Self-adaptive pressure regulation and control system and method for motor pump set

Abstract

The invention discloses a pressure self-adaptive regulation and control system and method for a motor pump set, and particularly relates to the field of general control regulation and control, and the method comprises the steps: collecting multi-dimensional data step by step, synchronously collecting basic data such as a pressure time domain sequence, temperature, vibration and the like and derivative parameters, and carrying out the time sequence synchronization processing to form a unified data matrix; extracting multi-dimensional features, extracting dimensionless core features according to four dimensions such as pressure features, and constructing standardized feature vectors; multi-dimensional feature function construction: respectively constructing four types of functions including pressure characteristics and pump body states based on the features of each dimension, and quantifying the performance of each dimension; deep fusion and decision making are carried out, a comprehensive pressure control index is constructed through dynamic weight and kernel function mapping, a swash plate angle adjusting amount is generated in combination with fuzzy reasoning and converted into a control current, and a pump body is driven to adjust the displacement; meanwhile, a three-level quality judgment standard and an exception handling mechanism are established, and pressure control quality monitoring and fault early warning are achieved; the method improves the voltage control precision and stability, and guarantees the safety of equipment.

Description

Technical Field

[0001] This invention relates to the field of general control and regulation technology, and more specifically, to a pressure adaptive regulation system and method for an electric motor pump unit. Background Technology

[0002] In industrial production, oil and gas drilling, and other fields, precise pressure control is directly related to production safety and efficiency. Traditional pressure control methods often rely on data from a single sensor, which is easily affected by equipment noise and environmental interference, resulting in insufficient data accuracy and slow response that makes it difficult to adapt to complex working conditions. For example, in oil and gas drilling, relying solely on pressure sensor feedback can easily lead to misjudgments of well kick and well leakage risks. In industrial production, pressure control logic supported by single flow data often suffers from overshoot problems.

[0003] With the development of IoT technology, it has become possible to acquire multi-dimensional data such as flow, temperature, and vibration. However, the heterogeneous data formats vary greatly and the quality is inconsistent, posing challenges to integrated applications. Existing fusion technologies mostly focus on data splicing and have not achieved deep feature mining, making it difficult to support intelligent decision-making. Therefore, it is urgent to build a multi-source data fusion framework that combines big data analysis and intelligent algorithms to improve the robustness and accuracy of pressure control systems and meet the dynamic pressure control needs in complex scenarios.

[0004] However, it still has some drawbacks in practical use, such as:

[0005] 1. Weak correlation of multi-source data: Existing solutions mostly process pressure, vibration and other data independently, without exploring the implicit correlation between parameters. Pressure control decisions rely on only a single indicator. The data lacks time-series synchronization and deep integration, which can easily lead to decision-making bias due to fluctuations in local parameters. It cannot adapt to the comprehensive pressure control needs under complex working conditions.

[0006] 2. Rigid weighting strategy: The weights of each pressure control parameter are fixed values, which cannot be dynamically adjusted according to pump status, energy consumption, and other scenarios. If the pump malfunctions, pressure control will still be applied according to the conventional weights, which can easily lead to equipment overload; when energy consumption exceeds the limit, it is difficult to optimize in a timely manner, resulting in poor scenario adaptability of the pressure control strategy.

[0007] 3. Insufficient decision-making accuracy: Pressure control commands are generated based on empirical thresholds, lacking in-depth feature extraction and quantitative modeling of the data. A precise mapping between features and adjustment amounts is not established, resulting in coarse adjustment actions when pressure fluctuations are large, making it difficult to achieve high-precision pressure control within ±0.1 MPa.

[0008] 4. Lack of risk prediction: Only pressure is monitored to ensure compliance, without a multi-dimensional quality assessment system. Risks such as pump malfunctions and excessive energy consumption cannot be identified in advance; once malfunctions occur, only passive shutdown is possible. The lack of a tiered early warning mechanism severely impacts production continuity. Summary of the Invention

[0009] In order to overcome the above-mentioned defects of the prior art, the present invention provides a pressure adaptive control system and method for motor pump sets, which solves the problems mentioned in the background art through the following solutions.

[0010] To achieve the above objectives, the present invention provides the following technical solution: a pressure adaptive control system and method for an electric motor pump unit, comprising:

[0011] S1 Multi-dimensional Data Step-by-Step Acquisition: Multi-dimensional basic data of the hydraulic system are synchronously acquired according to a preset sampling frequency. After time-series synchronization processing of all acquired multi-dimensional basic data, a unified data matrix with time-series alignment is formed.

[0012] S2 Multi-Dimensional Feature Extraction: Based on a unified data matrix, features of each dimension are extracted according to four dimensions: pressure characteristics, pump status, displacement response, and energy efficiency, forming a standardized feature vector.

[0013] S3 Multi-Dimensional Feature Function Construction: Based on standardized feature vectors, feature functions are constructed for the four dimensions of pressure characteristics, pump body state, displacement response, and energy efficiency, respectively, including: pressure characteristic function, pump body state function, displacement response function, and energy efficiency function.

[0014] S4 Deep Fusion and Intelligent Pressure Control Decision: Based on four-dimensional feature functions, a comprehensive pressure control index is constructed through dynamic weight allocation and nonlinear mapping. Fuzzy reasoning is performed by combining the comprehensive pressure control index with the feature functions of each dimension to generate the swashplate angle adjustment amount and convert it into control current to drive the actuator to adjust the displacement. At the same time, based on the preset thresholds of the comprehensive pressure control index and the feature functions of each dimension, a quality judgment standard and anomaly handling mechanism are established to realize real-time monitoring of pressure control quality and pressure control execution.

[0015] The technical effects and advantages of this invention are as follows:

[0016] 1. Deep fusion of multiple data: Simultaneously collect multi-dimensional data and align them in time series, strengthen parameter correlation through kernel function mapping, and construct a comprehensive control pressure index; the fusion result fully reflects the system status, solves the problem of data fragmentation in existing technologies, and provides comprehensive support for decision-making;

[0017] 2. Strong adaptability of dynamic weights: The weights of each dimension are dynamically adjusted according to the working conditions. When the pump body is abnormal, the equipment protection weight is increased, and the energy consumption weight is optimized when the energy consumption is low. It breaks the limitations of fixed weights, so that the pressure control strategy can flexibly adapt to different scenarios and ensure the stable and economical operation of the system.

[0018] 3. More precise pressure control decisions: Performance across various dimensions is quantified using feature functions, and adjustment instructions are generated using a fuzzy inference system. Inference rules cover the entire scenario, and precise adjustment values ​​are output after defuzzification, improving pressure control accuracy.

[0019] 4. Predictable Quality Risks: A three-tiered quality assessment standard is established, and comprehensive indices and characteristic functions are monitored in real time. Parameters are recorded for optimal operating conditions, and audible and visual alarms are triggered and faults are located when conditions are not met, thus achieving a closed loop of prediction, early warning, and handling, and improving production continuity. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall structure of the present invention.

[0021] Figure 2 This is a schematic diagram of the S1-S3 process of the present invention.

[0022] Figure 3 This is a schematic diagram of the S4 process of the present invention.

[0023] Figure 4 This is a schematic diagram showing the relationship between the pressure characteristic function value and the comprehensive pressure control index of the present invention. Detailed Implementation

[0024] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0025] refer to Figures 1-4 The pressure adaptive control system and method for an electric motor pump unit shown herein include:

[0026] S1: Multi-dimensional data collection in stages:

[0027] By deploying multiple types of high-precision sensors at key locations in the variable displacement piston pump and hydraulic system, raw data is collected step-by-step according to four dimensions: pressure characteristics, pump status, displacement response, and energy efficiency. A new time-series synchronization processing step is added to ensure data time alignment. Through synchronous sampling and time-series calibration, the integrity, consistency, and temporal correlation of the data are guaranteed, providing comprehensive, reliable, and time-consistent data source support for subsequent feature extraction, function construction, and pressure control decisions. The specific steps are as follows:

[0028] S101: Pressure Characteristics Data Acquisition

[0029] The real-time pressure at the pump outlet is collected by a high-precision diffused silicon pressure sensor deployed on the inner wall of the variable displacement plunger pump outlet pipe and recorded as the pump outlet pressure. ;

[0030] The real-time load pressure is collected by a pressure sensor of the same model deployed in the inlet pipe of the hydraulic actuator and recorded as the load pressure. ;

[0031] The pressure change rate is calculated based on the ratio of the pressure difference between two consecutive adjacent sampling points to the sampling interval. , Where i is the sampling number, The sampling interval;

[0032] Synchronous acquisition continuous The pressure time-domain sequence of each sampling point is denoted as... ,in This ensures that the time intervals of the sequences are uniform.

[0033] S102: Pump body status dimension acquisition:

[0034] The real-time temperature of the pump body stator is collected by a PT100 platinum resistance temperature sensor embedded in the stator winding coil of the pump body, and is denoted as pump body temperature T.

[0035] The vibration acceleration of the bearing housing is collected by a triaxial piezoelectric vibration sensor attached to the end face of the pump body bearing housing, and is denoted as vibration acceleration A. The collection direction is consistent with the tangential direction of bearing rotation.

[0036] Perform a Fast Fourier Transform on the raw voltage signal acquired by the vibration sensor to extract the frequency component with the largest amplitude in the vibration spectrum, which is denoted as the dominant vibration frequency. The rate of temperature change is calculated based on the ratio of the temperature difference between two consecutive adjacent sampling points to the sampling interval. , .

[0037] Synchronous acquisition continuous Temperature sequence of each sampling point With vibration acceleration sequence The timestamp is consistent with S101.

[0038] S103: Displacement Response Dimension Data Collection:

[0039] By using a rod-type displacement sensor mounted on the piston rod of the swashplate drive cylinder in the pump body, and combining the mechanical transmission relationship between the swashplate angle and displacement (transmission ratio: 1mm corresponds to 0.6°), the real-time deflection angle of the swashplate is calculated and recorded as the swashplate angle. ;

[0040] The real-time discharge rate is collected by an electromagnetic flow sensor deployed at the pump's discharge outlet and recorded as the real-time discharge rate Q (unit: L / min); the timing of the pressure control command issuance is also recorded. With the swashplate angle stabilized at the target value The moment The displacement adjustment lag time was calculated. (Unit: ms) Its stability criterion is: the angle change of three consecutive sampling points is ≤0.05°.

[0041] Synchronous acquisition continuous Slope angle sequence of sampling points With displacement sequence The timestamp is strictly aligned with S101.

[0042] S104: Energy efficiency data collection:

[0043] The real-time line current of the motor is collected by a Hall current sensor installed in the power input circuit of the drive motor, and the average value of the three-phase current is recorded as the motor current I.

[0044] The real-time line voltage of the motor is collected by a voltage sensor deployed at the motor input terminal, and the average value of the three-phase voltage is recorded as the motor voltage U.

[0045] The real-time power factor of the motor is acquired by a power factor transmitter connected in series in the motor control circuit, and recorded as the power factor. (dimensionless);

[0046] The real-time input power of the motor is calculated based on the collected electrical parameters and denoted as the input power. Unit: kW For calculating three-phase power;

[0047] Combined with pump outlet pressure The pump output power is calculated based on the real-time displacement Q using the hydraulic power formula and denoted as the output power. Unit: kW. Since 1 MPa·L / min = 60 kW, divide by 60 to achieve unit conversion.

[0048] Synchronous acquisition continuous Current sequence at each sampling point Voltage sequence Input power sequence and output power sequence The timestamp is consistent with the preceding dimension.

[0049] S105: Data Acquisition Timing Synchronization Processing:

[0050] To address the data timing discrepancies caused by multi-sensor sampling delays and clock asynchrony, timing calibration and alignment are performed on all collected dimensional data. The specific steps are as follows:

[0051] Uniform timestamp annotation: Based on the system's globally unified clock (accuracy ±1μs), a unique timestamp is added to each sampling point of each sensor. The timestamp format is as follows: , The start time of data collection. Ensure the time traceability of the original data;

[0052] Reference time axis established: The sampling timestamp of the pressure sensor in S101 is used as the reference time axis. Since the pressure signal is the core feedback quantity for pressure control, its timing accuracy has the highest priority.

[0053] Asynchronous data interpolation completion: For sampling data from temperature, vibration, displacement, flow, and electrical parameter sensors, if the deviation between the timestamp of a sampling point and the reference time axis exceeds 2μs, linear interpolation is used for completion.

[0054] Data deviation verification and rejection: Calculate the sampling timestamps of each sensor. Corresponding time to the reference time axis deviation ,like If 1% of the sampling period is considered invalid, the data at that sampling point is removed and then re-filled by interpolation to avoid abnormal time series data affecting subsequent analysis.

[0055] Unified data matrix construction: All time-series aligned data are organized into a unified data matrix according to the baseline time axis. Each row corresponds to a time node. Each column corresponds to a collected physical quantity, and the matrix form is as follows:

[0056]

[0057] Ensure that subsequent feature extraction and function construction are carried out based on a standardized dataset with the same time dimension to avoid fusion errors caused by temporal misalignment.

[0058] S2: Multi-dimensional feature extraction:

[0059] Based on the unified data matrix D after S1 time synchronization, mathematical transformations, statistical analysis, and normalization are performed on the raw data of four dimensions: pressure characteristics, pump status, displacement response, and energy efficiency. Feature quantities that can accurately characterize the core performance of each dimension are extracted. The high-dimensional, multi-unit raw data are transformed into low-dimensional, strongly correlated, dimensionless standardized feature indicators, eliminating the influence of dimensional differences on subsequent function construction and fusion analysis. The specific steps are as follows:

[0060] S201: Pressure Characteristic Dimension Feature Extraction:

[0061] Pressure time-domain sequence acquired by S101 The core features reflecting pressure stability and trends are extracted, and the specific steps are as follows:

[0062] Pressure fluctuation entropy extraction: First, the pressure time-domain sequence is normalized to extract the original pump outlet pressure value. Mapping to the [0, 1] interval yields the normalized pressure value. The normalization formula is: Based on information entropy theory, the fluctuation entropy of the normalized pressure time-domain sequence is calculated. This is used to characterize the randomness and stability of pressure signals. The smaller the fluctuation entropy, the more stable the pressure. The calculation formula is: ,in To find the minimum value, avoid ln(0) becoming meaningless;

[0063] Pressure Second Derivative Extraction: Pressure Change Rate Calculated Based on S101 For i = 1, 2, ..., 9, the second derivative of pressure is calculated using the quadratic difference method. (Unit: MPa / s²), representing the trend intensity and acceleration of pressure change, is calculated using the following formula: ;

[0064] Dimensionless processing of features: The second derivative of pressure is normalized and mapped to the interval [0, 1] to obtain the dimensionless second derivative of pressure. The normalization formula is: The core feature of the final extracted pressure characteristic dimension is: fluctuation entropy. Mean of the second derivative of pressure .

[0065] S202: Pump body state dimension feature extraction:

[0066] Temperature sequence acquired by S102 Vibration acceleration sequence and vibration frequency The core features reflecting the thermal state and structural vibration state of the pump body are extracted, and the specific steps are as follows:

[0067] Temperature gradient normalization: Temperature change rate calculated based on S101 For i = 1, 2, ..., 9, take their absolute values ​​and then normalize them to obtain the dimensionless temperature gradient. This characterizes the degree of drastic change in pump body temperature, and the calculation formula is: The mean of the nine sampling points is taken as one of the features of this dimension, i.e. ;

[0068] Mean square value extraction of vibration acceleration: for vibration acceleration sequences Calculate the mean square value The unit is g, which characterizes the energy intensity of the vibration signal. The larger the mean square value, the more intense the vibration. The calculation formula is: ,right After normalization, the dimensionless mean square value of vibration acceleration is obtained. The normalization formula is: ,in This represents the maximum measurement range of the vibration sensor.

[0069] Vibration frequency deviation extraction: based on the pump body's natural frequency Calculate the dominant frequency of vibration Relative deviation from the inherent main frequency The degree of abnormality in the pump body vibration state is characterized by the following formula: The core feature of the pump body state dimension extracted in the end is: mean temperature gradient. , mean square value of vibration acceleration Vibration dominant frequency deviation .

[0070] S203: Displacement Response Dimension Feature Extraction:

[0071] Based on the swashplate angle sequence, displacement sequence, and adjustment lag time collected by S103, the core features reflecting the displacement adjustment accuracy and response speed are extracted. The specific steps are as follows:

[0072] Displacement error accumulation extraction: based on swashplate angle With displacement coefficient Calculate the theoretical displacement at each time point. and with real-time displacement Calculate the absolute error The unit is L / min; the cumulative displacement error is obtained by integrating the error sequence. This characterizes the accuracy loss of displacement adjustment, with the integration range being the time interval (0-45ms) corresponding to 10 sampling points. The calculation formula is as follows: ,right After normalization, the dimensionless cumulative displacement error is obtained. The normalization formula is: ,in, , This is the maximum deflection angle of the swashplate, i.e. ;

[0073] Response lag rate extraction: based on rated lag time Calculate the actual lag time The ratio of the rated lag time This ratio characterizes the hysteresis properties of displacement adjustment; the smaller the ratio, the faster the response. The calculation formula is: ,like ,but Upper limit truncation to avoid the influence of outliers;

[0074] Displacement-Response Consistency Extraction: Calculating Displacement Sequence coefficient of variation The coefficient of variation characterizes the stability of displacement within the sampling period. The smaller the coefficient of variation, the more stable the response. The calculation formula is: ,in The standard deviation of the displacement series. This represents the average displacement.

[0075] The core feature of the finally extracted displacement response dimension is: cumulative displacement error. Response lag rate Displacement variation coefficient .

[0076] S204: Energy efficiency dimension feature extraction:

[0077] For the input power sequence acquired by S104 The core features reflecting energy utilization efficiency and power stability are extracted, and the specific steps are as follows:

[0078] Energy efficiency ratio extraction: Calculate the energy efficiency ratio of the pump at each sampling time. The mean of 10 sampling points is taken as the core feature of this dimension, namely the average energy efficiency ratio. The higher the energy efficiency ratio, the higher the energy conversion efficiency.

[0079] Power fluctuation coefficient extraction: Calculate the fluctuation coefficient of the input power sequence. The fluctuation coefficient characterizes the stability of the input power; the smaller the fluctuation coefficient, the more stable the power. The calculation formula is: ,in This represents the average input power.

[0080] Output power percentage extraction: Calculate the average output power With the average input power ratio To further quantify energy consumption efficiency, the calculation formula is as follows: ;

[0081] The core feature of the final extracted energy efficiency dimension is: average energy efficiency ratio. Power fluctuation coefficient Output power ratio .

[0082] Feature extraction results summary: This step ultimately outputs 11 features across 4 dimensions, forming a standardized feature vector. All feature values ​​are within the range of

[0083] S3: Construction of multi-dimensional feature functions:

[0084] Based on the standardized feature vector F output by S2, the relevant components in the feature vector are selectively fused according to four dimensions: pressure characteristics, pump status, displacement response, and energy efficiency. Feature functions for each dimension are constructed using deep mathematical methods such as coupled weighting, nonlinear transformation, and exponential correction. The specific steps are as follows:

[0085] S301: Constructing the pressure characteristic function :

[0086] Based on two pressure-related components in the standardized feature vector F: pressure fluctuation entropy. With the mean of the second derivative of pressure A pressure characteristic function is constructed using a dynamic weight coupling method to balance the influence weights of stability and trend. The specific process is as follows:

[0087] Weighting coefficient determination: Introducing a pressure stability priority coefficient Priority coefficient for pressure change trend This highlights the crucial role of pressure stability in pressure control.

[0088] The pressure characteristic function is constructed by combining weighted coupling and nonlinear correction. , where tanh() is the hyperbolic tangent function, used to nonlinearly compress the mean of the second derivative of pressure to avoid extreme values ​​dominating the function result, and at the same time to map it to the interval [0,1] to ensure the uniformity of the overall dimensions of the function; The value range is [0, 1]. The smaller the value, the more stable the pressure time-domain sequence, the smoother the pressure change trend, and the better the pressure characteristics.

[0089] S302: Constructing the pump body state function :

[0090] Based on the three pump state-related components in the standardized feature vector F—mean temperature gradient, mean square vibration acceleration, and vibration dominant frequency deviation—a pump state function is constructed using product superposition and threshold normalization to reflect the coupled influence of temperature and vibration. The specific process is as follows:

[0091] Temperature anomalies and vibration anomalies can exacerbate pump body losses. A product-based approach is used to characterize their synergistic coupling relationship, while a linear term is introduced to reflect the independent impact of vibration frequency deviation, ensuring comprehensive coverage of the pump body's operating state. The pump body state function is constructed as follows: The denominator is the theoretical maximum value of the combined features, and this normalization process ensures... The value range is strictly controlled within [0, 1]; A smaller value indicates a healthier pump body, while a larger value suggests a risk of overheating, severe vibration, or abnormal vibration frequency. If the pump body is determined to be close to an abnormal state, the adjustment range should be limited in subsequent pressure control decisions to protect the equipment.

[0092] S303: Constructing the displacement response function :

[0093] Based on the three displacement-response-related components in the standardized eigenvector F—displacement error accumulation, response hysteresis rate, and displacement variation coefficient—a displacement response function is constructed by combining integral effect enhancement and exponential penalty, highlighting the core requirements of accuracy and response speed. The specific process is as follows:

[0094] The cumulative displacement error is weighted using a squared term to enhance accuracy loss; the response hysteresis rate is penalized using an exponential term for non-linearity; and the displacement coefficient of variation is weighted using a linear term to reflect stability. The sum of the weights of these three terms is 1 to ensure function balance. The displacement response function is constructed as follows: ,in , , These are the weighting coefficients for each item. Mapping the lag rate to the interval [0, 1] makes the larger the lag rate, the smaller the output value of this term, and the more obvious the effect of the penalty term. The value range is [0, 1]. The smaller the value, the higher the displacement adjustment accuracy, the faster the response speed, and the more stable the displacement output.

[0095] S304: Constructing the energy efficiency function Based on the three energy efficiency-related components in the standardized feature vector F—average energy efficiency ratio, power fluctuation coefficient, and output power ratio—an energy efficiency function is constructed using a combination of linear weighting and positive excitation, emphasizing the principle of prioritizing energy efficiency. The specific process is as follows:

[0096] The average energy efficiency ratio and the output power ratio are positively correlated, so a positive weighting method is used to assign them high weights. The power fluctuation coefficient is negatively correlated, so a negative correction method is used to weaken its adverse effects. At the same time, a reference offset is introduced to avoid the function value being negative. An energy efficiency function is constructed. : ,in , , These are the weighting coefficients. As the reference offset, it is normalized. Mapping to the interval [0, 1], the normalization formula is: The normalized values ​​were used in the subsequent fusion analysis. Still recorded as To ensure that the dimensions are consistent with those of other characteristic functions; A higher value indicates higher energy efficiency, more stable input power, and better energy economy in the voltage control process.

[0097] S4: Deep Integration and Intelligent Pressure Control Decision-Making

[0098] Based on four feature functions constructed using S3, a deep fusion model of dynamic weight allocation and kernel function mapping is introduced to transform scattered dimensional indicators into a comprehensive pressure control evaluation index. This is then combined with a multi-input fuzzy inference system to generate refined pressure control instructions, while simultaneously linking with an online quality detection mechanism. This achieves an integrated pressure control logic encompassing fusion, decision-making, execution, and quality inspection. The specific steps are as follows:

[0099] S401: Construction of a Deep Fusion Model for Four-Dimensional Feature Functions

[0100] To balance the core priorities of features across various dimensions with dynamic adaptability to different scenarios, a two-level fusion strategy of dynamic weights and kernel function mapping is adopted to construct a comprehensive pressure control index. The specific process is as follows:

[0101] Dynamic weight determination logic: Define pressure characteristic weight coefficients Displacement response weighting coefficient Constraint weight coefficients Energy efficiency weighting coefficient The sum is 1;

[0102] Dynamically adjust weights based on priority of pressure control scenarios: Under normal pressure control scenarios , , , Emphasizing pressure and displacement as key parameters; pump body condition approaching abnormality. Automatically adjust to , , , Strengthen equipment protection; energy efficiency is too low. When, adjust to , , , Optimize energy consumption.

[0103] Kernel function mapping and fusion formula:

[0104] A radial basis function kernel is used to perform nonlinear mapping on each characteristic function to strengthen the implicit correlation between dimensions. Then, combined with dynamic weighted summation, a comprehensive pressure control index is constructed. : ,in For radial basis kernel functions, The mean, Standard deviation; The value range is [0, 1]. The larger the value, the better the overall pressure control status, providing a single and comprehensive evaluation basis for subsequent decision-making.

[0105] It should be further noted that, to verify the relationship between the pressure characteristic function and the comprehensive pressure control index, a single variable method was used, fixing... , , Simply by adjusting the system load A uniform gradient change was applied within the interval [0, 1]. The comprehensive pressure control index was calculated using the fusion formula to verify the relationship between the two. The experiment used a normal pressure control scenario, and the results are shown in the table below:

[0106]

[0107] S402: Generation of multi-input fuzzy inference control commands:

[0108] Introducing a multi-input single-output fuzzy inference system, to , , As the input variable, the swashplate angle adjustment amount To generate output variables, dynamic pressure control commands are generated using a refined rule base. The specific process is as follows:

[0109] Input / output fuzzification:

[0110] Input variable fuzzy subset definition:

[0111] : {Excellent (U) [0, 0.2], Good (G) [0.1, 0.4], Average (M) [0.3, 0.6], Poor (B) [0.5, 0.8], Inferior (W) [0.7, 1.0]};

[0112] : {Stable (S) [0, 0.2], Relatively Stable (RS) [0.1, 0.4], Fluctuating (F) [0.3, 0.6], Large Fluctuation (LF) [0.5, 1.0]};

[0113] : {Accuracy (A) [0, 0.2], Relatively Accurate (RA) [0.1, 0.4], Deviation (D) [0.3, 0.6], Large Deviation (LD) [0.5, 1.0]};

[0114] Output variables : {Significant decrease (LD-) [-2.0, -1.2], Slight decrease (SD-) [-1.5, -0.5], Maintain (H) [-0.3, 0.3], Slight increase (SD+) [0.5, 1.5], Significant increase (LD+) [1.2, 2.0]};

[0115] All variables are converted from precise values ​​to fuzzy quantities using triangular membership functions.

[0116] Fuzzy rule base design: 36 fuzzy rules covering the entire scenario are constructed, designed based on the principle of prioritizing core indicators and correcting constraint indicators. Some key rules are as follows:

[0117] Rule 1: If For excellence For stability and As the standard, then To maintain;

[0118] Rule 2: If For good For more stable and As the standard, then The increase was slight (+0.8°).

[0119] Rule 3: If For good For fluctuation and For greater accuracy, then It increased slightly (+1.0°).

[0120] Rule 4: If For China For fluctuation and If it is a deviation, then The increase was slight (+1.2°).

[0121] Rule 5: If For China For stability and For a large deviation, then This represents a significant increase (+1.6°).

[0122] Rule 6: If For difference, For large fluctuations and If it is a deviation, then To significantly reduce (-1.5°), priority should be given to stabilizing the pressure;

[0123] Rule 7: If For difference, For more stable and For a large deviation, then To significantly increase (+1.8°), priority will be given to supplementing emissions;

[0124] Rule 8: If inferior For large fluctuations and For a large deviation, then To significantly reduce (-1.8°) and pump body protection commands;

[0125] Defuzzification and instruction conversion: The centroid method is used to convert the fuzzy inference results into precise swashplate angle adjustment values. To ensure smooth output;

[0126] Using the linear mapping formula Converted to the control current I (unit: mA) of the electro-hydraulic proportional valve: ,in This represents the current swashplate angle (unit: °). , , The control current is output to the electro-hydraulic proportional valve through the PLC, which drives the variable piston pump to adjust the displacement and achieve precise closed-loop pressure control.

[0127] S403: Online Quality Inspection Linkage Mechanism

[0128] Based on comprehensive pressure control index By combining threshold values ​​with characteristic functions of various dimensions, a three-level quality judgment standard is established to achieve real-time monitoring and early warning of abnormalities in pressure control quality, as detailed below:

[0129] Quality grade classification:

[0130] Grade I (Excellent): ≥0.7、 ≤0.2、 ≤0.2、 ≤0.3 and ≥0.6; Corresponding indicators: pressure accuracy within ±0.1MPa, energy efficiency ≥85%, product qualification rate ≥99%;

[0131] Level II (Qualified): 0.4≤ <0.7、 ≤0.4、 ≤0.4、 ≤0.5 and ≥0.4; Corresponding indicators: pressure accuracy ±0.1~0.2MPa, energy efficiency 75%-85%, product qualification rate 95%-99%;

[0132] Level III (Unqualified): <0.4 or >0.6 or >0.6 or >0.7 or <0.4; Corresponding indicators: pressure accuracy exceeds ±0.2MPa, energy efficiency <75%, product non-conformity risk >5%;

[0133] Exception handling mechanism:

[0134] Level I: Maintain current pressure control parameters and record optimal operating parameters to the database;

[0135] Level II: Output quality focus prompts, dynamic fine-tuning of energy consumption weights. Increased to 0.2, optimizing energy efficiency;

[0136] Level III: Immediately triggers audible and visual alarms, suspends pressure regulation, outputs fault diagnosis information, and resumes automatic pressure control after technicians have checked the sensors, pump body, or parameter configuration.

[0137] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0138] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A pressure adaptive control system and method for an electric motor pump set, characterized in that, include: S1 Multi-dimensional Data Step-by-Step Acquisition: Multi-dimensional basic data of the hydraulic system are synchronously acquired according to a preset sampling frequency. After time-series synchronization processing of all acquired multi-dimensional basic data, a unified data matrix with time-series alignment is formed. S2 Multi-Dimensional Feature Extraction: Based on a unified data matrix, features of each dimension are extracted according to four dimensions: pressure characteristics, pump status, displacement response, and energy efficiency, forming a standardized feature vector. S3 Multi-Dimensional Feature Function Construction: Based on standardized feature vectors, feature functions are constructed for the four dimensions of pressure characteristics, pump body state, displacement response, and energy efficiency, respectively, including: pressure characteristic function, pump body state function, displacement response function, and energy efficiency function. S4 Deep Fusion and Intelligent Pressure Control Decision: Based on four-dimensional feature functions, a comprehensive pressure control index is constructed through dynamic weight allocation and nonlinear mapping. Fuzzy reasoning is performed by combining the comprehensive pressure control index with the feature functions of each dimension to generate the swashplate angle adjustment amount and convert it into control current to drive the actuator to adjust the displacement. At the same time, based on the preset thresholds of the comprehensive pressure control index and the feature functions of each dimension, a quality judgment standard and anomaly handling mechanism are established to realize real-time monitoring of pressure control quality and pressure control execution.

2. The pressure adaptive control system and method for a motor pump set according to claim 1, characterized in that: The unified data matrix includes: A unified precision timestamp is added to each sampling point of all multi-dimensional basic data, and a reference time axis is established based on the sampling timestamp of the pressure time domain sequence. For sampling points in other data whose timestamps deviate from the reference time axis by a preset threshold, linear interpolation is used to complete the data. The deviation between the sampling timestamp of each data and the corresponding time of the reference time axis is calculated, invalid sampling points with deviations exceeding the set threshold are removed, and the data is re-interpolated and completed. Finally, a time-aligned unified data matrix is ​​formed, with each row corresponding to a sampling time and each column corresponding to a multi-dimensional basic data item.

3. The pressure adaptive control system and method for a motor pump set according to claim 1, characterized in that: The standardized feature vector includes: Based on a unified data matrix, dimensionless processing of features across various dimensions is achieved through normalization, statistical analysis, and mathematical transformation. Core features are extracted according to preset dimensions. Among them, features extracted for the pressure characteristic dimension include pressure fluctuation entropy and the mean of the second derivative of pressure; features extracted for the pump body state dimension include the mean of temperature gradient, the mean square value of vibration acceleration, and the deviation of the dominant vibration frequency; features extracted for the displacement response dimension include the cumulative displacement error, the response hysteresis rate, and the displacement variation coefficient; and features extracted for the energy efficiency dimension include the average energy efficiency ratio, the power fluctuation coefficient, and the output power ratio. All of the above features are arranged in a preset order to form a standardized feature vector with a unified format.

4. The pressure adaptive control system and method for a motor pump set according to claim 1, characterized in that: The pressure characteristic function includes: The pressure fluctuation entropy and the mean of the second derivative of pressure in the pressure characteristic dimension of the standardized feature vector are used as the basis for construction; the priority weights of pressure stability and pressure change trend are set, and the sum of the weights of the two is 1. The above two features are fused through weighted coupling; the mean of the second derivative of pressure is nonlinearly compressed and corrected by the hyperbolic tangent function, and finally the pressure characteristic function is constructed.

5. The pressure adaptive control system and method for an electric motor pump set according to claim 1, characterized in that: The pump body state function includes: The pump body state function is constructed based on the mean temperature gradient, mean square vibration acceleration, and vibration dominant frequency deviation in the standardized feature vector of the pump body state dimension. The dimensionless mean temperature gradient and the dimensionless mean square vibration acceleration are fused by product superposition, and a linear term of vibration dominant frequency deviation is introduced. The pump body state function is then normalized based on the theoretical maximum value of the combined features.

6. The pressure adaptive control system and method for an electric motor pump set according to claim 1, characterized in that: The displacement response function includes: Based on the displacement response dimension of the standardized feature vector, the cumulative displacement error, response lag rate, and displacement coefficient of variation are used as the foundation. The cumulative displacement error is processed with a square term to strengthen the weight of the impact of accuracy loss, the response lag rate is processed with an exponential penalty term to highlight the negative impact of lag characteristics, and the displacement coefficient of variation is processed with a linear term to reflect the impact of displacement stability. Weight coefficients are set for the three features, and the sum of the three weight coefficients is 1. The features processed above are fused by a weighted summation method. Finally, a dimensionless displacement response function is constructed.

7. The pressure adaptive control system and method for an electric motor pump unit according to claim 1, characterized in that: The energy efficiency function includes: The standardization feature vector is based on the average energy efficiency ratio, power fluctuation coefficient, and output power ratio. The average energy efficiency ratio and output power ratio are positively weighted to enhance their positive contribution to energy efficiency, while the power fluctuation coefficient is negatively corrected to weaken its adverse effect on energy efficiency. Weight coefficients are set for the three features, and a benchmark offset is introduced to avoid negative function values. The processed features are then fused by weighted summation. The fused results are normalized to finally construct a dimensionless energy efficiency function.

8. The pressure adaptive control system and method for a motor pump set according to claim 1, characterized in that: The deep fusion includes: Based on the pressure characteristic function, pump state function, displacement response function, and energy efficiency function, core weight coefficients and constraint weight coefficients are set for each function, and the sum of all weight coefficients is 1. The weight allocation is dynamically adjusted according to the hydraulic system pressure control scenario: under normal pressure control scenario, the weight ratio of pressure characteristic function and displacement response function is higher than that of pump state function and energy efficiency function; when the value of pump state function exceeds the preset health threshold, the weight ratio of pump state function is increased; when the value of energy efficiency function is lower than the preset economic threshold, the weight ratio of energy efficiency function is increased. A radial basis function kernel function is used to perform nonlinear mapping on each characteristic function to strengthen the implicit correlation between dimensions. The value inversion transformation is used for pump state function to keep its optimization trend consistent with other functions. The nonlinearly mapped characteristic functions and their corresponding dynamic weight coefficients are weighted and summed, and the summation result is normalized to construct a comprehensive pressure control index.

9. The pressure adaptive control system and method for a motor pump set according to claim 1, characterized in that: The fuzzy reasoning includes: The comprehensive pressure control index, pressure characteristic function, and displacement response function are used as input variables, and the swashplate angle adjustment is used as the output variable. The input variables all take values ​​of [0, 1], and the swashplate angle adjustment takes values ​​of [-2.0°, 2.0°]. The input variables are divided into preset fuzzy subsets. The comprehensive pressure control index is divided into: excellent, good, medium, poor, and bad. The pressure characteristic function is divided into: stable, relatively stable, fluctuating, and large fluctuation. The displacement response function is divided into: accurate, relatively accurate, deviation, and large deviation. The output variables are divided into: large decrease, small decrease, maintain, small increase, and large increase. All variables are converted from precise values ​​to fuzzy quantities through the triangular membership function. A rule base is constructed, and the center of gravity method is used to transform the fuzzy inference results into precise swashplate angle adjustment amounts. Through a linear mapping formula with parameters of minimum control current 4mA, maximum control current 20mA, and maximum swashplate deflection angle 18°, the adjustment amount is converted into the control current of the electro-hydraulic proportional valve, which drives the variable displacement piston pump to adjust the displacement and achieve closed-loop pressure control.

10. The pressure adaptive control system and method for a motor pump set according to claim 1, characterized in that: The quality assessment criteria and anomaly handling mechanism include: Three-level quality assessment: Based on the comprehensive pressure control index, pressure characteristic function, pump body state function, displacement response function, and energy consumption efficiency function, three levels of standards are defined: Level I Excellent, Level II Qualified, and Level III Unqualified. Graded processing mechanism: Level I records high-quality operating parameters; Level II outputs parameter optimization instructions, fine-tuning the energy efficiency weight to 0.2; Level III immediately triggers audible and visual alarms, suspends pressure control, and outputs fault information containing out-of-standard indicators, and resumes automatic pressure control after fault investigation.

Images