Water conservancy gate intelligent diagnosis and early warning system based on state monitoring

Abstract

The invention relates to a water conservancy gate intelligent diagnosis and early warning system based on state monitoring. The system comprises a data obtaining and preprocessing module used for obtaining operation state data in the gate operation process and preprocessing the operation state data to obtain preprocessed operation state data; the feature extraction module is used for performing feature extraction on the preprocessed operation state data to generate a dynamic feature vector; the fault anomaly analysis module is used for performing fault anomaly analysis on the dynamic feature vector in combination with a preset fault anomaly model to obtain a fault anomaly analysis result; the fault index module is used for calculating a fault severity index by combining a preset fault library according to the fault abnormity analysis result; and the fault grading early warning module is used for generating a fault early warning signal according to the fault severity index. By adopting the system, the adaptability and accuracy of diagnosis and early warning can be improved, the gate operation safety is guaranteed, and fault hidden dangers are reduced.

Description

Technical Field

[0001] This invention belongs to the field of water conservancy engineering technology, and in particular relates to an intelligent diagnosis and early warning system for water conservancy gates based on condition monitoring. Background Technology

[0002] Water conservancy projects are crucial infrastructure for ensuring water resource allocation, flood control, drought relief, and ecological protection. As a core component of water conservancy projects, the operational status of gates directly impacts the safety and reliability of the entire project. With the widespread application of modern water conservancy facilities, intelligent monitoring and fault diagnosis of gates have become key to ensuring the long-term stable operation of these projects.

[0003] However, traditional monitoring methods often fail to meet the demands for real-time performance and accuracy under complex operating conditions, revealing significant technical bottlenecks and necessitating innovative methods to improve diagnostic capabilities. Currently, gate condition monitoring mainly relies on the individual analysis of vibration signals, displacement data, or torque data, but these methods have limitations when dealing with complex faults. The analysis of a single signal cannot fully reflect the multidimensional dynamic characteristics of gate operation. For example, complex faults such as wear in the transmission system or abnormal gear meshing often require a comprehensive analysis of the dynamic changes of multiple signals for accurate diagnosis.

[0004] Furthermore, existing technologies lack in-depth analysis of the inertial patterns during gate movement when extracting signal features, resulting in insufficient ability to identify early signs of faults and a tendency to miss critical warning opportunities. During gate operation, the inertial characteristics of motion are one of the core technical attributes for fault diagnosis. Inertial characteristics reflect the dynamic patterns of acceleration changes, velocity fluctuations, and position responses during gate opening and closing, which are directly related to the health status of components such as the transmission system, bearings, and gears. However, accurately extracting these inertial characteristics faces technical challenges: First, the gate is affected by water flow, load, and environmental interference during opening and closing, resulting in signals containing a large amount of noise, making it difficult to accurately separate the inertial parameters reflecting faults. Second, the correspondence between the dynamic changes of inertial parameters and fault types is complex, lacking a systematic feature library to support rapid and accurate fault location. For example, in actual operation, abnormal gear meshing may cause abnormal acceleration fluctuations, but these fluctuations may be masked by water flow disturbances, making it difficult to accurately determine the fault location. Summary of the Invention

[0005] Therefore, it is necessary to provide a condition-monitoring-based intelligent diagnosis and early warning system for hydraulic gates that can improve the accuracy of fault location, addressing the aforementioned technical problems.

[0006] Firstly, this application provides an intelligent diagnostic and early warning system for hydraulic gates based on condition monitoring, including:

[0007] The data acquisition and preprocessing module is used to acquire the operating status data during the gate's operation and preprocess the operating status data to obtain preprocessed operating status data; the operating status data includes vibration acceleration signals, displacement signals, and motor torque signals;

[0008] The feature extraction module is used to extract features from the preprocessed operating status data and generate dynamic feature vectors; the dynamic feature vectors are used to reflect the motion inertia of the gate.

[0009] The fault anomaly analysis module is used to perform fault anomaly analysis on dynamic feature vectors by combining a preset fault anomaly model, and obtain fault anomaly analysis results. The fault anomaly analysis results include fault type, confidence level and fault time.

[0010] The fault index module is used to calculate the fault severity index based on the fault anomaly analysis results and a pre-set fault database; the fault severity index is used to characterize the severity of the fault anomaly analysis results.

[0011] The fault classification and early warning module is used to generate fault warning signals based on the fault severity index; the fault warning signals are used to characterize the degree of fault in the gate's operating status.

[0012] In one embodiment, the data acquisition and preprocessing module is further configured to:

[0013] Collect operational status data during the gate's operation;

[0014] Wavelet threshold denoising is performed on the vibration acceleration signal to obtain the denoised vibration acceleration signal;

[0015] The displacement signal is smoothed by Kalman filtering to obtain a smoothed displacement signal;

[0016] The motor torque signal is subjected to median filtering to obtain the filtered motor torque signal.

[0017] The noise-reduced vibration acceleration signal, the smoothed displacement signal, and the filtered motor torque signal are time-aligned to obtain preprocessed operating status data.

[0018] In one embodiment, the feature extraction module is further configured to:

[0019] Time-domain feature extraction is performed on the vibration acceleration signal in the preprocessed operational status data to obtain a time-domain feature vector;

[0020] Frequency domain feature extraction can be performed on the vibration acceleration signal in the preprocessed operating status data to obtain the frequency domain feature vector;

[0021] The kinematic parameters of the displacement signals in the preprocessed running status data are calculated to obtain the kinematic feature vectors.

[0022] A fluctuation analysis was performed on the motor torque signal in the preprocessed operating status data to obtain the torque fluctuation feature vector;

[0023] The time-domain feature vector, frequency-domain feature vector, kinematic feature vector, and torque fluctuation feature vector are concatenated to generate a dynamic feature vector.

[0024] In one embodiment, the fault anomaly analysis module is further configured to:

[0025] The dynamic feature vector is input into the preset fault anomaly model, and the fault type is identified by the support vector machine classifier in the fault anomaly model.

[0026] The confidence level of the fault type is calculated using the following formula through the probability output layer in the fault anomaly model:

[0027]

[0028] in, For confidence level, For the weight vector, For dynamic feature vectors, For bias terms;

[0029] Based on the timestamp of the dynamic feature vector, the time when the fault type occurred is determined, and the fault time is obtained;

[0030] Based on the fault type, confidence level, and fault time, the fault anomaly analysis results are obtained.

[0031] In one embodiment, the fault index module is further configured to:

[0032] Based on the fault type, the corresponding weight coefficient is retrieved from the preset fault database;

[0033] The basic severity index is calculated based on confidence level and weighting coefficients;

[0034] Based on the failure time, the time decay factor is calculated using the following formula:

[0035]

[0036] in, The time decay factor, The attenuation rate, For the current time, For the time of failure;

[0037] The fault severity index is obtained by combining the basic severity index and the time decay factor.

[0038] In one embodiment, the fault classification and early warning module is further used for:

[0039] Based on preset warning threshold rules, the fault severity index is classified to obtain the fault warning level;

[0040] Based on the fault warning level, the corresponding warning strategy is matched from the preset warning strategy library to obtain the target warning strategy; the target warning strategy includes the frequency of warning notification, the notification target and the notification method.

[0041] Based on the target early warning strategy and the results of fault anomaly analysis, an early warning signal is generated.

[0042] In one embodiment, the system further includes a model self-updating module for:

[0043] Regularly collect fault events and corresponding operational status data to create new samples;

[0044] Add new samples to the training sample set to incrementally train the fault anomaly model and obtain the updated fault anomaly model.

[0045] Based on the actual evolution of faults in the newly added samples, the weight coefficients of the corresponding fault types in the fault database are dynamically adjusted to obtain the updated fault database.

[0046] Secondly, this application also provides a method for intelligent diagnosis and early warning of hydraulic gates based on condition monitoring, including:

[0047] The system acquires operational status data during the gate's operation and preprocesses this data to obtain preprocessed operational status data. The operational status data includes vibration acceleration signals, displacement signals, and motor torque signals.

[0048] Feature extraction is performed on the preprocessed operating status data to generate dynamic feature vectors; the dynamic feature vectors are used to reflect the motion inertia of the gate.

[0049] Based on a pre-defined fault anomaly model, fault anomaly analysis is performed on the dynamic feature vector to obtain fault anomaly analysis results; the fault anomaly analysis results include fault type, confidence level, and fault time.

[0050] Based on the fault anomaly analysis results and combined with the preset fault database, the fault severity index is calculated; the fault severity index is used to characterize the severity of the fault anomaly analysis results.

[0051] Based on the fault severity index, a fault warning signal is generated; the fault warning signal is used to characterize the degree of fault in the gate's operating status.

[0052] 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 implement the system described in the first aspect.

[0053] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the system described in the first aspect.

[0054] The aforementioned intelligent diagnosis and early warning system for hydraulic gates based on condition monitoring effectively eliminates noise interference and time deviations through data acquisition and preprocessing modules, providing high-quality data support for subsequent analysis. A dynamic feature vector is formed through a feature extraction module, which, combined with a fault anomaly analysis module, accurately identifies fault types, quantifies confidence levels, and pinpoints fault times. A fault index module, using a pre-set fault database, calculates a fault severity index to reflect the degree of fault hazard. A fault classification and early warning module generates appropriate early warning signals and notification strategies based on the index classification, ensuring that relevant parties promptly grasp fault information and respond quickly. Simultaneously, through incremental model training and dynamic adjustment of fault database weight coefficients, the system continuously optimizes diagnostic and early warning performance, thereby improving the accuracy of gate fault identification and the timeliness of early warning, reducing potential fault risks, and ensuring stable gate operation. Attached Figure Description

[0055] To more clearly illustrate the technical solutions in the embodiments or related technologies of this application, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0056] Figure 1 This is a schematic diagram of a hydraulic gate intelligent diagnosis and early warning system based on condition monitoring in one embodiment.

[0057] Figure 2 This is a flowchart illustrating a method for intelligent diagnosis and early warning of hydraulic gates based on condition monitoring in one embodiment.

[0058] Figure 3 This is a schematic diagram of a computer device in one embodiment. Detailed Implementation

[0059] 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.

[0060] In one embodiment, such as Figure 1 As shown, a smart diagnostic and early warning system 10 for hydraulic gates based on condition monitoring is provided. This embodiment illustrates the system's application to a terminal. It is understood that this system can also be applied to a server, or to a system including both a terminal and a server, and is implemented through interaction between the terminal and the server. In this embodiment, the system includes:

[0061] The data acquisition and preprocessing module 11 is used to acquire the operating status data during the gate operation process and preprocess the operating status data to obtain preprocessed operating status data.

[0062] For example, piezoelectric accelerometers, wire displacement sensors, and torque telemetry devices are arranged on the gate body, hoist, and motor housing to synchronously collect three types of raw signals—vibration acceleration, displacement, and motor torque—during the entire gate opening and closing cycle. These collected raw signals are then transmitted to the gate early warning terminal. The operational status data includes vibration acceleration, displacement, and motor torque signals. The vibration acceleration signal reflects the dynamic response of the gate due to structural vibration during operation; the displacement signal characterizes the positional changes of the gate during opening and closing; and the motor torque signal reflects the output torque of the motor driving the gate. After acquiring the data, the gate early warning terminal preprocesses the operational status data to eliminate interference and improve data quality. By employing differentiated processing methods based on the characteristics of different signal types during preprocessing, the preprocessed operational status data more accurately reflects the actual operating status of the gate.

[0063] The feature extraction module 12 is used to extract features from the preprocessed running status data and generate dynamic feature vectors.

[0064] For example, the gate early warning terminal performs feature extraction on the pre-processed operating status data. Feature extraction is used to extract information that reflects the essential characteristics of the gate's operating status from massive amounts of raw data, transforming high-dimensional raw data into low-dimensional feature vectors. For the pre-processed vibration acceleration signal, displacement signal, and motor torque signal, the gate early warning terminal employs different feature extraction methods to extract key feature parameters from each type of signal. The terminal integrates the extracted feature parameters to generate a dynamic feature vector. This dynamic feature vector contains information on various aspects of the gate's operation, including vibration characteristics, displacement variation patterns, and motor torque fluctuations, comprehensively reflecting the gate's motion inertia and improving the targeting and accuracy of fault diagnosis.

[0065] The fault anomaly analysis module 13 is used to perform fault anomaly analysis on dynamic feature vectors in combination with a preset fault anomaly model to obtain fault anomaly analysis results.

[0066] For example, the gate early warning terminal combines a preset fault anomaly model to conduct fault anomaly analysis on dynamic feature vectors. The preset fault anomaly model is a model trained using machine learning algorithms based on a large amount of gate fault sample data. It is used to identify fault types, calculate confidence levels, and determine fault times. The fault anomaly model includes functional modules such as fault classification and probability calculation. During the analysis, the gate early warning terminal first inputs the dynamic feature vectors into the fault anomaly model. The model processes the feature vectors through its internal algorithm module, first identifying the possible fault types of the gate, and then calculating the confidence level for that fault type. The confidence level is used to measure the reliability of the fault type identification result. Simultaneously, the timestamp of the dynamic feature vector is combined to determine the time of the fault occurrence. The terminal integrates the identified fault type, the calculated confidence level, and the determined fault time to form the fault anomaly analysis result. The fault anomaly analysis result includes the fault type, confidence level, and fault time.

[0067] The fault index module 14 is used to calculate the fault severity index based on the fault anomaly analysis results and a preset fault database.

[0068] The pre-built fault database is a pre-established database containing information on various gate faults. It stores information such as weight coefficients for different fault types, the scope of fault impact, and historical handling cases. The weight coefficients measure the impact of different fault types on gate operational safety; the more severe the fault type, the larger the corresponding weight coefficient. The gate early warning terminal first queries the fault database for the corresponding weight coefficient based on the fault type in the fault anomaly analysis results. Then, it calculates a basic severity index based on the confidence level of the fault type, which initially reflects the severity of the fault. The terminal then calculates a time decay factor based on the fault time. This time decay factor considers the impact of the interval between the fault occurrence time and the current time on the fault severity. The longer the fault occurs, the greater the potential risk may be if it is not handled promptly. Finally, the terminal combines the basic severity index and the time decay factor to obtain the final fault severity index. The fault severity index comprehensively reflects the current severity of the fault.

[0069] The fault classification and early warning module 15 is used to generate fault early warning signals based on the fault severity index.

[0070] For example, the gate early warning terminal generates a fault early warning signal based on the calculated fault severity index. The purpose of generating this signal is to send corresponding warning information to relevant personnel or systems based on the severity of the fault, reminding them to take timely action to prevent the fault from escalating and affecting the gate's operational safety. The generation process requires combining preset rules and a strategy library to achieve accurate fault classification and targeted notification. The gate early warning terminal classifies the fault severity index according to preset warning threshold rules, clarifying the fault early warning level; the terminal matches the corresponding warning strategy based on the warning level, determining the notification frequency, recipients, and methods; finally, it integrates the warning strategy with the fault anomaly analysis results to generate a complete fault early warning signal, ensuring that relevant parties can obtain fault information promptly and accurately and take countermeasures.

[0071] The aforementioned intelligent diagnosis and early warning system for hydraulic gates based on condition monitoring effectively improves data quality through precise collection and differentiated preprocessing of operational status data, laying a reliable foundation for subsequent analysis. Multi-dimensional feature extraction and integration form a dynamic feature vector comprehensively reflecting the gate's operational status, enhancing the accuracy of fault identification. Combined with a fault anomaly model, it achieves comprehensive analysis of fault type, confidence level, and time, and calculates a fault severity index using a fault database, ensuring scientific and reasonable fault assessment. Based on the index, it generates tiered early warning signals and matches them with targeted strategies, ensuring relevant parties promptly grasp fault information. Simultaneously, by regularly collecting samples to optimize the model and fault database, it continuously improves the accuracy and adaptability of diagnosis and early warning, effectively ensuring gate operational safety, reducing fault risks, and improving operation and maintenance efficiency.

[0072] In an optional embodiment, the data acquisition and preprocessing module 11 is further configured to:

[0073] S111 collects the operating status data during the gate's operation.

[0074] For example, the gate early warning terminal collects operational status data through sensors deployed at key parts of the gate. Vibration acceleration signals are collected by acceleration sensors installed at key structural locations such as the gate's main beam and support arms, which can capture vibration changes during gate operation in real time. Displacement signals are obtained through gate travel sensors or laser displacement sensors, which can accurately record the gate's position information at every instant during opening and closing. Motor torque signals are collected from the gate drive motor control system to reflect the torque output of the motor driving the gate during operation. The gate early warning terminal controls each sensor to collect data synchronously according to a preset sampling frequency, ensuring that the different types of signals collected are consistent in the time dimension, avoiding data deviations caused by asynchronous collection, and providing a time-consistent data foundation for subsequent data processing and analysis.

[0075] S112, perform wavelet threshold denoising on the vibration acceleration signal to obtain the denoised vibration acceleration signal.

[0076] For example, the gate early warning terminal performs wavelet threshold denoising processing on the collected vibration acceleration signal. The principle of wavelet threshold denoising technology is to use wavelet transform to decompose the signal into wavelet coefficients of different scales. Wavelet coefficients containing effective signal information have larger amplitudes, while wavelet coefficients generated by noise have smaller amplitudes. The gate early warning terminal processes the decomposed wavelet coefficients according to preset threshold rules, setting wavelet coefficients with amplitudes less than the threshold to zero and retaining wavelet coefficients with amplitudes greater than the threshold. Then, it performs inverse wavelet transform on the processed wavelet coefficients to reconstruct the denoised vibration acceleration signal. Through denoising processing, useless information generated by environmental interference, sensor noise, and other factors in the vibration acceleration signal can be effectively removed, making the denoised vibration acceleration signal more accurately reflect the true state of the gate structure vibration and providing high-quality signal data.

[0077] S113 performs Kalman filtering on the displacement signal to smooth it, resulting in a smoothed displacement signal.

[0078] For example, the gate early warning terminal uses Kalman filtering to smooth the displacement signal. Kalman filtering, based on a state-space model, smooths the dynamic signal through two steps: prediction and update. In the prediction step, the gate early warning terminal predicts the current displacement state and error covariance based on historical data and motion patterns of the displacement signal. In the update step, the predicted displacement state is corrected by combining the currently acquired displacement measurement value to obtain the optimal displacement estimate. Through Kalman filtering, fluctuations in the displacement signal caused by measurement noise and external interference can be effectively suppressed, making the smoothed displacement signal more consistent with the actual movement trajectory of the gate and accurately reflecting the positional change trend of the gate during opening and closing.

[0079] S114 performs median filtering on the motor torque signal to obtain the filtered motor torque signal.

[0080] For example, the gate early warning terminal performs median filtering on the motor torque signal. Median filtering is a nonlinear filtering method. Its principle is to divide the motor torque signal into several sliding windows in chronological order, sort the torque data within each sliding window, and select the data in the middle position after sorting as the output value of that window. This process is repeated for all sliding windows to obtain the filtered motor torque signal. Through filtering, isolated noise points caused by instantaneous load fluctuations and electromagnetic interference in the motor torque signal can be effectively removed, avoiding the influence of noise on the torque signal analysis results. The filtered motor torque signal more accurately reflects the torque variation pattern during the operation of the motor-driven gate.

[0081] S115 performs time alignment on the denoised vibration acceleration signal, the smoothed displacement signal, and the filtered motor torque signal to obtain preprocessed operating status data.

[0082] Specifically, since the vibration acceleration signal, displacement signal, and motor torque signal are collected by different sensors, the three signals may deviate in the time dimension due to factors such as sensor response speed and data transmission delay. Therefore, the gate early warning terminal needs to perform time alignment processing on the denoised vibration acceleration signal, smoothed displacement signal, and filtered motor torque signal. During the processing, the gate early warning terminal uses the timestamp of each signal acquisition as a reference and adjusts the three signals to a unified time scale through interpolation, synchronization calibration, and other methods to ensure that at the same point in time, the three signals can reflect different parameters of the gate under the same operating state. The time-aligned signals form preprocessed operating status data, providing time-consistent data assurance for subsequent synchronous feature extraction and fault analysis.

[0083] In an optional embodiment, the feature extraction module 12 is further configured to:

[0084] S121, extract time-domain features from the vibration acceleration signal in the preprocessed operating status data to obtain a time-domain feature vector.

[0085] For example, the gate early warning terminal performs time-domain feature extraction on the pre-processed vibration acceleration signal. Time-domain feature extraction calculates characteristic parameters based on the signal's amplitude variation and statistical characteristics in the time domain. Specifically, the gate early warning terminal calculates time-domain statistical parameters of the vibration acceleration signal, including mean, variance, standard deviation, peak value, peak-to-peak value, kurtosis, and skewness. The mean reflects the average level of the signal, the variance and standard deviation reflect the signal's dispersion, the peak value and peak-to-peak value characterize the maximum fluctuation range of the signal, and the kurtosis and skewness reflect the morphological characteristics of the signal amplitude distribution. By calculating these time-domain characteristic parameters, the gate early warning terminal integrates them to form a time-domain feature vector. This vector reflects the basic characteristics of the vibration acceleration signal in the time dimension, providing fundamental feature information for subsequent comprehensive analysis of the gate's vibration state and determination of whether anomalies exist.

[0086] S122 extracts frequency domain features from the vibration acceleration signal in the preprocessed operating status data to obtain a frequency domain feature vector.

[0087] For example, the gate early warning terminal performs frequency domain feature extraction on the preprocessed vibration acceleration signal. Frequency domain feature extraction first requires converting the vibration acceleration time-domain signal into a frequency-domain signal using Fourier transform to obtain the amplitude distribution of the signal at different frequency components. Subsequently, the gate early warning terminal calculates frequency domain feature parameters such as the centroid frequency, mean square frequency, frequency variance, and peak frequency of the frequency domain signal. The centroid frequency reflects the center frequency position where the signal energy is concentrated; the mean square frequency and frequency variance reflect the degree of dispersion of the frequency distribution; and the peak frequency is the frequency corresponding to the maximum signal amplitude. These frequency domain feature parameters are integrated to form a frequency domain feature vector. This vector reveals the distribution pattern of different frequency components in the vibration acceleration signal, helping to identify abnormal vibrations at specific frequencies caused by gate component failures (such as bearing wear or structural loosening), providing frequency domain-level feature support for fault diagnosis.

[0088] S123, calculate the kinematic parameters of the displacement signal in the preprocessed running status data to obtain the kinematic feature vector.

[0089] For example, the gate early warning terminal calculates kinematic parameters for the pre-processed displacement signal. This calculation, based on the displacement signal and combined with time information, derives the kinematic characteristics of the gate's operation. Specifically, the terminal calculates the gate's operating speed by performing a first-order derivative on the displacement signal, and its operating acceleration by performing a second-order derivative. It also calculates parameters such as the maximum displacement value, minimum displacement value, rate of change of displacement, and displacement fluctuation amplitude. These parameters reflect the gate's motion state from different perspectives. For instance, velocity and acceleration reflect the dynamic trend of the gate's motion, the maximum and minimum displacement values ​​determine the gate's range of motion, and the rate of change of displacement and fluctuation amplitude reflect the smoothness of the gate's motion. These kinematic parameters are integrated to form a kinematic feature vector. This vector comprehensively characterizes the gate's kinematic characteristics during operation, providing a kinematic basis for judging whether there are faults such as gate jamming or abnormal speed.

[0090] S124, perform fluctuation analysis on the motor torque signal in the preprocessed operating status data to obtain the torque fluctuation feature vector.

[0091] For example, the gate early warning terminal performs fluctuation analysis on the pre-processed motor torque signal. This fluctuation analysis aims to uncover the fluctuation patterns and characteristics of the motor torque signal over time. The gate early warning terminal calculates parameters such as the torque change amplitude, torque change frequency, standard deviation of torque fluctuation, and the slopes of the rising and falling edges of the torque signal. These parameters effectively reflect the stability of the torque output during motor-driven gate operation. An abnormally large increase in torque fluctuation amplitude or frequency may indicate increased resistance or component jamming in the gate operation. By integrating these fluctuation parameters to form a torque fluctuation feature vector, characteristic information at the torque fluctuation level is provided for subsequent comprehensive judgment of whether the gate's operating status is normal. Specifically, the torque change amplitude represents the difference between the maximum and minimum torque within the same time period; the torque change frequency is the number of torque fluctuations per unit time; the standard deviation of torque fluctuation reflects the dispersion of torque fluctuation; and the slopes of the rising and falling edges reflect the speed of torque change.

[0092] S125 concatenates the time-domain feature vector, frequency-domain feature vector, kinematic feature vector, and torque fluctuation feature vector to generate a dynamic feature vector.

[0093] For example, after obtaining the time-domain feature vector, frequency-domain feature vector, kinematic feature vector, and torque fluctuation feature vector respectively, the gate early warning terminal needs to perform a concatenation operation on these four types of feature vectors. During the concatenation process, the gate early warning terminal sequentially concatenates the feature parameters of the four types of vectors according to a preset feature order, forming a vector with higher dimension and more comprehensive information, namely, the dynamic feature vector. The time-domain feature vector provides time-domain statistical information of vibration, the frequency-domain feature vector provides frequency-domain distribution information of vibration, the kinematic feature vector provides dynamic characteristic information of gate motion, and the torque fluctuation feature vector provides stability information of motor torque. After concatenation of the four types of vectors, the dynamic feature vector can completely reflect the operating state and motion inertia of the gate from multiple dimensions such as vibration, motion, and driving torque, providing comprehensive and systematic feature data support for subsequent input of fault anomaly models for fault diagnosis, ensuring the accuracy and comprehensiveness of fault diagnosis.

[0094] In an optional embodiment, the fault anomaly analysis module 13 is further configured to:

[0095] S131, input the dynamic feature vector into the preset fault anomaly model, and use the support vector machine classifier in the fault anomaly model to identify the fault type and obtain the fault type.

[0096] For example, the gate early warning terminal inputs dynamic feature vectors into a preset fault anomaly model. The Support Vector Machine (SVM) classifier is a machine learning classification algorithm based on statistical learning theory, which classifies data by finding the optimal classification hyperplane. During the training phase, the SVM classifier learns the feature patterns and classification boundaries corresponding to different fault types through a large number of dynamic feature vector samples labeled with fault types. During classification, the SVM classifier maps the input dynamic feature vectors to a high-dimensional feature space. Based on the position of the vector relative to the classification boundary in the high-dimensional space, it determines the fault category to which it belongs, thus obtaining the fault type. This classification process has high generalization ability and can effectively handle small sample sizes and high-dimensional data, ensuring accurate identification of the gate's fault type even with high-dimensional dynamic feature vectors.

[0097] S132, using the probability output layer in the fault anomaly model, calculate the confidence level of the fault type using the following formula:

[0098]

[0099] in, For confidence level, For the weight vector, For dynamic feature vectors, This is a bias term.

[0100] Specifically, after obtaining the fault type, the gate early warning terminal calculates the confidence level of the fault type through the probability output layer in the fault anomaly model. In the formula for calculating the confidence level mentioned above, Confidence level, used to quantify the reliability of fault type identification results. For the weight vector, For dynamic feature vectors, The bias term is used to adjust the model's output baseline, ensuring that the confidence score is consistent with reality when the feature vector has specific values. By operating the dynamic feature vector and weight vector, and combining this with the bias term, the gate warning terminal can obtain the confidence score corresponding to the fault type. A higher confidence score indicates a more reliable fault type identification result. The weight vector consists of parameters learned during model training, used to measure the influence of each feature parameter in the dynamic feature vector on the fault type identification result. Different feature parameters have different weight values ​​and contribute different amounts to the confidence score.

[0101] S133, based on the timestamp of the dynamic feature vector, determine the time when the fault type occurred, and obtain the fault time.

[0102] For example, during the process of collecting operational status data and generating dynamic feature vectors, the gate early warning terminal adds a corresponding timestamp to each set of dynamic feature vectors. The timestamp accurately records the collection time of the gate operational status data corresponding to that set of feature vectors. After determining the fault type, the gate early warning terminal can determine the collection time of the operational status data corresponding to that fault type based on the timestamp of the dynamic feature vector, and thus infer the time when the fault type occurred, obtaining the fault time. Determining the fault time provides a temporal basis for staff to trace the specific moment the fault occurred and analyze the background and cause of the fault; furthermore, the fault time is also a key parameter when calculating the fault severity index, directly affecting the calculation result of the time decay factor, and thus affecting the accuracy of the fault severity index.

[0103] S134. Based on the fault type, confidence level, and fault time, the fault anomaly analysis results are obtained.

[0104] Specifically, after obtaining the fault type, confidence level, and fault time, the gate early warning terminal integrates these three pieces of information to form a fault anomaly analysis result. The fault type clearly identifies the possible faults currently existing in the gate, serving as the basis for subsequent fault handling solutions and impact assessment. The confidence level quantifies the reliability of the fault type identification result, allowing staff to determine whether further verification of the fault is necessary, avoiding unnecessary maintenance operations due to misjudgment. The fault time determines the specific moment the fault occurred, providing a basis for analyzing the fault's development trend and assessing the duration of its impact on gate operation. These three pieces of information are interconnected and complementary, collectively forming a complete fault anomaly analysis result that reflects the current fault status of the gate.

[0105] In an optional embodiment, the fault index module 14 is further configured to:

[0106] S141, Based on the fault type, query the corresponding weight coefficient from the preset fault database.

[0107] For example, the gate early warning terminal queries a pre-set fault database based on the fault type determined in the fault anomaly analysis results to obtain the corresponding weight coefficient for that fault type. During the establishment of the pre-set fault database, a risk assessment is conducted on various common gate faults. This assessment comprehensively considers factors such as the impact of the fault on the gate's operational safety and stability, the difficulty and cost of maintenance after the fault occurs, and the potential chain reactions that the fault may trigger. Based on the assessment results, a corresponding weight coefficient is assigned to each fault type. The greater the impact of the fault on gate operation, the higher the maintenance difficulty, and the stronger the potential risk, the larger the corresponding weight coefficient; conversely, the smaller the weight coefficient. By accurately querying the weight coefficient corresponding to the fault type from the fault database, key parameters are provided for the subsequent calculation of the basic severity index, ensuring that the basic severity index can reasonably reflect the inherent severity differences of different fault types. Common faults may include gate jamming, abnormal motor torque, and structural loosening.

[0108] S142, calculates the basic severity index based on confidence level and weighting coefficients.

[0109] Specifically, after obtaining the weight coefficients corresponding to the fault types, the gate early warning terminal calculates a basic severity index based on the confidence level in the fault anomaly analysis results and the retrieved weight coefficients. The confidence level reflects the reliability of the fault type identification results, while the weight coefficients reflect the severity of the fault type itself. The calculation of the basic severity index is a comprehensive consideration of these two parameters. For example, during the calculation process, the gate early warning terminal uses a preset mathematical operation method to convert the confidence level and weight coefficients into a quantified basic severity index value. If the confidence level is high and the weight coefficient is large, it indicates that the fault type identification is reliable and the fault itself is severe, and the basic severity index will be at a high level; if the confidence level is low or the weight coefficient is small, the basic severity index will be relatively low. The basic severity index initially quantifies the severity of the fault, providing a basis for further refining the fault severity by incorporating time factors.

[0110] S143, Calculate the time decay factor using the following formula based on the failure time:

[0111]

[0112] in, The time decay factor, The attenuation rate, For the current time, This refers to the time of failure.

[0113] Specifically, the gate early warning terminal calculates a time decay factor based on the fault time in the fault anomaly analysis results and a preset formula. The formula for calculating the time decay factor includes... The time decay factor, The attenuation rate, For the current time, The attenuation rate is a parameter preset based on the gate fault development pattern and actual operation and maintenance experience. Different types of faults correspond to different attenuation rates, and its value reflects the rate at which the severity of the fault changes over time. If the potential risk of the fault increases over time (such as structural crack faults), the attenuation rate can be set to a specific positive value; if the impact of the fault gradually weakens over time (such as temporary faults caused by transient interference), the attenuation rate can be set to a specific negative value or zero. The current time is the system time when the gate early warning terminal calculates the time attenuation factor, and the fault time is the time when the fault occurred. By calculating the difference between the current time and the fault time, and combining it with the attenuation rate, the time attenuation factor is obtained, which is used to reflect the impact of the fault occurrence time on the current severity of the fault.

[0114] S144, combined with the basic severity index and the time decay factor, yields the fault severity index.

[0115] For example, the gate early warning terminal combines the calculated basic severity index and time decay factor, and obtains the final fault severity index through a preset calculation rule. The calculation rule can be multiplying the basic severity index by the time decay factor, or adding the effect of the time decay factor to the basic severity index. If the time decay factor is greater than 1, it indicates that the severity of the fault has increased over time, and the final fault severity index will be higher than the basic severity index; if the time decay factor is less than 1, it indicates that the severity of the fault has decreased over time, and the final fault severity index will be lower than the basic severity index; if the time decay factor is equal to 1, the severity of the fault has not changed over time, and the fault severity index is consistent with the basic severity index. The fault severity index comprehensively considers the inherent severity of the fault type, the reliability of fault identification, and the impact of the fault occurrence time, enabling a more accurate and comprehensive quantification of the threat level of the current fault to gate operation, providing precise quantitative basis for subsequently formulating reasonable fault early warning strategies.

[0116] In an optional embodiment, the fault classification and early warning module 15 is further configured to:

[0117] S151, based on the preset warning threshold rules, classifies the fault severity index to obtain the fault warning level.

[0118] Specifically, the gate early warning terminal classifies the calculated fault severity index based on preset early warning threshold rules to obtain the fault early warning level. The preset early warning threshold rules are formulated by combining gate operation safety standards, historical fault handling experience, and industry specifications. These rules contain multiple different threshold ranges, each corresponding to a specific early warning level. For example, when the fault severity index is in the lowest threshold range, it corresponds to a minor early warning level, indicating that the fault has a relatively small impact on gate operation; when the fault severity index is in the medium threshold range, it corresponds to a general early warning level, indicating that the fault requires attention and timely handling; when the fault severity index exceeds the highest threshold, it corresponds to an emergency early warning level, meaning that the fault has seriously threatened gate safety and immediate emergency measures are required. The gate early warning terminal determines its threshold range and thus the corresponding fault early warning level by comparing the current fault severity index with the thresholds in the early warning threshold rules.

[0119] S152, based on the fault warning level, match the corresponding warning strategy from the preset warning strategy library to obtain the target warning strategy.

[0120] Specifically, after obtaining the fault warning level, the gate warning terminal matches the corresponding warning strategy from the preset warning strategy library to obtain the target warning strategy. The warning strategy library is pre-built and stores standardized warning processing schemes corresponding to different warning levels. The target warning strategy includes the frequency of warning notifications, the notification recipients, and the notification method. Regarding the notification frequency, minor warnings may be set to a daily summary notification, while emergency warnings are set to real-time continuous notifications, ensuring that information is transmitted at a reasonable frequency for faults of different severity. As for the notification recipients, minor warnings may only notify frontline maintenance personnel, general warnings notify the maintenance manager, and emergency warnings require simultaneous notification to the maintenance team, management department, and emergency command personnel to ensure timely awareness by relevant parties. Notification methods can include SMS, system pop-ups, dedicated warning APP push notifications, and telephone voice notifications, achieving reasonable allocation of warning resources.

[0121] S153 generates an early warning signal based on the target early warning strategy and the results of fault anomaly analysis.

[0122] Specifically, the gate early warning terminal generates a complete early warning signal based on the target early warning strategy and fault anomaly analysis results. First, the terminal structures key information from the fault anomaly analysis results, such as fault type, confidence level, and fault time, converting it into easily understandable text descriptions. Following the notification format specified in the target early warning strategy, the terminal integrates the structured fault information with the early warning level identifier to form the main body of the early warning content. Finally, based on the notification frequency, recipients, and notification methods of the target early warning strategy, the terminal adds triggering conditions to the early warning signal, generating an early warning signal that can directly trigger the notification mechanism. These triggering conditions can include notification time intervals, recipient lists, and transmission channel identifiers. The early warning signal not only contains the core information of the fault but also clearly defines how this information is transmitted, ensuring that relevant personnel can receive and accurately understand the fault situation in a timely manner according to the strategy, providing clear guidance for subsequent fault handling.

[0123] In an optional embodiment, the system further includes a model self-updating module 16, for:

[0124] S161, periodically collect fault events and corresponding operational status data to form new samples.

[0125] Specifically, during long-term operation, the gate early warning terminal periodically collects actual fault events and corresponding operational status data to form new samples. The collection cycle can be preset according to the gate's operational intensity and fault frequency, such as once a month or once a quarter, to ensure continuous accumulation of effective data. The collected fault event information must include the actual fault type verified on-site, the specific process of the fault occurrence, the fault handling results, and the final impact assessment to ensure the authenticity and completeness of the fault information. The corresponding operational status data consists of the original operational status data (and pre-processed data) for a period of time before and after the fault occurrence, and must precisely correspond to the timeline of the fault event to ensure that the data accurately reflects the changes in the gate's status before and after the fault. By associating and storing the fault event information and the corresponding operational status data, a structured new sample is formed.

[0126] S162, add the new samples to the training sample set, and incrementally train the fault anomaly model to obtain the updated fault anomaly model.

[0127] Specifically, the gate early warning terminal adds new samples to the existing training sample set and incrementally trains the preset fault and anomaly model to obtain an updated fault and anomaly model. Incremental training differs from traditional full retraining. Its core is to fine-tune the model parameters using only new samples while retaining the original training results, avoiding resource waste and excessively long training cycles caused by full training. During training, the new samples are first validated to ensure that the sample data is complete, without missing or abnormal data, and consistent with the format and feature dimensions of the original training samples. The new samples are then input into the fault and anomaly model, and the gradient descent optimization algorithm is used to iteratively update the classification boundary parameters of the support vector machine classifier, the weight vector of the probability output layer, and the bias terms, enabling the model to learn new fault modes or new features of existing fault modes contained in the new samples. After training, the updated model is tested using a preset validation dataset to verify whether its fault identification accuracy and confidence calculation precision have improved. If the preset performance indicators are met, the updated fault and anomaly model is determined to be effective and can be used for subsequent fault diagnosis, achieving continuous model optimization.

[0128] S163. Based on the actual evolution process of the faults in the newly added samples, dynamically adjust the weight coefficients of the corresponding fault types in the fault database to obtain the updated fault database.

[0129] Specifically, the gate early warning terminal dynamically adjusts the weight coefficients of corresponding fault types in the fault database based on the actual evolution of faults in newly added samples, resulting in an updated fault database. The actual evolution of a fault refers to the change in its severity over time and its actual impact on gate operation throughout the entire lifecycle of a fault in a newly added sample, from occurrence to resolution. For example, if a motor torque fluctuation fault, which originally had a low weight coefficient, repeatedly causes gate opening and closing delays in newly added samples, and its actual impact exceeds expectations, its weight coefficient needs to be adjusted. During the adjustment process, the actual impact of faults in newly added samples is first analyzed, such as downtime, maintenance costs, and safety risk levels, establishing a correspondence between the degree of impact and the weight coefficients. Then, the original weight coefficients for this fault type in the fault database are queried, and the adjustment range is calculated based on the actual impact analysis results. If the actual impact is greater than expected, the weight coefficient is increased; otherwise, it is decreased. Finally, the adjusted weight coefficients are updated in the fault database, and the basis for the adjustment is recorded to ensure that the weight coefficients in the fault database reflect the actual severity of the fault in real time, providing a guarantee for the accuracy of subsequent fault severity index calculations.

[0130] In the aforementioned intelligent diagnosis and early warning system for hydraulic gates based on condition monitoring, the gate early warning terminal performs targeted preprocessing on operational status data such as vibration acceleration, displacement, and motor torque to effectively filter out interference information and ensure data reliability. A dynamic feature vector is formed through multi-dimensional feature extraction and integration, which, combined with a fault anomaly model, accurately identifies the fault type, calculates confidence levels, and determines the fault time. A scientific fault severity index is derived by using a pre-set fault database. Based on the index, tiered early warning signals are generated and matched with corresponding notification strategies to ensure that relevant parties receive fault information in a timely manner. Simultaneously, by periodically collecting incremental samples to optimize the model and dynamically adjusting the fault database weight coefficients, the adaptability and accuracy of the diagnosis and early warning are continuously improved, effectively ensuring gate operation safety and reducing potential fault hazards.

[0131] 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.

[0132] Based on the same inventive concept, this application also provides a method for implementing the aforementioned intelligent diagnosis and early warning system for hydraulic gates based on condition monitoring. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations in one or more embodiments of the intelligent diagnosis and early warning method for hydraulic gates based on condition monitoring provided below can be found in the limitations of the intelligent diagnosis and early warning system for hydraulic gates based on condition monitoring described above, and will not be repeated here.

[0133] In one exemplary embodiment, such as Figure 2 As shown, a method for intelligent diagnosis and early warning of hydraulic gates based on condition monitoring is provided, including:

[0134] S1, acquire the operating status data during the gate operation process, and preprocess the operating status data to obtain preprocessed operating status data; the operating status data includes vibration acceleration signal, displacement signal and motor torque signal;

[0135] S2, extract features from the preprocessed operating status data to generate dynamic feature vectors; the dynamic feature vectors are used to reflect the motion inertia of the gate;

[0136] S3, combined with the preset fault anomaly model, performs fault anomaly analysis on the dynamic feature vector to obtain fault anomaly analysis results; the fault anomaly analysis results include fault type, confidence level and fault time;

[0137] S4. Based on the fault anomaly analysis results and the preset fault database, calculate the fault severity index; the fault severity index is used to characterize the severity of the fault anomaly analysis results.

[0138] S5 generates a fault warning signal based on the fault severity index; the fault warning signal is used to characterize the fault degree of the gate's operating status.

[0139] Furthermore, the operating status data during the gate's operation is acquired, and the operating status data is preprocessed to obtain preprocessed operating status data, including the following steps:

[0140] Collect operational status data during the gate's operation;

[0141] Wavelet threshold denoising is performed on the vibration acceleration signal to obtain the denoised vibration acceleration signal;

[0142] The displacement signal is smoothed by Kalman filtering to obtain a smoothed displacement signal;

[0143] The motor torque signal is subjected to median filtering to obtain the filtered motor torque signal.

[0144] The noise-reduced vibration acceleration signal, the smoothed displacement signal, and the filtered motor torque signal are time-aligned to obtain preprocessed operating status data.

[0145] Furthermore, feature extraction is performed on the preprocessed runtime data to generate dynamic feature vectors, including the following steps:

[0146] Time-domain feature extraction is performed on the vibration acceleration signal in the preprocessed operational status data to obtain a time-domain feature vector;

[0147] Frequency domain feature extraction can be performed on the vibration acceleration signal in the preprocessed operating status data to obtain the frequency domain feature vector;

[0148] The kinematic parameters of the displacement signals in the preprocessed running status data are calculated to obtain the kinematic feature vectors.

[0149] A fluctuation analysis was performed on the motor torque signal in the preprocessed operating status data to obtain the torque fluctuation feature vector;

[0150] The time-domain feature vector, frequency-domain feature vector, kinematic feature vector, and torque fluctuation feature vector are concatenated to generate a dynamic feature vector.

[0151] Furthermore, based on a pre-defined fault anomaly model, fault anomaly analysis is performed on the dynamic feature vectors to obtain the fault anomaly analysis results, including the following steps:

[0152] The dynamic feature vector is input into the preset fault anomaly model, and the fault type is identified by the support vector machine classifier in the fault anomaly model.

[0153] The confidence level of the fault type is calculated using the following formula through the probability output layer in the fault anomaly model:

[0154]

[0155] in, For confidence level, For the weight vector, For dynamic feature vectors, For bias terms;

[0156] Based on the timestamp of the dynamic feature vector, the time when the fault type occurred is determined, and the fault time is obtained;

[0157] Based on the fault type, confidence level, and fault time, the fault anomaly analysis results are obtained.

[0158] Furthermore, based on the fault anomaly analysis results and in conjunction with a pre-defined fault database, the fault severity index is calculated, including the following steps:

[0159] Based on the fault type, the corresponding weight coefficient is retrieved from the preset fault database;

[0160] The basic severity index is calculated based on confidence level and weighting coefficients;

[0161] Based on the failure time, the time decay factor is calculated using the following formula:

[0162]

[0163] in, The time decay factor, The attenuation rate, For the current time, For the time of failure;

[0164] The fault severity index is obtained by combining the basic severity index and the time decay factor.

[0165] Furthermore, based on the fault severity index, a fault warning signal is generated, including the following steps:

[0166] Based on preset warning threshold rules, the fault severity index is classified to obtain the fault warning level;

[0167] Based on the fault warning level, the corresponding warning strategy is matched from the preset warning strategy library to obtain the target warning strategy; the target warning strategy includes the frequency of warning notification, the notification target and the notification method.

[0168] Based on the target early warning strategy and the results of fault anomaly analysis, an early warning signal is generated.

[0169] Furthermore, the method also includes the following steps:

[0170] Regularly collect fault events and corresponding operational status data to create new samples;

[0171] Add new samples to the training sample set to incrementally train the fault anomaly model and obtain the updated fault anomaly model.

[0172] Based on the actual evolution of faults in the newly added samples, the weight coefficients of the corresponding fault types in the fault database are dynamically adjusted to obtain the updated fault database.

[0173] In one embodiment, such as Figure 3 A computer device 300 is provided, comprising:

[0174] At least one processor 301, and at least one memory 302 communicatively connected to said processor 301; said memory stores application code executable by said processor, said application code being executed by said processor to enable said processor to execute the above-described intelligent diagnosis and early warning method for hydraulic gates based on status monitoring;

[0175] The computer device may also include: sensor 303;

[0176] The processor 301, memory 301, and sensor 303 can be connected via bus 304 or other means. The figure shows an example of connection via bus 304. Figure 3 The character is represented by a single thick line, but this does not mean that there is only one bus or a type of bus.

[0177] 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 in the above method embodiments.

[0178] For the device embodiments, since they basically correspond to the method embodiments, the relevant parts can be referred to in the description of the method embodiments. The device embodiments described above are merely illustrative. The components described as separate parts may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this disclosure according to actual needs. Those skilled in the art can understand and implement this without creative effort.

[0179] The above-described embodiments are merely illustrative of several implementation methods of the embodiments of this application, and their descriptions are relatively specific and detailed. However, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the embodiments of this application, and these modifications and improvements all fall within the protection scope of the embodiments of this application.

Claims

1. A smart diagnostic and early warning system for hydraulic gates based on condition monitoring, characterized in that, The system includes: The data acquisition and preprocessing module is used to acquire the operating status data during the gate's operation and preprocess the operating status data to obtain preprocessed operating status data; the operating status data includes vibration acceleration signals, displacement signals, and motor torque signals; The feature extraction module is used to extract features from the preprocessed operating status data and generate a dynamic feature vector; the dynamic feature vector is used to reflect the motion inertia of the gate. The fault anomaly analysis module is used to perform fault anomaly analysis on the dynamic feature vector in combination with a preset fault anomaly model to obtain fault anomaly analysis results; the fault anomaly analysis results include fault type, confidence level and fault time. The fault index module is used to calculate the fault severity index based on the fault anomaly analysis results and a preset fault database; the fault severity index is used to characterize the severity of the fault anomaly analysis results. The fault classification and early warning module is used to generate a fault early warning signal based on the fault severity index; the fault early warning signal is used to characterize the fault degree of the gate's operating status.

2. The system according to claim 1, characterized in that, The data acquisition and preprocessing module is also used for: Collect the operating status data during the gate's operation; The vibration acceleration signal is subjected to wavelet threshold denoising to obtain the denoised vibration acceleration signal. The displacement signal is smoothed by Kalman filtering to obtain a smoothed displacement signal; The motor torque signal is subjected to median filtering to obtain the filtered motor torque signal; The denoised vibration acceleration signal, smoothed displacement signal, and filtered motor torque signal are time-aligned to obtain the preprocessed operating status data.

3. The system according to claim 1, characterized in that, The feature extraction module is also used for: Time-domain feature extraction is performed on the vibration acceleration signal in the preprocessed operating status data to obtain a time-domain feature vector; Frequency domain feature extraction can be performed on the vibration acceleration signal in the preprocessed operating status data to obtain a frequency domain feature vector. The kinematic parameters of the displacement signals in the preprocessed running status data are calculated to obtain the kinematic feature vectors. A fluctuation analysis is performed on the motor torque signal in the preprocessed operating status data to obtain a torque fluctuation feature vector; The time-domain feature vector, the frequency-domain feature vector, the kinematic feature vector, and the torque fluctuation feature vector are concatenated to generate a dynamic feature vector.

4. The system according to claim 1, characterized in that, The fault anomaly analysis module is also used for: The dynamic feature vector is input into a preset fault anomaly model, and the fault type is identified by the support vector machine classifier in the fault anomaly model. The confidence level of the fault type is calculated using the following formula through the probability output layer in the fault anomaly model: in, For the confidence level, For the weight vector, For the dynamic feature vector, For bias terms; Based on the timestamp of the dynamic feature vector, the time when the fault type occurred is determined, and the fault time is obtained; The fault anomaly analysis results are obtained based on the fault type, the confidence level, and the fault time.

5. The system according to claim 4, characterized in that, The fault index module is also used for: Based on the fault type, the corresponding weight coefficient is retrieved from the preset fault database; Based on the confidence level and the weighting coefficients, the basic severity index is calculated; Based on the fault time, the time decay factor is calculated using the following formula: in, The time decay factor is... The attenuation rate, For the current time, The fault time; The fault severity index is obtained by combining the basic severity index and the time decay factor.

6. The system according to claim 1, characterized in that, The fault classification and early warning module is also used for: Based on preset warning threshold rules, the fault severity index is classified to obtain the fault warning level; Based on the fault warning level, a corresponding warning strategy is matched from a preset warning strategy library to obtain the target warning strategy; The target early warning strategy includes the frequency of early warning notifications, the notification recipients, and the notification methods; The warning signal is generated based on the target warning strategy and the fault anomaly analysis results.

7. The system according to claim 1, characterized in that, The system also includes a model self-updating module, used for: Regularly collect fault events and corresponding operational status data to create new samples; The newly added samples are added to the training sample set, and the fault anomaly model is incrementally trained to obtain the updated fault anomaly model. Based on the actual evolution of faults in the newly added samples, the weight coefficients of the corresponding fault types in the fault database are dynamically adjusted to obtain the updated fault database.

8. A method for intelligent diagnosis and early warning of hydraulic gates based on condition monitoring, characterized in that, The method includes: The system acquires operational status data during the gate's operation and preprocesses the data to obtain preprocessed operational status data. The operational status data includes vibration acceleration signals, displacement signals, and motor torque signals. Feature extraction is performed on the preprocessed operating status data to generate a dynamic feature vector; the dynamic feature vector is used to reflect the motion inertia of the gate. Based on a preset fault anomaly model, fault anomaly analysis is performed on the dynamic feature vector to obtain fault anomaly analysis results; the fault anomaly analysis results include fault type, confidence level, and fault time. Based on the fault anomaly analysis results and combined with a preset fault database, a fault severity index is calculated; the fault severity index is used to characterize the severity of the fault anomaly analysis results. A fault warning signal is generated based on the fault severity index; the fault warning signal is used to characterize the degree of fault in the operating state of the gate.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the system comprising any one of claims 1-7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the system comprising any one of claims 1-7.

Images