Intelligent operation optimization and fault processing method for hydraulic power unit

Abstract

The invention provides an intelligent operation optimization and fault processing method for a hydraulic power unit, and the method comprises the steps: collecting real-time pressure data and flow data from the hydraulic power unit through a sensor network, and carrying out the classification processing of the collected pressure data and flow data through a support vector machine algorithm, obtaining a current operation state classification result of the hydraulic power unit; through power redundancy configuration of the started hydraulic power unit, synchronous signal data are collected for the pressure regulating valve group, a neural network algorithm is adopted to conduct prediction processing on the synchronous signal data, and an expected response time sequence of pressure regulation is obtained; according to the obtained filter mechanism optimization result with the minimized oil path interruption, real-time monitoring is conducted on filter element blockage data in the high-dust environment, anomaly detection is conducted on the filter element blockage data through a support vector machine algorithm, and a predicted deviation value of pressure fluctuation is obtained.

Description

Technical Field

[0001] This invention relates to the field of industrial automation and fault prediction technology, and in particular to an intelligent operation optimization and fault handling method for hydraulic power units. Background Technology

[0002] In the energy and power sector, thermal power generation serves as a core pillar for ensuring stable grid operation and the absorption of new energy sources. The reliability and response speed of its equipment are crucial to regional energy security. Especially in energy bases like Xinjiang, thermal power plants frequently need to respond to grid load fluctuations and ensure rapid peak shaving to support the grid connection needs of new energy sources such as wind and solar power. However, the performance of key components in existing thermal power equipment, such as the hydraulic system of the coal mill, directly affects boiler combustion efficiency and unit load regulation capabilities, necessitating urgent technological upgrades. Improving the stability and rapid response capabilities of hydraulic systems in complex environments has become key to ensuring the reliable operation of the power grid.

[0003] Existing hydraulic systems for coal mills generally suffer from complex design and difficult maintenance. Traditional systems rely on multiple sets of valves and complex pipelines, resulting in numerous interfaces and a high susceptibility to leaks in high-temperature, high-dust environments. For example, the excessive number of pipeline interfaces in hydraulic stations accelerates the aging of seals under extreme climates, and frequent leaks increase maintenance costs and downtime risks. Furthermore, the system's response speed is insufficient, making it difficult to meet the requirements of rapid peak shaving by the power grid. Especially during rapid load changes, lag in pressure regulation leads to unstable output, affecting boiler combustion efficiency. These problems are particularly pronounced in the arid and dusty environment of Xinjiang, limiting the operating efficiency of thermal power plants.

[0004] The core technical challenges of hydraulic systems lie in the reliability of the power unit and the precision of pressure regulation. First, existing single-pump configurations lack a backup mechanism; if a pump or motor fails, the system will immediately shut down, preventing the coal mill from operating normally. For example, in actual operation, overheating of the single-pump motor bearings can cause a sudden drop in the output of the pulverizing system, forcing the unit load to decrease from 150 MW to 100 MW, affecting grid stability. Second, the response speed and synchronization of pressure regulation are insufficient. Traditional valve group designs are complex, with excessively long signal response times, making it difficult to achieve rapid and precise pressure control. Especially under high-load conditions with frequent start-stop cycles, asynchronous cylinder movements cause uneven pressure on the grinding rollers, further exacerbating deviations in coal powder fineness.

[0005] Furthermore, the filtration mechanism of the hydraulic system also faces challenges. Traditional filters are prone to causing brief interruptions in the oil circuit during switching, leading to pressure fluctuations and affecting the stable operation of the grinding rollers. For example, in high-dust environments, filter element clogging is frequent, and if the oil flow is discontinuous during switching, it may cause pressure imbalance between the grinding rollers and the grinding disc, thereby affecting the quality of pulverized coal and increasing the instability of boiler combustion. This dual deficiency in power reliability and pressure regulation precision further amplifies the operational risks of the system under extreme conditions.

[0006] Therefore, ensuring the continuous operation and pressure stability of the hydraulic system in frequent peak shaving and harsh environments has become a key issue in improving the deep-shaving capability and operational reliability of thermal power plants. Summary of the Invention

[0007] The technical problem to be solved by the present invention is to address the above-mentioned defects in the prior art. The technical problem to be solved by the present invention is: how to improve the operational reliability of the hydraulic power unit, the pressure regulation response speed and the continuity of the filtration system under complex working conditions such as high dust and high load, so as to effectively reduce the system failure rate and operational risks.

[0008] To address the aforementioned technical problems, this invention provides an intelligent operation optimization and fault handling method for hydraulic power units, mainly comprising the following steps: Real-time pressure and flow data are collected from the hydraulic power unit through a sensor network. The pressure and flow data are then classified using a support vector machine algorithm to obtain the classification result of the current operating status of the hydraulic power unit. Based on the current operating status classification result, the activation conditions of the standby pump group are obtained. If the current operating status classification result shows a single pump failure, the standby pump group is activated to switch to dual pump parallel mode, and it is determined that the power redundancy configuration of the hydraulic power unit has been enabled. By using the enabled power redundancy configuration, synchronization signal data is collected for the pressure regulating valve group, and a neural network algorithm is used to predict the synchronization signal data to obtain the expected response time series of pressure regulation. Based on the expected response time sequence, the continuity of the oil circuit during the filter switching process is determined. If the expected response time sequence exceeds a preset threshold, the opening parameter of the filter switching valve is adjusted to obtain the optimized filtration mechanism result with minimal oil circuit interruption. Based on the optimization results of the filtration mechanism, real-time monitoring of filter clogging data in high-dust environments is performed. The support vector machine algorithm is used to detect anomalies in the filter clogging data and obtain the predicted deviation value of pressure fluctuation. Based on the predicted deviation value, the overall control parameters of the hydraulic power unit are adjusted. If the predicted deviation value is lower than a preset threshold, the current configuration is maintained; otherwise, the power redundancy and the filtering mechanism are iteratively optimized until the operational risk is reduced to an acceptable level.

[0009] Beneficial effects of the present invention The present invention discloses an intelligent operation optimization and fault handling method for hydraulic power units, which has the following beneficial effects: 1. Intelligent fault diagnosis and redundancy switching: The system uses a support vector machine algorithm to accurately and quickly classify and identify the operating status of the hydraulic power unit. When a single pump fails, the backup pump group can be activated immediately, achieving a seamless switch to the dual-pump parallel mode, which greatly improves the power reliability and continuity of the system.

[0010] 2. Precise pressure regulation prediction: The neural network algorithm is used to predict the synchronization signal of the pressure regulating valve group, generating a high-precision expected response time series, which provides a reliable data foundation for subsequent oil circuit continuity judgment and valve control, effectively improving the response speed and synchronization of pressure regulation.

[0011] 3. Minimize oil circuit interruption: By analyzing the expected response time series, the opening of the filter switching valve is dynamically adjusted, the filter switching logic is optimized, the duration of oil circuit interruption is significantly reduced, and the stable operation of the system under high pressure and high dust conditions is ensured.

[0012] 4. Adaptive Optimization and Risk Control: By continuously monitoring the filter element's clogging status and calculating pressure fluctuation prediction deviations, a closed-loop control parameter iterative optimization mechanism was constructed. This mechanism can adaptively adjust power redundancy and filtration mechanisms to ensure that system operational risks are always controlled within an acceptable level, achieving intelligent operation and maintenance.

[0013] 5. Comprehensive improvement of system performance: This invention integrates fault handling, operation optimization and risk prediction, systematically solves several key technical problems faced by hydraulic power units in complex industrial environments, and significantly improves their overall stability, reliability and service life. Detailed Implementation

[0014] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0015] It should be noted that the terms "including" and "containing" used in this article are open-ended expressions, meaning that they include but are not limited to.

[0016] This embodiment provides an intelligent operation optimization and fault handling method for hydraulic power units, which specifically includes the following steps: Step S101: Classification of operating status Real-time pressure and flow data are collected from the hydraulic power unit via a sensor network to obtain real-time monitoring data. Data preprocessing methods (such as filtering and normalization) are used to clean the real-time monitoring data, resulting in standardized data. A support vector machine (SVM) algorithm is then used to train a classification model on the standardized data. If the output value of the classification model is greater than a preset threshold, the hydraulic power unit is determined to be in normal operation; if the output value is less than or equal to the preset threshold, it is determined to be in an abnormal operating state, resulting in a state classification result. Based on the state classification result, time series data of the abnormal operating states are extracted to obtain an abnormal time series. Time series analysis methods (such as STL decomposition and feature extraction) are used to extract features from the abnormal time series, resulting in an abnormal feature set. A decision tree algorithm is then used to classify the abnormal feature set to accurately determine the anomaly type (such as single pump failure, valve jamming, etc.).

[0017] Step S102: Enable power redundancy configuration The system acquires sensor data from the hydraulic power unit and uses a fault detection algorithm to determine its current operating status, resulting in a status classification. If the classification indicates a single pump failure, the activation conditions for the standby pump group are immediately triggered, generating a pump group switching command. Based on the switching command, the hydraulic power unit is switched to a dual-pump parallel mode, and the mode transition is confirmed. Operating parameters (such as pressure, flow rate, and current) in the dual-pump parallel mode are acquired through a real-time monitoring module to determine if the power redundancy configuration has been successfully enabled. If the power redundancy configuration is enabled, stable operating data of the hydraulic power unit is obtained through a status feedback mechanism to confirm that the system has stabilized. Based on the system stability data, a support vector machine algorithm is used again to classify the operating status of the hydraulic power unit, yielding a classification result. Based on this classification result, the operating parameters of the dual-pump parallel mode (such as pump speed distribution) are fine-tuned, ultimately confirming the completion of the power redundancy configuration optimization for the hydraulic power unit.

[0018] Step S103: Pressure Regulation Prediction By utilizing the redundant configuration of the activated hydraulic power unit, synchronous signal data of the pressure regulating valve group is acquired. A sensor array is used to collect multi-channel pressure signals in real time, resulting in a raw signal dataset. Based on this raw dataset, data preprocessing techniques (such as wavelet denoising and Z-score normalization) are employed to denoise and normalize the signals, generating a standardized signal dataset. A convolutional neural network is used to extract deep features from the standardized signal dataset, yielding a temporal feature set. If the fluctuation amplitude of the feature value sequence exceeds a preset threshold (indicating high signal noise or interference), principal component analysis and other dimensionality reduction processing are performed on the feature set to generate an optimized feature dataset. Using the optimized feature dataset, a long short-term memory network is employed to predict the time series, obtaining the expected time series for pressure regulation. The expected time series is compared with the actual acquired response sequence. If the deviation between the predicted and actual values ​​exceeds a preset threshold, a backpropagation algorithm is used to adaptively adjust the model parameters, generating an optimized prediction model. Finally, using this optimized prediction model, precise control commands for the pressure regulating valve group are generated in real time, resulting in the final pressure regulation response sequence.

[0019] Step S104: Optimize the filtering mechanism Obtain the expected response time series of pressure regulation obtained in step S103, and use time series analysis methods (such as calculating variance and range) to determine the fluctuation characteristics of the response time series. If the fluctuation characteristics exceed a preset threshold, the degree of abnormality in oil circuit continuity is quantitatively judged by comparing the difference between the fluctuation amplitude and the preset threshold. Based on the degree of abnormality in oil circuit continuity, a linear regression algorithm is used to establish a mapping relationship between the degree of abnormality and the valve opening adjustment amount, and the optimized value of the filter switching valve opening is calculated. Using the adjusted valve opening parameters, the filter switching process is simulated in a simulation environment or actual system, and a new response time series is obtained, focusing on determining the duration of oil circuit interruption. If the duration of oil circuit interruption does not reach the minimization target (e.g., the target is less than 10 milliseconds), the valve opening parameters are iteratively adjusted, and the simulation is repeated to obtain the response time series until the degree of improvement in oil circuit continuity meets the requirements. Based on the degree of improvement, a K-means clustering algorithm is used to classify the response time series obtained from multiple iterations, and the optimal result that minimizes oil circuit interruption is selected as the optimization result of the filtering mechanism. Based on this optimization result, the control logic of the filter switching process is updated to obtain the final oil circuit continuity parameters, thereby determining the system's operational stability.

[0020] Step S105: Filter Cartridge Clogging Monitoring and Abnormal Detection Pressure fluctuation data at the inlet and outlet of the filter element in a high-dust environment is collected by pressure sensors to obtain a real-time monitoring dataset. If the pressure difference or fluctuation frequency in the real-time monitoring dataset exceeds a preset threshold, a support vector machine (SVM) algorithm is used to classify the pressure fluctuation data to determine whether the filter element is clogged and the severity of the blockage. Based on the classification results of the SVM algorithm, a predicted deviation value for filter element blockage is output. This value is used to quantify the degree of deviation between the current state and the normal state, thereby determining the anomaly detection result. Through the anomaly detection result, the changing trend of pressure fluctuations (such as shorter fluctuation period and increased amplitude) is further analyzed to obtain the dynamic characteristics of filter element blockage. Based on these dynamic characteristics, the parameters of the filtration mechanism are adjusted, such as triggering the backwashing procedure in advance or adjusting the activation strategy of the standby filter, to obtain an optimized filtration configuration. The operating parameters of the oil circuit system are updated with the optimized filtration configuration, and it is observed whether the system returns to a stable oil circuit operating state. If the pressure fluctuation index still exceeds the preset threshold under a stable oil circuit operating state, data is collected again and the above process is repeated until a new and effective optimized configuration is obtained.

[0021] Step S106: System Iterative Optimization and Risk Control Real-time pressure fluctuation data of the hydraulic system is acquired, and its predicted deviation value is determined using time series analysis techniques (such as the ARIMA model). If the predicted deviation value is lower than a preset threshold, it indicates that the system is operating smoothly, and the current control parameters of the hydraulic power unit are maintained to enter a stable operating state. If the predicted deviation value is higher than the preset threshold, a support vector machine algorithm is first used to optimize the power redundancy configuration, such as dynamically adjusting the load distribution or switching strategy of the dual pumps, to determine the adjusted redundancy parameters. Based on the adjusted redundancy parameters, the control signal of the filtering mechanism is then optimized using a Kalman filter algorithm to filter out noise interference and obtain smoother, more realistic pressure fluctuation data. Key operational risk indicators (such as fluctuation variance and number of exceedances) are extracted from the smoothed pressure fluctuation data, and it is determined whether they have reached an acceptable level. If the operational risk has not reached an acceptable level, the overall control parameters of the system (including power and filtering parameters) are iteratively adjusted, and new pressure fluctuation data is acquired. Based on the new pressure fluctuation data, the above deviation judgment and optimization process is repeated to form a closed-loop control until the operational risk is confirmed to have dropped to an acceptable level, at which point the final stable control parameter set is determined.

Claims

1. A method for intelligent operation optimization and fault handling for a hydraulic power unit, characterized in that, The method comprises: Collecting real-time pressure data and flow data from the hydraulic power unit through a sensor network, and performing classification processing on the pressure data and flow data using a support vector machine algorithm to obtain a current operating state classification result of the hydraulic power unit; According to the current operating state classification result, an activation condition of a backup pump group is obtained, if the current operating state classification result shows single pump failure, the backup pump group is activated to switch to a double pump parallel mode, and it is determined that the power redundancy configuration of the hydraulic power unit has been enabled; Through the enabled power redundancy configuration, synchronous signal data is collected for a pressure regulating valve group, and prediction processing is performed on the synchronous signal data using a neural network algorithm to obtain an expected response time sequence of pressure regulation; According to the expected response time sequence, the oil line continuity in the filter switching process is judged, if the expected response time sequence exceeds a preset threshold, the opening parameter of the filter switching valve is adjusted to obtain a filter mechanism optimization result of minimizing oil line interruption; Through the filter mechanism optimization result, real-time monitoring is performed on the filter core blockage data in a high dust environment, and abnormal detection is performed on the filter core blockage data using a support vector machine algorithm to obtain a prediction deviation value of pressure fluctuation; According to the prediction deviation value, the overall control parameter of the hydraulic power unit is adjusted, if the prediction deviation value is lower than a preset threshold, the current configuration is maintained, otherwise the power redundancy and the filter mechanism are iteratively optimized until the operating risk is reduced to an acceptable level.

2. The method of claim 1, wherein, The method comprises: Collecting pressure data and flow data from the hydraulic power unit through the sensor network to obtain real-time monitoring data; Using a data preprocessing method to clean the real-time monitoring data to obtain standardized data; Performing classification training on the standardized data through a support vector machine algorithm to obtain a classification model; If the output value of the classification model is greater than a preset threshold, it is determined that the hydraulic power unit is in a normal operating state; If the output value is less than or equal to the preset threshold, it is determined to be an abnormal operating state, and a state classification result is obtained; According to the state classification result, time series data of the abnormal operating state is extracted to obtain an abnormal time series; Using a time series analysis method to extract features from the abnormal time series to obtain an abnormal feature set; Performing classification on the abnormal feature set through a decision tree algorithm to determine the abnormal type.

3. The method of claim 1, wherein, The method comprises: Collecting sensor data of the hydraulic power unit, determining the current operating state through a fault detection algorithm to obtain a state classification result; If the state classification result shows single pump failure, an activation condition of the standby pump group is triggered, and a pump group switching instruction is generated; According to the pump group switching instruction, the hydraulic power unit is controlled to switch to a double-pump parallel mode, and it is determined whether the running mode conversion is completed; Through a real-time monitoring module, running parameters in the double-pump parallel mode are obtained, and it is judged whether the power redundancy configuration is enabled; If the power redundancy configuration has been enabled, stable running data of the hydraulic power unit is obtained through a state feedback mechanism, and system stability is determined; According to the system stability data, a support vector machine algorithm is used to classify the running state of the hydraulic power unit, and a classification result is obtained; According to the classification result, the running parameters of the double-pump parallel mode are adjusted, and it is determined that the power redundancy configuration optimization of the hydraulic power unit is completed.

4. The method of claim 1, wherein, Through the enabled power redundancy configuration, synchronous signal data of the pressure regulating valve group is collected, a neural network algorithm is used for prediction processing of the synchronous signal data, and an expected response time sequence of pressure regulation is obtained, including: Through the redundancy configuration of the hydraulic power unit, the synchronous signal data of the pressure regulating valve group is obtained, a sensor array is used to collect multi-channel pressure signals in real time, and an original signal data set is obtained; The original signal data set is denoised and standardized to generate a standardized signal data set; Through a convolutional neural network, the standardized signal data set is feature extracted to obtain a time sequence feature set of the signal; If the fluctuation amplitude of the feature value sequence in the time sequence feature set exceeds a preset threshold, the time sequence feature set is dimensionally reduced to generate an optimized feature data set; Through a long short-term memory network, the optimized feature data set is time sequence predicted to obtain an expected time sequence of pressure regulation; If the deviation between the predicted value and the actual value of the expected time sequence exceeds a preset threshold, the model parameters are adaptively adjusted to generate an optimized prediction model; Through the optimized prediction model, control instructions of the pressure regulating valve group are generated in real time to obtain a final pressure regulation response sequence.

5. The method of claim 1, wherein, According to the expected response time sequence, the oil line continuity in the filter switching process is judged, if the expected response time sequence exceeds a preset threshold, the opening parameter of the filter switching valve is adjusted, the optimization result of the filter mechanism with minimized oil line interruption is obtained, including: The expected response time sequence is obtained, and the fluctuation characteristics thereof are determined by using a time sequence analysis method; If the fluctuation characteristics exceed a preset threshold, the abnormality degree of the oil line continuity is judged by comparing the difference between the fluctuation amplitude and the preset threshold; According to the abnormality degree, the opening parameter adjustment amount of the filter switching valve is calculated by using a linear regression algorithm to obtain an optimized value of the valve opening; Through the adjusted valve opening parameter, the filter switching process is simulated to obtain a new response time sequence, and the duration of the oil line interruption is judged; If the duration of the oil line interruption does not reach the minimum target, the valve opening parameter is iteratively adjusted, the response time sequence is reacquired, and the improvement degree of the oil line continuity is determined. According to the improvement degree, the response time sequence is classified by using a K-means clustering algorithm to obtain a filter mechanism optimization result of oil circuit interruption minimization; The control logic of the filter switching process is updated through the optimization result to obtain a final oil circuit continuity parameter and determine system operation stability.

6. The method of claim 1, wherein, The filter mechanism optimization result is used to monitor filter element blockage data in a high-dust environment in real time, and a support vector machine algorithm is used to detect the filter element blockage data for abnormality to obtain a predicted deviation value of pressure fluctuation, including: Pressure fluctuation data of the filter element in a high-dust environment are collected by a sensor to obtain a real-time monitoring data set; If the pressure value in the real-time monitoring data set exceeds a preset threshold value, a support vector machine algorithm is used to classify the pressure fluctuation data to determine the filter element blockage state; According to the classification result, a predicted deviation value of filter element blockage is obtained to determine an abnormality detection result; The change trend of pressure fluctuation is analyzed through the abnormality detection result to obtain dynamic characteristics of filter element blockage; According to the dynamic characteristics, the parameters of the filter mechanism are adjusted to obtain an optimized filter configuration; The operating parameters of the oil circuit system are updated through the optimized filter configuration to obtain a stable oil circuit operating state; If the pressure fluctuation still exceeds the preset threshold value in the stable oil circuit operating state, data are re-collected and the above process is repeated to obtain a new optimized configuration.

7. The method of claim 1, wherein, According to the predicted deviation value, the overall control parameters of the hydraulic power unit are adjusted, if the predicted deviation value is lower than a preset threshold value, the current configuration is maintained, otherwise the power redundancy and the filter mechanism are iteratively optimized until the operating risk is reduced to an acceptable level, including: Pressure fluctuation data are obtained, and a predicted deviation value is determined through time series analysis; If the predicted deviation value is lower than a preset threshold value, the current control parameters of the hydraulic power unit are maintained to obtain a stable operating state; If the predicted deviation value is higher than a preset threshold value, a support vector machine algorithm is used to optimize the power redundancy configuration to determine adjusted redundancy parameters; According to the adjusted redundancy parameters, the filter mechanism is optimized by a Kalman filtering algorithm to obtain smoothed pressure fluctuation data; From the smoothed pressure fluctuation data, an operating risk indicator is extracted to determine whether an acceptable level is reached; If the operating risk has not reached an acceptable level, the control parameters are iteratively adjusted to obtain new pressure fluctuation data; According to the new pressure fluctuation data, the deviation judgment and optimization process are repeated to determine the final stable control parameters.