the condition is met, otherwise takes 0); this formula can correct the impact of historical high-risk failures on the current risk level, avoid ignoring recurring potential failures, and improve the accuracy of risk assessment.

[0137] After obtaining the statistical number of days, a risk correction factor is calculated based on this number of days and a preset historical risk impact coefficient. The more high-risk events in the history, the larger the correction factor. Simultaneously, historical risk events more recent to the current time have a greater weight in the correction factor, while the weight of earlier events gradually decreases. In practice, a weighting method similar to exponential decay or harmonic series can be used. Through this correction, equipment that has a currently low probability of failure but has repeatedly exhibited anomalies recently will not be easily judged as safe, thus effectively avoiding missed assessments due to instantaneous fluctuations in failure probability, and improving the robustness and practical guiding significance of risk assessment.

[0138] In this embodiment, the failure probability is mapped to an initial risk level in a direct proportional relationship. Then, the number of occurrences of medium- and high-risk events over several consecutive days is statistically analyzed, and a correction factor is calculated using a weighting principle where closer events have greater weight. Finally, the initial risk level is multiplied by the correction factor to obtain the final risk level. This method effectively integrates the current instantaneous failure probability with the cumulative effect of historical risks, avoiding misjudgments or omissions caused by probability fluctuations. This allows for the timely upgrading of risk levels for equipment exhibiting recurring anomalies, providing maintenance personnel with a more accurate and robust decision-making basis.

[0139] In one exemplary embodiment, the method further includes:

[0140] Map the final risk level to a visual color parameter;

[0141] Based on the weight coefficients, failure probabilities, and standardized valid data of each monitoring dimension, the visualization weights corresponding to each monitoring dimension are calculated, and the corresponding components of the hoisting equipment are highlighted differently according to their visualization weights.

[0142] Specifically, since the final risk level is a discrete value (e.g., level 1 to 5), and color in the human-machine interface is the most intuitive way to convey risk, the system establishes a color mapping rule: risk levels from low to high correspond to a color spectrum from green to red. Specifically, when the final risk level is level 1 (no risk), the corresponding color parameter is set to pure green; when it is level 5 (extremely high risk), it is set to pure red; and levels 2, 3, and 4 correspond to transitional colors between green and red, such as yellowish-green, yellow, and orange-red. This color mapping allows maintenance personnel to immediately determine the overall risk status of the equipment based on color when scanning the monitoring interface, without needing to read specific numerical values.

[0143] in, Level 1 risk corresponds to green (255,0,0), Level 5 risk corresponds to red (0,255,0), and the intermediate levels show a gradient change, which makes it easy for maintenance personnel to quickly judge the risk level of the equipment by color.

[0144] Building upon the overall risk color mapping described above, a differentiated highlighting mechanism is implemented for each monitoring dimension (i.e., each component or physical quantity) to assist maintenance personnel in quickly locating specific fault sources. Specifically, the system calculates the visualization weight for each monitoring dimension based on preset weight coefficients, the current fault probability, and standardized valid data. The weight coefficients reflect the importance of different dimensions in influencing equipment failure; for example, motor current and encoder readings typically have higher weights than ambient temperature. The fault probability is the value output by the Long Short-Term Memory network, representing the overall likelihood of equipment failure. The standardized valid data reflects the position of the current measured value relative to the normal range for that dimension. The visualization weight is obtained by multiplying these three factors; dimensions with higher importance, greater current fault probability, and more deviations from the normal range have higher visualization weights.

[0145] in, Weight W {\text{vis}}(i,t) The higher the value, the more prominently the corresponding components of the hoist and airlock are highlighted in the visualization interface, enabling rapid location of the fault and assisting maintenance personnel in efficiently troubleshooting.

[0146] After calculating the visualization weights for each dimension, the corresponding components of the hoisting equipment are highlighted differently according to their weights. Highlighting methods may include, but are not limited to: changing the border thickness of the component icon, adjusting the saturation of the fill color, adding a flashing effect, or displaying a dynamic indicator bar next to the component. Components with higher weights are highlighted more prominently, thus guiding maintenance personnel to focus their attention first on the location most likely to experience a fault.

[0147] In this embodiment, by mapping the final risk level to a color parameter ranging from green to red, an intuitive visual communication of risk levels is achieved. Simultaneously, visualization weights are calculated based on dimensional weights, failure probabilities, and standardized data, and each component of the ring crane is highlighted differently according to its weight, making key failure sources prominently displayed on the interface. This method guides maintenance personnel to quickly locate high-risk components, reduces data retrieval time, and significantly improves fault diagnosis efficiency and the intuitiveness of human-computer interaction.

[0148] In one exemplary embodiment, the method further includes:

[0149] Full life cycle monitoring is performed on at least one of the loss mechanisms in the hoisting equipment, including frequency converter, brake contactor, motor, and holding brake. The running time and number of actions are recorded, and an early warning signal is output when the running time or number of actions reaches a preset threshold.

[0150] In addition, the on / off status of the power circuit circuit breaker and contactor is displayed in the form of a single-line diagram, and the on / off status of each network node in the DP bus and Ethernet is displayed in real time in the form of a node list.

[0151] Specifically, lifting equipment contains components that are subject to high workload, frequent operation, and are crucial for safe operation, such as frequency converters, brake contactors, motors, and holding brakes. These components gradually experience mechanical wear or electrical aging with accumulated operating time and number of operations, and their failure probability is often positively correlated with cumulative usage. Therefore, for at least one of these wear-prone mechanisms, the operating time (e.g., cumulative motor operating hours) and number of operations (e.g., the number of times the brake contactor engages and disengages, and the number of times the holding brake opens and closes) are continuously recorded. Based on this, the system presets thresholds corresponding to the equipment's design life or maintenance procedures. When the monitored operating time or number of operations reaches or exceeds these thresholds, an early warning signal is automatically output.

[0152] On the other hand, the on / off status of the main circuit circuit breakers and contactors in the power circuit is presented intuitively in the form of a single-line diagram. A single-line diagram is a simplified electrical wiring diagram that uses a single line to represent a three-phase circuit and labels the corresponding locations with graphic symbols for circuit breakers and contactors. By mapping real-time acquired switching status (such as the closing / opening signals of circuit breaker auxiliary contacts) onto the single-line diagram, maintenance personnel can clearly see the current status of each switching device on the monitoring interface. This graphical display method allows maintenance personnel to quickly locate fault points or abnormal states in the power circuit without referring to complex electrical drawings or going to the site for inspection, significantly improving fault diagnosis efficiency.

[0153] Meanwhile, the on / off status of each network node in the DP bus and Ethernet is displayed in real time as a node list. The control system of a hoisting machine typically uses PROFIBUS-DP (DP bus for short) as the communication network between field devices and the master station, while Ethernet is used to exchange data between the PLC and the upper-level monitoring system. These networks connect multiple nodes, such as frequency converters, encoders, remote I / O modules, and HMI control panels. If any network node experiences a disconnection, communication timeout, or configuration error, it may cause partial malfunction of the hoisting machine control system. This method uses network status monitoring to periodically send heartbeat detection or read diagnostic information to each node, and presents the online / offline status of each node in a list format on a visual interface.

[0154] In this embodiment, by monitoring the operating time and number of actions of loss-generating mechanisms such as frequency converters and brake contactors throughout their entire life cycle, early warnings can be issued when preset thresholds are reached, enabling preventative maintenance. At the same time, the on / off status of power circuit breaker and contactor is displayed intuitively with a single-line diagram, and the on / off status of DP bus and Ethernet nodes is displayed in real time with a node list, facilitating rapid location of electrical faults and network anomalies.

[0155] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0156] Based on the same inventive concept, this application also provides a nuclear power plant ring crane fault detection visualization system for implementing the aforementioned nuclear power plant ring crane fault detection method. The solution provided by this system is similar to the implementation scheme described in the above method; therefore, the specific limitations of one or more embodiments of the nuclear power plant ring crane fault detection visualization system provided below can be found in the limitations of the nuclear power plant ring crane fault detection method described above, and will not be repeated here.

[0157] In one exemplary embodiment, such as Figure 4 As shown, a nuclear power plant ring crane fault detection visualization system is provided, including:

[0158] The data acquisition module 402 is used to establish communication with the PLC of the hoisting equipment and collect the operating data of the hoisting equipment. The operating data includes time-series data of multiple monitoring dimensions.

[0159] The feature extraction module 404 is used to preprocess the collected operational data to obtain standardized effective data, and extract time trend features and spatial coupling features based on the effective data; among them, the time trend features are used to characterize the data change pattern of a single monitoring dimension within a time window, and the spatial coupling features are used to characterize the correlation between different monitoring dimensions.

[0160] The fault prediction and risk assessment module 406 is used to input the time trend features and spatial coupling features into the long short-term memory network, selectively memorize and update the historical state and the current input through the forget gate, input gate and output gate of the long short-term memory network, and output the fault probability of the device at the current moment; the corresponding risk level is obtained according to the fault probability mapping, and the risk level is corrected based on historical risk events, and the corrected final risk level is output.

[0161] The visualization module 408 is used to generate and display a visualization interface containing equipment operating status, fault information and risk level based on the final risk level and operating data.

[0162] like Figure 5 As shown, without changing the original PLC program, by collecting data from the PLC program and using host computer software to program and process the data, all alarm information and related signal screen information of the 20-meter platform can be seen on the touch screen in the electrical room. On the basis of this information, the working frequency records of each mechanism brake, the speed and torque curves and speed and torque values ​​of each mechanism, the encoder trigger values, trigger times and trigger times of each lifting mechanism encoder fault, effective rotation fault, and overspeed fault, etc. can be exported as files for viewing and analysis.

[0163] The PLC communicates with the host system via Ethernet TCP / IP. The PLC system consists of an OPLC and a PTMS, while the host system uses classic WinCC software. The OPLC's IP address is 192.168.0.1, and the PTMS's IP address is 192.168.0.2. Data variables for communication with the OPLC and PTMS are established in the WinCC host system, containing nearly a thousand word variables and thousands of bit variables. Through programming of this data information, displays such as data curves, numerical values, calculation information, archived information, count information, and fault information are generated.

[0164] In an exemplary embodiment, the data acquisition module 402 establishes communication with the PLC of the hoisting equipment via the Ethernet TCP / IP protocol. The PLC includes OPLC and PTMS. The operating data collected by the data acquisition module 402 includes: current, voltage, speed, and torque of the drive device, limit switch status, encoder reading, inverter operating parameters, and the operating frequency, cumulative working time, and number of actions of the brake.

[0165] In one exemplary embodiment, the visualization presentation module 408 includes:

[0166] The fault alarm display unit is used to display fault codes and fault alarm records in real time in the form of a text list;

[0167] The trend curve display unit is used to provide the display and export functions of real-time trend curves and historical trend curves of each monitored variable;

[0168] The status monitoring and display unit is used to display the on / off status of the power circuit circuit breaker and contactor in the form of a single-line diagram, and to display the on / off status of each network node in the DP bus and Ethernet in the form of node status.

[0169] The brake status display unit is used to display the real-time status, operating frequency, and cumulative operating time of the brake.

[0170] The parameter input unit provides an interface for users to input preset thresholds to support the setting of early warning conditions in the whole life cycle monitoring.

[0171] The modules in the aforementioned nuclear power plant hoist fault detection visualization system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in the processor of a computer device in hardware form or independent of it, or stored in the memory of the computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0172] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 6As shown, this computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores data. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network. When the computer program is executed by the processor, it implements a method for detecting faults in a nuclear power plant's overhead crane.

[0173] Those skilled in the art will understand that Figure 6 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0174] In one exemplary embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of the method described above.

[0175] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described method.

[0176] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps of the method described above.

[0177] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0178] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0179] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0180] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for detecting faults in a nuclear power plant's ring crane, characterized in that, The method includes: Collect operational data of the ring crane equipment, which includes time-series data from multiple monitoring dimensions; The collected operational data is preprocessed to obtain standardized effective data, and time trend features and spatial coupling features are extracted based on the effective data; wherein, the time trend features are used to characterize the data change pattern of a single monitoring dimension within a time window, and the spatial coupling features are used to characterize the correlation between different monitoring dimensions; The time trend features and spatial coupling features are input into the Long Short-Term Memory (LSTM) network. Through the forget gate, input gate, and output gate of the LSM network, historical state data and currently input feature data are selectively memorized and updated, and the failure probability of the device at the current moment is output. The corresponding risk level is obtained based on the failure probability mapping, and the risk level is corrected based on historical risk events.

2. The method according to claim 1, characterized in that, The preprocessing of the collected operational data includes: The three-standard-deviation criterion is used to identify outliers in each monitoring dimension. Values ​​that deviate from the mean by more than three standard deviations are judged as outliers and corrected accordingly. The data for each monitoring dimension after correction are standardized and mapped to a preset range to obtain the standardized effective data.

3. The method according to claim 1, characterized in that, The extraction of time trend features based on the effective data includes: A sliding window is used to calculate the trend slope of a preset number of time steps prior to the current moment, and the trend slope is used as the time trend feature; the length of the sliding window corresponds to the number of time steps of a preset duration.

4. The method according to claim 1, characterized in that, The extraction of spatial coupling features based on the effective data includes: Calculate the mutual information between different monitoring dimensions, whereby the mutual information is used to characterize the degree of interdependence between two monitoring dimensions; Based on the preset weight coefficients of each monitoring dimension, the calculated mutual information from multiple sets is weighted and summed to obtain the spatial coupling feature.

5. The method according to claim 1, characterized in that, The step of obtaining the corresponding risk level based on the failure probability mapping and correcting the risk level based on historical risk events includes: According to a preset mapping relationship between multiple probability intervals and risk levels, the failure probability is converted into an initial risk level; in the mapping relationship, the failure probability and the corresponding risk level are directly proportional. Obtain the historical risk levels for multiple consecutive days prior to the current moment, and count the number of days that reached the medium risk level or above; Based on the number of days obtained from statistics and the preset historical risk impact coefficient, a risk correction factor is calculated, and the product of the initial risk level and the risk correction factor is taken as the final risk level; wherein, the more recent the historical risk event is, the greater its weight in the correction factor.

6. The method according to claim 5, characterized in that, The method further includes: The final risk level is mapped to a visual color parameter; Based on the weight coefficients of each monitoring dimension, the failure probability, and the standardized valid data, the visualization weight corresponding to each monitoring dimension is calculated, and the corresponding components of the hoisting equipment are highlighted differently according to the level of the visualization weight.

7. The method according to claim 1, characterized in that, The method further includes: The system performs full lifecycle monitoring on at least one of the following components in the hoisting equipment: frequency converter, brake contactor, motor, and holding brake. It records the running time and number of actions, and outputs an early warning signal when the running time or number of actions reaches a preset threshold. In addition, the on / off status of the power circuit circuit breaker and contactor is displayed in the form of a single-line diagram, and the on / off status of each network node in the DP bus and Ethernet is displayed in real time in the form of a node list.

8. A visualization system for detecting faults in a nuclear power plant's ring crane, characterized in that, The system includes: The data acquisition module is used to establish communication with the PLC of the hoisting equipment and collect the operating data of the hoisting equipment, which includes time-series data of multiple monitoring dimensions. The feature extraction module is used to preprocess the collected operational data to obtain standardized effective data, and extract time trend features and spatial coupling features based on the effective data; wherein, the time trend features are used to characterize the data change pattern of a single monitoring dimension within a time window, and the spatial coupling features are used to characterize the correlation between different monitoring dimensions; The fault prediction and risk assessment module is used to input the time trend features and the spatial coupling features into the long short-term memory network, selectively memorize and update the historical state and the current input through the forget gate, input gate and output gate of the long short-term memory network, and output the fault probability of the device at the current moment; the corresponding risk level is obtained according to the fault probability mapping, and the risk level is corrected based on historical risk events, and the corrected final risk level is output. The visualization module is used to generate and display a visualization interface containing equipment operating status, fault information and risk level based on the final risk level and the operating data.

9. The visualization system according to claim 8, characterized in that, The data acquisition module is used to establish communication with the PLC of the hoisting equipment via the Ethernet TCP / IP protocol. The PLC includes OPLC and PTMS. The operating data collected by the data acquisition module includes: current, voltage, speed, torque of the drive device, limit switch status, encoder reading, frequency converter operating parameters, and brake operating frequency, cumulative working time and number of actions.

10. The visualization system according to claim 8, characterized in that, The visualization module includes: The fault alarm display unit is used to display fault codes and fault alarm records in real time in the form of a text list; The trend curve display unit is used to provide the display and export functions of real-time trend curves and historical trend curves of each monitored variable; The status monitoring and display unit is used to display the on / off status of the power circuit circuit breaker and contactor in the form of a single-line diagram, and to display the on / off status of each network node in the DP bus and Ethernet in the form of node status. The brake status display unit is used to display the real-time status, operating frequency, and cumulative operating time of the brake. The parameter input unit provides an interface for users to input preset thresholds to support the setting of early warning conditions in the whole life cycle monitoring.

Images