A hydraulic cylinder intelligent oil supply control system and method applied to water conservancy projects

Abstract

The application relates to the technical field of hydraulic engineering, and discloses a hydraulic cylinder intelligent oil supply control system and method applied to hydraulic engineering. The system comprises a hydraulic parameter acquisition module, a data fusion processing module, a manifold state modeling module, a dynamic risk prediction module and a self-adaptive oil supply control module. The hydraulic parameter acquisition module acquires multi-point oil pressure, piston displacement and oil supply flow monitoring signals in real time; the data fusion processing module fuses multi-source heterogeneous signals; the manifold state modeling module constructs a hydraulic cylinder operation state Riemann manifold geometric model based on the fused signals; the dynamic risk prediction module calculates a real-time abnormal aggregation index according to a geometric characteristic vector; and the self-adaptive oil supply control module generates an oil supply control instruction according to the abnormal aggregation index. The system can comprehensively perceive the operation state of the hydraulic cylinder, accurately model and predict risks, dynamically adjust oil supply parameters, and meet the operation requirements of modern hydraulic engineering.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of hydraulic engineering, and particularly relates to an intelligent oil supply control system and method of a hydraulic cylinder applied to hydraulic engineering. BACKGROUND

[0002] In the field of hydraulic engineering, as a core execution element of key operations such as gate opening and closing, dam regulation and channel water control, the stable operation of the oil supply system of the hydraulic cylinder is directly related to the operation accuracy and safety performance of the entire hydraulic facility. The traditional oil supply control of the hydraulic cylinder adopts a fixed mode based on preset parameters, that is, the oil supply pressure and flow threshold are manually set, and a simple on-off valve or proportional valve is used for rough regulation, which is difficult to adapt to the complex and variable working condition requirements of hydraulic engineering.

[0003] The working environment of hydraulic engineering is often accompanied by uncertain factors such as water flow impact, water pressure fluctuation and temperature change. These factors can cause the oil pressure in the oil supply pipeline of the hydraulic cylinder to present nonlinear fluctuation, the piston displacement accuracy to be disturbed, and the oil supply flow to deviate from the actual demand. For example, during the gate opening and closing process, when the water flow speed suddenly increases, the hydraulic cylinder needs to bear the instantaneously increased load. At this time, if the traditional oil supply control system still supplies oil according to the fixed parameters, oil pressure shortage may occur, causing the gate opening and closing to be stalled, or oil pressure may be too high, causing pipeline leakage and damage to the sealing element.

[0004] The monitoring of the traditional control system on the running state of the hydraulic cylinder is mostly limited to single parameter acquisition, such as only monitoring the oil pressure or only monitoring the displacement, and lacks comprehensive analysis capability of multi-source monitoring data. When an abnormality occurs in a parameter, it is difficult to quickly determine whether the abnormality is caused by a fault in the oil supply system itself or by external working condition changes, resulting in delayed fault diagnosis and untimely maintenance response, further increasing the safety risk and maintenance cost of the hydraulic engineering operation.

[0005] The traditional oil supply control mode cannot predict potential risks in the running process of the hydraulic cylinder, and can only passively handle the faults after they occur. For example, when the oil supply pipeline is slightly worn, causing the oil pressure to slowly decrease, the traditional system cannot identify this gradual abnormality, and will not trigger an alarm until the wear intensifies and causes obvious faults. At this time, the normal operation of the hydraulic engineering may have been affected, and even a safety accident may be caused. With the development of large-scale and intelligent hydraulic engineering, the deficiencies of the traditional hydraulic cylinder oil supply control system in control accuracy, self-adaptive ability and risk prediction ability are increasingly prominent, and it is difficult to meet the needs of modern hydraulic engineering for efficient, safe and stable operation. SUMMARY

[0006] The purpose of the present application is to provide an intelligent oil supply control system of a hydraulic cylinder applied to hydraulic engineering to solve the problems raised in the background.

[0007] To achieve the above object, the application provides a hydraulic cylinder intelligent oil supply control system applied to water conservancy projects.

[0008] A hydraulic parameter acquisition module is configured to acquire in real time a multi-point oil pressure monitoring signal of a hydraulic cylinder oil supply pipeline, a piston displacement monitoring signal fed back by a displacement sensor, and an oil supply flow monitoring signal collected by a flowmeter.

[0009] A data fusion processing module is configured to perform multi-source heterogeneous data fusion processing on the oil pressure monitoring signal, the displacement monitoring signal, and the flow monitoring signal output by the hydraulic parameter acquisition module.

[0010] A manifold state modeling module is configured to construct a Riemann manifold geometric model of the hydraulic cylinder operating state according to the fused monitoring signal sequence output by the data fusion processing module.

[0011] A dynamic risk prediction module is configured to calculate a real-time abnormality aggregation index of the hydraulic cylinder oil supply system based on a geometric feature vector output by the Riemann manifold geometric model.

[0012] An adaptive oil supply control module is configured to generate an oil supply control instruction of the hydraulic cylinder according to the real-time abnormality aggregation index output by the dynamic risk prediction module.

[0013] Preferably, the hydraulic parameter acquisition module comprises:

[0014] A first oil pressure sensor distributed at the inlet section of the oil supply pipeline, a second oil pressure sensor distributed at the middle section of the pipeline, and a third oil pressure sensor distributed at the outlet section of the pipeline;

[0015] A displacement sensor installed on the hydraulic cylinder piston rod;

[0016] An ultrasonic flowmeter arranged at the outlet of the oil supply pump;

[0017] The first oil pressure sensor, the second oil pressure sensor, and the third oil pressure sensor output oil pressure monitoring signals respectively, the displacement sensor outputs a displacement monitoring signal, and the ultrasonic flowmeter outputs a flow monitoring signal.

[0018] Preferably, when the data fusion processing module performs multi-source heterogeneous data fusion processing, it:

[0019] Calculates the propagation time lag between the monitoring signals of the first oil pressure sensor and the second oil pressure sensor by using a wavelet coherence analysis method;

[0020] Performs time domain alignment processing on the displacement monitoring signal and the flow monitoring signal according to the propagation time lag;

[0021] Combines the time domain aligned oil pressure monitoring signal, displacement monitoring signal, and flow monitoring signal into a multi-dimensional monitoring signal matrix.

[0022] Preferably, the manifold state modeling module constructs the Riemannian manifold geometric model when:

[0023] Performing Hilbert-Huang transform on the multi-dimensional monitoring signal matrix to extract intrinsic mode function components of each monitoring signal;

[0024] Mapping the instantaneous frequency and instantaneous amplitude of the intrinsic mode function components to tangent space coordinates of the Riemannian manifold;

[0025] Constructing a geometric feature vector of the hydraulic cylinder operating state according to the curvature variation of the tangent space coordinates.

[0026] Preferably, the dynamic risk prediction module calculates the real-time anomaly aggregation degree index when:

[0027] Using a topological data analysis algorithm to identify the high-dimensional clustering structure of the geometric feature vector on the Riemannian manifold;

[0028] Calculating the aggregation density distribution of the geometric feature vector in the manifold space when historical failures occur;

[0029] Quantifying the real-time anomaly aggregation degree index based on the Hausdorff distance between the current geometric feature vector and the historical aggregation density distribution.

[0030] Preferably, the dynamic risk prediction module further comprises:

[0031] Using a causal discovery algorithm to identify the causal correlation strength between oil pressure fluctuations and displacement lag from the geometric feature vector;

[0032] According to the causal correlation strength, the weight coefficient of the real-time anomaly aggregation degree index is corrected.

[0033] Preferably, the adaptive oil supply control module generates an oil supply control instruction when:

[0034] Using a three-valued logic system to map the real-time anomaly aggregation degree index to an oil supply pressure adjustment amount, a flow compensation amount, and a displacement correction amount;

[0035] Adjusting the opening command of the proportional overflow valve according to the oil supply pressure adjustment amount;

[0036] Generating a speed control signal for the variable frequency pump according to the flow compensation amount;

[0037] Outputting a piston position calibration instruction according to the displacement correction amount.

[0038] Preferably, the adaptive oil supply control module further comprises:

[0039] Predicting the oil supply parameter drift trend of the next three operation periods through the time series causal inference model;

[0040] When the oil supply parameter drift trend is detected to exceed the preset threshold, triggering the parameter self-calibration mechanism of the ternary logic system.

[0041] Preferably, the system further comprises:

[0042] A control instruction execution module for parsing the oil supply control instructions generated by the adaptive oil supply control module;

[0043] The control instruction execution module drives the proportional overflow valve to perform the opening degree adjustment action;

[0044] The control instruction execution module controls the variable frequency pump to complete the speed adjustment operation;

[0045] The control instruction execution module sends the piston position calibration instruction to the hydraulic cylinder servo controller.

[0046] Preferably, the present application further comprises a hydraulic cylinder intelligent oil supply control method applied to water conservancy projects, which comprises all the modules and method processes of the hydraulic cylinder intelligent oil supply control system applied to water conservancy projects.

[0047] Compared with the prior art, the present application has the following beneficial effects:

[0048] The hydraulic cylinder intelligent oil supply control system applied to water conservancy projects realizes real-time monitoring of the multi-point oil pressure, piston displacement and oil supply flow of the hydraulic cylinder oil supply pipeline through the hydraulic parameter acquisition module. Compared with the single parameter monitoring mode of the traditional control system, it can comprehensively capture the key state information in the running process of the hydraulic cylinder, cover all links from pipeline pressure to actuator displacement to medium flow, make the system's perception of the hydraulic cylinder's running state more complete and accurate, and timely discover the subtle abnormalities that are easily ignored in the traditional monitoring mode, providing comprehensive data support for subsequent control and risk judgment.

[0049] The data fusion processing module fuses and processes the multi-source heterogeneous monitoring signals, effectively solving the problem of independent existence of each monitoring parameter and poor data correlation in the traditional system. Since the oil pressure, displacement and flow signals may produce noise due to environmental interference during the acquisition process, and the dimensions and change laws of different parameters are different, it is difficult to accurately reflect the real running state of the hydraulic cylinder by directly using a single parameter. Through multi-source data fusion, the data noise interference can be eliminated, the correlation information between different parameters can be integrated, the system operation rules hidden behind multiple parameters can be mined, the fused signals are closer to the actual working condition, a high-quality data foundation is provided for subsequent state modeling, the judgment of the hydraulic cylinder running state is more scientific, and control deviation caused by single parameter misjudgment is avoided.

[0050] The manifold state modeling module constructs a Riemann manifold geometry model based on the fused monitoring signal sequence, breaking through the limitations of traditional linear modeling methods in dealing with the nonlinear and dynamically changing operating state of the hydraulic cylinder. Under complex working conditions of water conservancy projects, the operating state of the hydraulic cylinder presents obvious nonlinear characteristics, and the traditional linear model cannot accurately describe the complex state change law. However, the Riemann manifold geometry model can map the change of the operating state of the hydraulic cylinder into a geometric space by using the characteristics of the geometric space, intuitively reflect the state evolution trend through the geometric feature vector, and better conform to the operating characteristics of the hydraulic cylinder in actual operation, making the modeling of the operating state more accurate and more practical.

[0051] The dynamic risk prediction module calculates a real-time abnormality aggregation degree index based on the geometric feature vector, realizing the early identification of potential risks of the hydraulic cylinder oil supply system. The traditional system can only passively alarm after a fault occurs, but this module can capture the aggregation trend of abnormal states by analyzing the geometric features of the operating state, and can perceive potential risks through the changes in the abnormality aggregation degree index before the fault fully appears, changing the traditional "after-treatment" mode and providing more response time for the staff, which helps to take intervention measures in advance, reduces the probability of failure, and reduces the impact of failure on water conservancy engineering operations.

[0052] The adaptive oil supply control module generates control instructions according to the real-time abnormality aggregation degree index, so that the oil supply control can be dynamically adjusted according to the change of the operating state of the hydraulic cylinder. In the scene where the working conditions of water conservancy projects fluctuate greatly, the traditional fixed parameter control method cannot timely adapt to the change of working conditions, and is prone to mismatch between oil supply and demand, while the adaptive control can judge whether there is a risk in the current operating state and the degree of risk according to the abnormality aggregation degree index, and then automatically adjust the oil supply pressure, flow and other parameters. For example, when the abnormality aggregation degree index rises, the oil supply parameters are adjusted in time to alleviate the abnormal trend, and when the index is normal, the optimal oil supply state is maintained to ensure that the hydraulic cylinder always operates in the best working condition. This improves the flexibility and adaptability of oil supply control, reduces the control deviation caused by changes in working conditions, reduces the probability of faults such as pipeline leakage and seal damage, and also reduces unnecessary energy consumption, improves the economy and stability of the hydraulic cylinder operation, and ensures that water conservancy related operations such as gate opening and closing, dam regulation, etc. can be carried out efficiently and smoothly, reducing maintenance costs and safety risks. BRIEF DESCRIPTION OF DRAWINGS

[0053] Figure 1 The timing diagram of the hydraulic cylinder intelligent oil supply control system applied to water conservancy projects described in the application;

[0054] Figure 2 The working principle flowchart of the hydraulic parameter acquisition module;

[0055] Figure 3 A work principle flow chart for the manifold state modeling module to construct a Riemann manifold geometry model;

[0056] Figure 4 A work principle flow chart for the dynamic risk prediction module to correct the real-time abnormality aggregation index weight coefficient. DETAILED DESCRIPTION

[0057] The technical solutions in the embodiments of the present application will be clearly and completely described with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work are within the protection scope of the present application.

[0058] Please refer to Figure 1 The present application provides a hydraulic cylinder intelligent oil supply control system and method applied to water conservancy projects, which comprises:

[0059] Key parameters in the hydraulic cylinder oil supply process are collected in real time by a multi-source sensor, and accurate modeling and dynamic risk prediction of the operating state are realized based on manifold geometry theory and multi-source data fusion technology, and finally adaptive oil supply control instructions are generated. The system comprises a hydraulic parameter acquisition module, a data fusion processing module, a manifold state modeling module, a dynamic risk prediction module and an adaptive oil supply control module. The hydraulic parameter acquisition module acquires in real time multi-point oil pressure monitoring signals of the hydraulic cylinder oil supply pipeline, piston displacement monitoring signals fed back by a displacement sensor and oil supply flow monitoring signals collected by a flowmeter. The data fusion processing module fuses and processes the above multi-source heterogeneous monitoring signals, eliminates time domain inconsistency and forms a unified multi-dimensional data matrix. The manifold state modeling module constructs a Riemann manifold geometry model describing the operating state of the hydraulic cylinder based on the fused signal sequence, and maps the nonlinear characteristics of the monitoring signals to the geometric structure of the manifold space. The dynamic risk prediction module calculates the real-time abnormality aggregation index of the system according to the geometric feature vector output by the manifold model, and quantifies the operating risk level. The adaptive oil supply control module generates oil supply control instructions according to the abnormality aggregation index, and realizes closed-loop regulation of the oil supply pressure, flow and piston position.

[0060] Embodiment 1: Please refer to Figure 2The hydraulic parameter acquisition module includes a set of oil pressure sensors arranged according to a specific spatial distribution, a high-precision displacement sensor, and an ultrasonic flow meter. The first oil pressure sensor is installed at the inlet section of the pipeline after the oil supply pump, as close as possible to the outlet flange of the pump, to capture the initial value of the oil supply pressure. The second oil pressure sensor is installed at the middle section of the pipeline, usually at the intermediate point after the pipeline direction changes significantly or the length exceeds a certain distance, to monitor the transmission state of the pressure in the pipeline. The third oil pressure sensor is directly installed near the oil inlet of the hydraulic cylinder, serving as the pressure monitoring point at the outlet section of the pipeline. The three sensors work in a synchronous sampling mode, and the output voltage analog signals are transmitted to the signal conditioning unit through shielded cables, and are converted to digital signals after amplification and filtering. The displacement sensor uses magnetostrictive or grating principle, with the measuring body fixed to the cylinder barrel of the hydraulic cylinder, and the movable part mechanically connected to the piston rod, to detect the absolute or relative displacement of the piston in real time. The ultrasonic flow meter uses time difference measurement principle, with the sensor probe clamped and installed on the outer wall of the straight pipe section at the outlet of the oil supply pump, to calculate the flow rate and flow by measuring the time difference of ultrasonic wave propagation in the fluid.

[0061] The data fusion processing module receives multiple digital signals from the above-mentioned sensors. Due to different response characteristics, sampling frequencies, and transmission delays of different sensors, there is natural asynchrony in the time domain. For example, there is a difference between the propagation speed of pressure signals in oil and the electrical transmission speed of displacement signals, and the measurement of flow signals also has a certain pipe flow state response time. To eliminate this time domain inconsistency, the module uses wavelet coherence analysis method to process the monitoring signals of the first and second oil pressure sensors. This method determines the propagation time lag of pressure waves from the pump outlet to the middle section of the pipeline by calculating the phase relationship of two pressure signals at different frequency components. This time lag not only reflects the physical properties of oil, but also includes the comprehensive effects of pipeline elasticity and fluid inertia.

[0062] After obtaining the propagation time lag, the data fusion processing module performs time domain alignment processing on the displacement monitoring signal and the flow monitoring signal. The displacement signal usually needs to be shifted forward or backward by a certain number of sampling points, and this shift amount is calculated according to the product of the pressure propagation time lag and the sampling frequency. The flow signal is adjusted similarly according to its coupling relationship with the pressure signal. Time domain alignment processing not only considers static time lag, but also pays attention to dynamic changes, such as when the oil temperature changes, causing the sound speed of oil to change, the propagation time lag will change slightly, and at this time the time lag estimation needs to be updated online.

[0063] After time alignment, each monitoring signal is combined into a multi-dimensional monitoring signal matrix, each row of which represents a synchronized time sampling point, and each column represents the measurement of a sensor. The first column of the matrix is the first oil pressure sensor signal after time delay correction, the second column is the second oil pressure sensor signal, the third column is the third oil pressure sensor signal, the fourth column is the displacement sensor signal, and the fifth column is the ultrasonic flowmeter signal. The formation of this matrix enables the originally independent multi-source heterogeneous data to be highly consistent in the time dimension, providing an accurate and consistent data basis for subsequent manifold state modeling. The matrix data is stored in the memory in the form of a ring buffer, which not only retains a long enough history data for trend analysis, but also ensures that new data can be updated in real time.

[0064] During the entire implementation process, the reliability of signal transmission is ensured through industrial Ethernet or fieldbus technology. Each sensor channel is provided with a self-diagnosis function, which can detect signal abnormalities such as wire breakage, excessive range or noise interference. When a sensor fails, the system can automatically switch to a redundant backup sensor or use an estimation algorithm to compensate for data loss, maintaining the continuous operation of the system. All processing is implemented on an embedded industrial computing platform, which has sufficient computing power to process multi-channel high-frequency sampling data and meet the strict real-time requirements of hydraulic engineering hydraulic systems.

[0065] Example 2: see Figure 3 The manifold state modeling module receives a multi-dimensional monitoring signal matrix from the data fusion processing module, which contains oil pressure, displacement and flow data after time alignment processing. The module first performs Hilbert-Huang transform processing on each monitoring signal channel in the matrix. This transform process includes two main stages: empirical mode decomposition and Hilbert spectrum analysis. For each monitoring signal, empirical mode decomposition decomposes it into a series of intrinsic mode function components, which are generated through an iterative screening process, each component satisfying certain conditions about extreme points and zero points, and having a clear instantaneous frequency physical meaning. Taking the inlet section oil pressure signal as an example, 5-8 intrinsic mode function components may be generated, corresponding to different frequency range fluctuation characteristics, from high-frequency pressure pulsation to low-frequency trend change.

[0066] After completing the empirical mode decomposition, Hilbert transform is performed on each intrinsic mode function component to calculate its analytic signal, and then the instantaneous frequency and instantaneous amplitude sequences are obtained. The instantaneous frequency reflects the frequency characteristics of the signal component at each time point, while the instantaneous amplitude represents the energy intensity of the frequency component. Taking an intrinsic mode function component of the middle section of the pipeline oil pressure signal as an example, its instantaneous frequency sequence may show a concentrated distribution of frequency values in a certain time interval, while the instantaneous amplitude sequence may show a pattern of gradually decaying or enhancing amplitude.

[0067] When mapping these instantaneous frequency and amplitude parameters onto a Riemannian manifold structure, a differential geometry approach is adopted. The instantaneous frequency and amplitude values of all the IMF components at each time point form a high-dimensional data point, and these data points collectively form a point set on the manifold. The local geometry of the manifold is established by computing the geodesic distances and tangent space projections between these points. The tangent space coordinates are determined by constructing a local coordinate system at each point, which projects the neighboring points onto the tangent plane at that point.

[0068] After the manifold structure is established, the curvature variation of the tangent space coordinates is computed. The curvature computation involves quantifying the degree of bending of the manifold surface, which is achieved by comparing the difference between the geodesic distances and Euclidean distances of neighboring points. High curvature regions correspond to rapid changes or transition phases of the system's operating state, while low curvature regions represent relatively stable operating states. Taking the reversing phase of the hydraulic cylinder piston movement as an example, the manifold curvature during this period usually exhibits significant changes, reflecting the transformation of the system's dynamic characteristics.

[0069] Based on the curvature analysis results, a geometric feature vector is constructed to describe the operating state of the hydraulic cylinder. This vector contains multiple dimensions of feature indicators, including the average curvature, Gaussian curvature, curvature variation rate, and other geometric features, as well as statistical features extracted from the instantaneous frequency and amplitude. These features collectively provide a multi-angle description of the system's operating state, enabling the capture of subtle changes in the system's dynamic behavior. For example, when there is a slight fluctuation in the supply oil pressure, certain components of the geometric feature vector will exhibit corresponding change patterns, which may be difficult to detect in traditional time or frequency domain analysis.

[0070] The entire manifold state modeling process is implemented on an embedded computing platform, using optimized numerical algorithms to handle large-scale data computations. A sliding window mechanism is used during the computation process to handle the latest block of monitoring data, ensuring the timeliness of the features and maintaining the computational load within a reasonable range. The generated geometric feature vector is output in the form of a time series, providing input data for the subsequent dynamic risk prediction module. This module has adaptive learning capabilities, allowing it to automatically adjust the manifold parameters based on the system's operating data, ensuring that the model always matches the actual operating state of the system.

[0071] Embodiment 3: The sequence of geometric feature vectors output by the manifold state modeling module is received and processed by the dynamic risk prediction module. The module first analyzes the distribution structure of these high-dimensional vectors on the Riemannian manifold using a topological data analysis algorithm. Topological data analysis identifies topological features in the data by constructing persistent homologies. In implementation, the Vietoris-Rips complex construction method is used to generate a series of simple complexes by setting different distance parameters ε. As the value of ε increases, connections between data points gradually form, producing topological features of different dimensions (such as connected components, voids, etc.). The birth time and death time of each topological feature are recorded in the persistence diagram. By analyzing the points far from the diagonal line in the persistence diagram, significant clustering structures of geometric feature vectors in the manifold space can be identified. These clustering structures often correspond to the operating modes of the system, including normal operation and various abnormal states.

[0072] Based on the identification of the clustering structure in the emergence phase, the module calculates the clustering density distribution of the geometric feature vectors in the manifold space when historical faults occur. This process uses kernel density estimation method to construct a probability density function with the feature vector point corresponding to each historical fault event as the center. Gaussian kernel function is usually used for such estimation, and its smoothing parameter is adaptively determined according to the distribution characteristics of the data. Historical fault data come from system operation logs and manually annotated records, containing various fault types such as oil line blockage, internal leakage, sensor failure, etc. Each fault type will form a density distribution pattern in the manifold space, and these patterns are stored in the knowledge base in the form of probability distribution.

[0073] Based on the relationship between the current geometric feature vector and the historical clustering density distribution, the module calculates the real-time abnormal clustering degree index. This index is quantified by Hausdorff distance, and its calculation formula is: Where: represents the point set formed by the current geometric feature vector in the manifold space, represents the clustering point set of the kth historical fault mode, represents the geodesic distance between two points in the manifold space. This distance measures the maximum and minimum separation between the current state point set and the historical fault cluster, and the smaller the value, the closer the current state to the historical fault mode.

[0074] The module applies a causal discovery algorithm to identify the causal association strength between oil pressure fluctuations and displacement lags from the geometric feature vectors. The transfer entropy method is used to calculate the causal influence measure between variables, which can capture nonlinear and asymmetric causal relationships. For the oil pressure signal and the displacement signal The transfer entropy calculation is based on the conditional probability distribution of the two signals, which measures the degree of prediction contribution of the oil pressure history information to the current displacement value under the condition of knowing the displacement history information. The strength value of this causal relationship is between 0 and 1, and the greater the value, the stronger the causal influence of the oil pressure fluctuation on the displacement lag.

[0075] According to the calculated causal correlation strength, the module corrects the weight coefficient of the real-time anomaly aggregation index. The correction process adopts a weighted function, and the causal strength is introduced as a weight factor into the anomaly index calculation. For a fault mode with strong causal correlation, the corresponding anomaly aggregation index will be given a higher weight, and vice versa. This correction mechanism enables the system to more accurately reflect the correlation between the current operating state and a specific fault mode, improving the accuracy of risk prediction.

[0076] The entire dynamic risk prediction process adopts a sliding time window mechanism to perform batch processing analysis on the geometric feature vectors in the recent period. The window length is adjusted according to the dynamic characteristics of the system, usually containing hundreds to thousands of sampling points. The analysis results include the real-time anomaly aggregation index value and the matching probability of each fault mode, which are transmitted to the adaptive oil supply control module as the basis for control decisions. The module maintains a dynamically updated knowledge base, continuously absorbing new operating data to optimize the historical fault density distribution model, enabling the system to have continuous learning ability. All calculations are implemented on an industrial-grade embedded platform, using numerical optimization algorithms to ensure computational efficiency and meet real-time requirements.

[0077] Embodiment 4: refer to Figure 4 The adaptive oil supply control module receives the real-time anomaly aggregation index value output by the dynamic risk prediction module, which ranges from 0 to 1 as a continuous value. The module uses a three-value logic system to discretize the index, dividing it into low, medium, and high risk level state intervals. The interval boundary values are determined based on historical operating data statistics, for example, setting [0, 0.3) as the low risk state, [0.3, 0.7) as the medium risk state, and [0.7, 1] as the high risk state. Each state interval corresponds to a different control rule library, and the anomaly aggregation index is mapped to specific oil supply pressure adjustment, flow compensation, and displacement correction amounts through table lookup. The following table shows a typical mapping relationship:

[0078] Anomaly aggregation index and control amount mapping table

[0079]

[0080] The specific value of the pressure adjustment amount is generated by a fuzzy inference mechanism. The system establishes a fuzzy rule base containing 49 rules, taking the feedforward oil pressure deviation and the abnormality aggregation index as input variables. For example, when the inlet pressure deviation is negative and the abnormality index is in the medium-risk interval, the "large amplitude pressure increase" rule is triggered, and the specific pressure adjustment value is output. This value is converted into the opening command of the proportional overflow valve, and the opening change amount ΔK is calculated by the PID controller, whose proportional coefficient is dynamically adjusted according to the abnormality aggregation index. The command is transmitted to the electromagnetic driver of the proportional overflow valve through a 4-20 mA analog signal, driving the armature displacement to change the valve core throttling area.

[0081] The generation of the flow compensation amount adopts a feedforward-feedback composite control strategy. The feedforward part calculates the theoretical flow demand according to the speed signal of the displacement sensor, and the feedback part compensates for the flow deviation according to the abnormality aggregation index. The final generated flow compensation value is converted into the speed control signal of the variable frequency pump, and the pulse width modulation technology is used to generate a three-phase alternating current driving waveform. For example, when the flow needs to be increased by 2.5 L / min, the control system calculates the corresponding motor speed increment Δn, adjusts the output voltage frequency by changing the IGBT conduction duty cycle, and makes the pump speed increase accurately by 85 rpm.

[0082] The calculation of the displacement correction amount combines the deviation between the actual position of the piston and the target position, as well as the influence factor of the abnormality aggregation index. The correction amount is converted into a piston position calibration command, which is transmitted to the hydraulic cylinder servo controller through the CAN bus in the standard CANopen protocol message. The message contains a 32-bit floating-point target position value and timestamp information. After parsing the command, the servo controller adopts a position-velocity-current three-closed-loop control strategy, adjusts the valve core displacement of the electro-hydraulic servo valve, changes the oil flow and direction into the hydraulic cylinder, and makes the piston move to the calibration position. For example, when a +0.3 mm displacement needs to be compensated, the servo valve opening area increases by 0.8 mm² and maintains for 200 ms.

[0083] The module predicts the drift trend of the oil supply parameters through a time series causal inference model. The model uses a long short-term memory network architecture, inputs the oil pressure, flow, displacement, and abnormality index sequence in the last 120 seconds, and outputs the parameter change prediction value in the next three control periods (about 15 seconds). The network contains 128 memory units, which capture the long-term dependence of time series through a gating mechanism. When the drift of any parameter is predicted to exceed the threshold (such as pressure drift > 0.2 MPa), the parameter self-calibration mechanism of the three-value logic system is triggered. This mechanism automatically adjusts the state interval boundary value and the control amount mapping relationship, for example, adjusting the lower limit of the medium-risk interval from 0.3 to 0.35, or increasing the flow compensation range of the high-risk state to ±6 L / min. The self-calibration parameters are stored in the non-volatile memory, ensuring that the updated control rules are maintained after system restart.

[0084] Example 5: The control instruction execution module receives the digital control instruction set generated by the adaptive oil supply control module, which contains three independent parameters: proportional overflow valve opening adjustment value, variable frequency pump rotating speed setting value and piston position calibration amount. The instruction analysis unit adopts a hierarchical decoding mechanism, first identifies the instruction header identifier to determine the control object type, and then extracts the 32-bit floating point number data in the parameter area. For the proportional overflow valve control instruction, the signal conversion unit converts the opening percentage in floating point format to a 0-10V analog voltage signal. This voltage signal is input to the current amplifier circuit, and a 4-20mA drive current proportional to the set value is generated through voltage-current conversion. The drive current is applied to the proportional electromagnet coil to generate an electromagnetic force corresponding to the current intensity. The electromagnetic force pushes the valve core to move axially against the spring pre-tightening force, changing the flow area of the throttle. For example, when the opening instruction value is 65%, the system outputs an 8.25V voltage signal, which is converted to generate a 16.5mA drive current, causing the valve core to displace to the middle position of the designed stroke, at which time the overflow pressure is set to 75% of the system rated pressure.

[0085] The execution process of the variable frequency pump rotating speed control instruction involves power electronics technology. The rotating speed setting value extracted by the instruction analysis unit is input to the digital signal processor, which generates a three-phase pulse width modulation waveform. The waveform generation adopts a space vector modulation algorithm, calculates the required output frequency according to the target rotating speed, and automatically adjusts the voltage amplitude based on the torque-speed characteristic curve of the pump. The modulation waveform drives the insulated gate bipolar transistor power module to convert the DC bus voltage into variable frequency and variable voltage three-phase alternating current. For example, when the target rotating speed is set to 1450rpm, the system outputs 48Hz three-phase alternating current with a voltage amplitude adjusted to 380V line voltage. This electric energy is transmitted to the pump three-phase asynchronous motor through a shielded power cable to drive the pump shaft to rotate. The rotating speed closed-loop control is realized through the encoder installed on the motor shaft end, and the encoder pulse signal is fed back to the digital signal processor, which dynamically adjusts the pulse width modulation waveform duty cycle after comparing with the set value, so that the actual rotating speed is stabilized within the range of ±5rpm of the set value.

[0086] The execution of the piston position calibration instruction relies on the hydraulic cylinder servo control system. The displacement correction amount is packaged into a standard CANopen protocol message by the instruction analysis unit, which contains a 16-bit node address, a 32-bit floating-point position value, and an 8-bit control mode identifier. The message is transmitted to the hydraulic cylinder servo controller through the controller area network bus at a transmission rate of 500 kbps. The communication interface of the servo controller uses optical isolation technology to prevent electromagnetic interference from affecting signal integrity. After message analysis, the position control module inputs the target position value into the position-speed-current three-loop control algorithm. The position loop generates a speed feedforward command, the speed loop outputs a current set value, and the current loop controls the servo valve coil current through a proportional valve drive circuit. For example, when receiving a +0.8 mm displacement correction instruction, the servo controller will push the electro-hydraulic servo valve spool to the corresponding opening position within 50 ms, causing the pressure oil to enter the rodless chamber of the hydraulic cylinder. The piston rod displacement is detected in real time by a magnetostrictive displacement sensor, and its feedback signal is compared with the target position to form a closed-loop control. When the actual position and target position difference is less than 0.02 mm, the controller enters the position holding mode, and compensates for the drift caused by internal leakage by adjusting the valve opening slightly.

[0087] System-level cooperative control is achieved through precise time synchronization mechanism. The control instruction execution module has a high-precision clock source, and all execution actions are scheduled with 1 ms time resolution. The proportional relief valve regulation, variable frequency pump speed regulation, and piston position calibration three control processes use time-sharing triggering strategy to avoid instantaneous power superposition causing power supply fluctuation. The execution state monitoring unit collects feedback signals from each execution mechanism in real time: proportional valve spool displacement sensor signal, variable frequency pump motor current waveform, and servo controller following error value. These signals are sampled through independent analog input channels and converted to digital quantities at a sampling frequency of 500 Hz. When the execution mechanism response delay exceeds the set threshold (such as valve spool action delay > 100 ms) or the following error continues to exceed the standard (such as position error > 0.1 mm for more than 2 seconds), the system automatically triggers the execution exception diagnosis process. This process first switches to the backup control channel, analyzes the fault characteristics, and distinguishes between sensor failure, actuator jamming, or mechanical wear, and uploads the diagnosis results to the central monitoring system.

[0088] Environmental adaptability design considers the special working conditions of the water conservancy project site, all electrical connections use IP67 protection level connectors, signal cables are equipped with double shielding structure. The actuator mounting base is provided with shock absorbing rubber pads, which effectively attenuate mechanical vibrations from the gate opening and closing operation. In low temperature environment, heating tape is wound around the hydraulic valve block to maintain oil viscosity; in high temperature working condition, heat dissipation fins are used to reduce the temperature of electronic components. Modular design allows quick replacement of faulty units, proportional valve drive board, frequency conversion power module and other key components support hot plug operation. Historical execution data is saved in ferroelectric memory in a circular storage mode, recording the control command and execution response curve in the last 200 hours, providing data basis for running state analysis.

[0089] It should be noted that the relational terms herein such as first and second and the like are used solely to distinguish one entity or action from another, without necessarily requiring or implying any such actual relationship or order between such entities or actions. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0090] Although embodiments of the present application have been shown and described, it is to be understood that various modifications, substitutions, replacements and variations can be made to these embodiments without departing from the principles and spirit of the present application, and the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A hydraulic cylinder intelligent oil supply control system applied in water conservancy projects, characterized in that, include: The hydraulic parameter acquisition module is used to acquire in real time multi-point oil pressure monitoring signals from the hydraulic cylinder oil supply line, piston displacement monitoring signals fed back by the displacement sensor, and oil supply flow monitoring signals collected by the flow meter. The data fusion processing module is used to perform multi-source heterogeneous data fusion processing on the oil pressure monitoring signal, displacement monitoring signal and flow monitoring signal output by the hydraulic parameter acquisition module; The manifold state modeling module is used to construct a Riemannian manifold geometric model of the hydraulic cylinder's operating state based on the monitoring signal sequence fused by the data fusion processing module. The dynamic risk prediction module is used to calculate the real-time anomaly clustering index of the hydraulic cylinder oil supply system based on the geometric feature vector output by the Riemannian manifold geometric model. An adaptive oil supply control module is used to generate oil supply control commands for the hydraulic cylinder based on the real-time anomaly clustering index output by the dynamic risk prediction module. The hydraulic parameter acquisition module includes: The first oil pressure sensor is located at the inlet section of the oil supply pipeline, the second oil pressure sensor is located in the middle section of the pipeline, and the third oil pressure sensor is located at the outlet section of the pipeline. Displacement sensor installed on the piston rod of a hydraulic cylinder; An ultrasonic flow meter installed at the outlet of the oil supply pump; The first oil pressure sensor, the second oil pressure sensor, and the third oil pressure sensor output oil pressure monitoring signals, the displacement sensor outputs displacement monitoring signals, and the ultrasonic flow meter outputs flow monitoring signals. When the data fusion processing module performs multi-source heterogeneous data fusion processing: The propagation time delay between the monitoring signals of the first oil pressure sensor and the second oil pressure sensor is calculated using wavelet coherence analysis. The displacement monitoring signal and the flow monitoring signal are time-domain aligned based on the propagation delay. The time-domain aligned oil pressure monitoring signal, displacement monitoring signal, and flow monitoring signal are combined into a multi-dimensional monitoring signal matrix.

2. The intelligent oil supply control system for hydraulic cylinders applied in water conservancy projects according to claim 1, characterized in that, When the manifold state modeling module constructs the Riemannian manifold geometric model: The Hilbert-Huang transform is performed on the multidimensional monitoring signal matrix to extract the intrinsic mode function components of each monitoring signal; The instantaneous frequency and instantaneous amplitude of the intrinsic mode function components are mapped to the tangent space coordinates of the Riemannian manifold; The geometric feature vector of the hydraulic cylinder's operating state is constructed based on the curvature change of the tangent space coordinates.

3. The intelligent oil supply control system for hydraulic cylinders applied in water conservancy projects according to claim 2, characterized in that, When the dynamic risk prediction module calculates the real-time anomaly clustering index: Topological data analysis algorithms are used to identify the high-dimensional clustering structure of the geometric feature vectors on the Riemannian manifold; Calculate the cluster density distribution of the geometric feature vectors in the manifold space when the historical faults occurred; The Hausdorff distance quantifies the real-time anomaly clustering index based on the current geometric feature vector and the historical clustering density distribution.

4. The intelligent oil supply control system for hydraulic cylinders applied in water conservancy projects according to claim 3, characterized in that, The dynamic risk prediction module also includes: The causal correlation strength between oil pressure fluctuations and displacement hysteresis is identified from the geometric feature vector using a causal discovery algorithm. The weighting coefficient of the real-time anomaly clustering index is adjusted based on the strength of the causal relationship.

5. The intelligent oil supply control system for hydraulic cylinders applied in water conservancy projects according to claim 1, characterized in that, When the adaptive fuel supply control module generates fuel supply control commands: A ternary logic system is used to map the real-time anomaly clustering index into oil supply pressure adjustment, flow compensation, and displacement correction. The opening degree of the proportional relief valve is adjusted according to the oil supply pressure adjustment amount; The variable frequency pump speed control signal is generated based on the flow compensation amount; Output piston position calibration command based on the displacement correction amount.

6. The intelligent oil supply control system for hydraulic cylinders applied in water conservancy projects according to claim 5, characterized in that, The adaptive fuel supply control module also includes: Predict the drift trend of fuel supply parameters for the next three operating cycles using a time-series causal reasoning model; When the detected trend of fuel supply parameter drift exceeds the preset threshold, the parameter self-calibration mechanism of the three-valued logic system is triggered.

7. The intelligent oil supply control system for hydraulic cylinders applied in water conservancy projects according to claim 1, characterized in that, Also includes: A control command execution module is used to parse the fuel supply control commands generated by the adaptive fuel supply control module. The control command execution module drives the proportional relief valve to perform an opening adjustment action; The control command execution module controls the variable frequency pump to complete the speed adjustment operation; The control command execution module sends a piston position calibration command to the hydraulic cylinder servo controller.

8. A method for intelligent oil supply control of hydraulic cylinders applied in water conservancy projects, characterized in that, It includes all modules and method flows of the intelligent oil supply control system for hydraulic cylinders applied to water conservancy projects as described in any one of claims 1 to 7.

Images