Oil pump working condition response control optimization method and system based on machine learning

Abstract

The invention discloses an oil pump working condition response control optimization method and system based on machine learning, and relates to the technical field of oil pump response control, and the method comprises the steps: obtaining historical working condition data of a multi-pump parallel system, analyzing the influence of load change on the output stability of each oil pump, and setting an initial main pump load distribution proportion and a reserve capacity reservation proportion; generating candidate inter-pump load distribution combinations based on the full-factor grid, applying the candidate inter-pump load distribution combinations to oil pump working condition response control one by one, and extracting load distribution characteristics; training an evaluation model by utilizing machine learning, and outputting a load distribution proportion evaluation value corresponding to each combination; mapping and interpolating the evaluation values to generate a smooth load distribution proportion display graph; the optimal inter-pump load distribution proportion is analyzed and screened out according to system load requirements, and oil pump dynamic optimization control is achieved; the problems that the system is lagged in pressure response, large in output fluctuation and unbalanced in energy consumption and mechanical wear under the working conditions of sudden load change and high dynamic are solved.

Description

Technical Field

[0001] This invention relates to the field of oil pump response control technology, and more specifically, to a machine learning-based method and system for optimizing oil pump operating condition response control. Background Technology

[0002] Existing hydraulic system pump operating condition response control methods encompass various technical approaches, such as constant pressure oil supply methods based on pressure feedback, variable frequency speed regulation methods based on load prediction, and displacement adjustment methods based on flow demand. These methods are widely used in the pump control of construction machinery, industrial hydraulic power units, and large equipment. Traditional pump control systems typically adjust pump speed, pump displacement, and the load distribution ratio between pumps using fixed logic or empirical parameters to meet the system's transient pressure requirements and stable output. However, these methods often rely on manually calibrated static parameters, which cannot adequately handle load fluctuations under complex operating conditions and are difficult to maintain optimal response performance during long-term operation.

[0003] In practical applications, the main problems faced by oil pump response control include: difficulty in ensuring the dynamic stability of system pressure when the load changes frequently, performance imbalance caused by differences in efficiency and wear between different pumps, and pressure fluctuations caused by pump speed regulation lag. Although traditional control methods can achieve relatively reliable oil supply under specific operating conditions, their high dependence on fixed parameter configuration makes it difficult for the system to obtain continuous and stable control performance in multi-condition and highly dynamic environments. This, to some extent, limits the intelligent and efficient development of oil pump response control.

[0004] The aforementioned disclosed technical solutions have at least the following technical problems: In multi-pump parallel systems, the setting of the load distribution ratio between pumps has a significant impact on system pressure stability and energy utilization efficiency, but existing systems generally use a fixed distribution ratio. However, if this load distribution ratio is not set properly, it will introduce additional system errors, affecting response performance and equipment lifespan.

[0005] If the load distribution ratio is set too high, although the main pump can quickly provide the necessary pressure and flow under high load demand, it will also cause the main pump to operate under high load for a long time, leading to problems such as accelerated temperature rise, increased energy consumption, and accelerated mechanical wear. Under certain operating conditions, this over-distribution may also cause pressure mismatch between the pump and the actuator, causing oscillations in the system response, thereby affecting the overall operational stability and control accuracy.

[0006] On the other hand, if the load distribution ratio is set too low, although it can reduce the pressure load on the main pump under light load conditions and improve the overall energy efficiency of the system, the main pump cannot output sufficient pressure in time during sudden load changes or high-speed operation, resulting in a delayed system response or even a momentary pressure drop. In this case, the auxiliary pump may be forced to bear an excessively high load for a short period of time, thereby causing an imbalance in the oil supply of the entire system and reducing the stability and reliability of the operating response.

[0007] Therefore, a balance must be found between two factors when selecting the load distribution ratio between pumps to ensure that the system has sufficient dynamic response capability while maintaining reasonable energy consumption, controllable temperature rise, and stable equipment life. However, existing technologies often rely too heavily on fixed coefficients or ignore the differences in pump conditions in load distribution control, resulting in unsatisfactory oil supply performance under different operating conditions and affecting the overall performance and stability of oil pump operating condition response control.

[0008] To address the above problems, this invention proposes a solution. Summary of the Invention

[0009] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a machine learning-based oil pump operating condition response control optimization method and system. By analyzing historical operating condition data of a multi-pump parallel system, a mapping model between pump load distribution characteristics and system response performance is established. Candidate load distribution ratios are generated based on a full-factor grid. The optimal distribution ratio is selected through machine learning evaluation and smoothing visualization to achieve dynamic load optimization of the main pump and auxiliary pump. This solves the problems of pressure response lag, large output fluctuations, and uneven energy consumption and mechanical wear under sudden load changes and high dynamic operating conditions.

[0010] To achieve the above objectives, the present invention provides the following technical solution:

[0011] A machine learning-based method for optimizing oil pump operating condition response control includes the following steps: acquiring historical data of a multi-pump parallel system under a target operating condition, analyzing the impact of load changes on the output stability of each oil pump under the target operating condition, and setting an initial inter-pump load allocation ratio; the inter-pump load allocation ratio includes a main pump load allocation ratio and a reserve capacity ratio; performing Cartesian product on the initial inter-pump load allocation ratio at several equal division points based on a full-factor grid to generate several first inter-pump load allocation ratios; applying the first inter-pump load allocation ratios one by one to the oil pump operating condition response control, and extracting features during the application process to obtain load allocation features; training a load allocation ratio evaluation model based on machine learning according to the load allocation features, and outputting the corresponding load allocation ratio evaluation value; mapping and smoothing the load allocation ratio display diagram based on each first inter-pump load allocation ratio and the corresponding load allocation ratio evaluation value to obtain a load allocation ratio display diagram; analyzing the load allocation ratio display diagram according to demand data to select the optimal inter-pump load allocation ratio for controlling the oil pump.

[0012] In a preferred embodiment, the step of acquiring historical data of the multi-pump parallel system under the target operating condition, analyzing the impact of load changes on the output stability of each pump under the target operating condition, and setting an initial inter-pump load allocation ratio specifically involves: setting the initial inter-pump load allocation ratio, including setting the main pump, auxiliary pumps, and setting the initial main pump load allocation ratio and the initial reserve capacity ratio; the auxiliary pump is not a single pump, but the sum of all pumps except the main pump; after determining the allocation ratio of a main pump, the next cycle begins, the main pump of the previous cycle is removed, and the auxiliary pump of the previous cycle is set as 1, and the main pump, auxiliary pumps, and the initial main pump load allocation ratio and the initial reserve capacity ratio are set again, and the cycle continues until all pumps are allocated; the total load of the auxiliary pumps in the next cycle is the remaining load after the main pumps share the load and the reserve capacity is reserved in the previous cycle.

[0013] In a preferred embodiment, the setting of the main pump and auxiliary pumps specifically involves: dividing the total system load sequence into segments according to a preset load variation range to obtain several load intervals, each interval containing oil pump output pressure, flow rate, and speed data for the corresponding time period; statistically analyzing the maximum, minimum, and average output pressure of the oil pumps in each load interval to determine the pressure fluctuation range; statistically analyzing the maximum, minimum, and average speed deviations and average speed of the oil pumps in each load interval to determine the speed stability range; statistically analyzing the maximum, minimum, and average output flow rates of the oil pumps in each load interval to determine the flow rate fluctuation range; and comparing the pressure fluctuation range, flow rate fluctuation range, and speed stability range of each oil pump in each load interval, determining the oil pump with the largest number of pumps and the most stable performance as the main pump, and combining the remaining oil pumps into an auxiliary pump group.

[0014] In a preferred embodiment, setting the initial main pump load allocation ratio and the initial standby capacity reserve ratio specifically involves: initially setting the main pump load allocation ratio to enable the main pump to carry as much flow as possible in a preset high stability range, while ensuring that the total system flow demand can be met in all load ranges, thus obtaining the initial main pump load allocation ratio; and analyzing the instantaneous load changes of the system during sudden load increases and instantaneous main pump efficiency degradation based on the historical total system load demand curve, and setting the initial standby capacity reserve ratio.

[0015] In a preferred embodiment, the step of generating several first pump load allocation ratios by performing Cartesian products on the initial pump load allocation ratios based on a full-factor grid at several equal division points is as follows: taking the initial main pump load allocation ratio and the initial reserve capacity ratio as the center, the variation range of the main pump load allocation ratio and the variation range of the reserve capacity ratio are preset according to percentages; the variation range of the main pump load allocation ratio is divided into several equal division points, and the variation range of the reserve capacity ratio is divided into several equal division points, resulting in a sequence of equal divisions of the main pump load allocation ratio and the reserve capacity ratio; the Cartesian products of the sequence of equal divisions of the main pump load allocation ratio and the sequence of equal divisions of the reserve capacity ratio are combined to obtain all combinations of the main pump load allocation ratio and the reserve capacity ratio, and each combination forms a first pump load allocation ratio.

[0016] In a preferred embodiment, the first pump load distribution ratio is applied one by one to the oil pump operating condition response control, and feature extraction is performed during the application process to obtain load distribution features. Specifically, the following steps are taken: quantify the characteristics of excessively high and low loads on the main pump; record the average load percentage of the main pump during the entire operating condition response process, the fluctuation amplitude of the main pump output flow and pressure, and the response lag time calculated based on the deviation of the main pump output pressure and flow from the system requirements during the high load change phase; and statistically analyze the recorded data according to the time series to obtain the maximum value, average value, and standard deviation to obtain the load distribution features.

[0017] In a preferred embodiment, the step of training a load allocation ratio evaluation model based on machine learning according to load allocation features and outputting the corresponding load allocation ratio evaluation value specifically involves: using the load allocation feature vector corresponding to each first pump room load allocation ratio as an input sample and the corresponding system response performance index as the model's target output value; dividing all input samples proportionally into a training set and a validation set for model training and performance verification; selecting a supervised learning model, inputting the training set into the model, and iteratively training the proportional parameters corresponding to each feature in the model to minimize the error between the load allocation ratio evaluation value output by the model and the target output value; and inputting the load allocation feature corresponding to each first pump room load allocation ratio into the trained model to obtain the corresponding load allocation ratio evaluation value.

[0018] In a preferred embodiment, the step of mapping and smoothing the load allocation ratio display diagram based on each first pump room load allocation ratio and the corresponding load allocation ratio evaluation value specifically involves: forming mapping points between each first pump room load allocation ratio and the corresponding load allocation ratio evaluation value, and drawing all mapping points in a three-dimensional coordinate system according to the coordinates of the main pump load allocation ratio and the reserve capacity ratio; interpolating the evaluation values ​​of the mapping points according to the coordinate order of the main pump load allocation ratio and the reserve capacity ratio to form a continuous evaluation value surface, thereby obtaining a smooth and continuous load allocation ratio display diagram.

[0019] In a preferred embodiment, the step of analyzing and selecting the optimal inter-pump load distribution ratio control pump based on demand data in the load distribution ratio display chart specifically involves: performing contour analysis on the smooth and continuous load distribution ratio display chart along the directions of the main pump load distribution ratio and the reserve capacity ratio to identify peak regions of the evaluation value within the local continuous area; applying the inter-pump load distribution ratio of the peak regions one by one to the historical load curve to simulate the pump response based on the total load demand and load fluctuation range of the actual system, and statistically analyzing the main pump output pressure, flow rate, and speed data; and using the main pump selection method to select the optimal inter-pump load distribution ratio.

[0020] A machine learning-based oil pump operating condition response control optimization system, characterized by comprising an initial inter-pump load allocation ratio module, a full-factor grid module, a feature extraction module, a load allocation ratio evaluation module, a graph quantization module, and a screening module; the initial inter-pump load allocation ratio module is used to acquire historical data of the multi-pump parallel system under the target operating condition, analyze the impact of load changes on the output stability of each oil pump under the target operating condition, and set the initial inter-pump load allocation ratio; the full-factor grid module is used to perform Cartesian products on the initial inter-pump load allocation ratio based on the full-factor grid, dividing it into several equal parts, to generate several first inter-pump load allocation ratios. For example, the feature extraction module is used to apply the load distribution ratio of the first pump room to the oil pump operating condition response control one by one, and to extract features during the application process to obtain load distribution features; based on the load distribution features, a load distribution ratio evaluation model is trained based on machine learning, and the corresponding load distribution ratio evaluation value is output; the load distribution ratio evaluation module is used to map and smooth the load distribution ratio display diagram based on each first pump room load distribution ratio and the corresponding load distribution ratio evaluation value; the graph quantization module and the filtering module are used to analyze the load distribution ratio display diagram according to the demand data and filter out the optimal pump room load distribution ratio to control the oil pump.

[0021] The technical effects and advantages of the machine learning-based oil pump operating condition response control optimization method and system of this invention are as follows:

[0022] 1. This invention acquires historical data of a multi-pump parallel system under target operating conditions, combines this data with the load distribution characteristics between pumps, and utilizes machine learning to establish a load distribution ratio evaluation model, thereby achieving intelligent optimization control of the oil pump's operating condition response. Compared with traditional methods of setting fixed distribution ratios or empirical parameters, this invention can dynamically adjust the main pump load distribution ratio and the reserve capacity ratio according to the actual load changes of the system, thus ensuring the stability of system pressure, flow rate, and speed under multiple operating conditions. This effectively alleviates the response lag problem under high load surges and low load / light load conditions, improves the output coordination between pumps and the dynamic performance of the overall system, and reduces problems such as excessive temperature rise, increased energy consumption, and accelerated mechanical wear caused by prolonged high-load operation of pumps.

[0023] 2. This invention maps the inter-pump load distribution ratio to evaluation values ​​and smoothly generates a load distribution ratio display chart. It then combines historical load curves to dynamically simulate local peak areas and select the main pump, thus achieving the optimal inter-pump load distribution ratio. This method not only theoretically yields the optimal combination but has also been verified in actual system operation to adapt to load fluctuations, ensuring the stability and response accuracy of each pump's output. Simultaneously, it considers energy efficiency optimization and equipment lifespan control, significantly improving the intelligent control level and engineering applicability of multi-pump parallel systems. Attached Figure Description

[0024] Figure 1 This is a flowchart illustrating a machine learning-based oil pump operating condition response control optimization method according to the present invention.

[0025] Figure 2 This is a schematic diagram of the structure of an oil pump operating condition response control optimization system based on machine learning according to the present invention. Detailed Implementation

[0026] 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0027] Example 1, Figure 1 This invention presents a machine learning-based method for optimizing oil pump operating condition response control, comprising the following steps:

[0028] S1, acquire historical data of the multi-pump parallel system under target operating conditions, analyze the impact of load changes on the output stability of each oil pump under target operating conditions, and set the initial inter-pump load distribution ratio;

[0029] In this embodiment, the acquisition of historical data of the multi-pump parallel system under the target operating condition, analysis of the impact of load changes on the output stability of each oil pump under the target operating condition, and setting of the initial inter-pump load distribution ratio are specifically as follows:

[0030] The setting of the initial pump load distribution ratio includes setting the main pump, auxiliary pump, and setting the initial main pump load distribution ratio and the initial standby capacity reserve ratio.

[0031] The auxiliary pump is not a single pump, but rather the sum of all pumps except the main pump;

[0032] After determining the allocation ratio of a main pump, the next cycle begins. The main pump from the previous cycle is removed, and the auxiliary pump from the previous cycle is set as 1. The main pump, auxiliary pump, and the initial main pump load allocation ratio and the initial standby capacity reserve ratio are set again. The cycle continues until all pumps are allocated.

[0033] The total load of the auxiliary pump in the next cycle is the remaining load after the main pump shares the load and the reserve capacity is reserved for the load in the previous cycle.

[0034] The pump load allocation ratio includes a main pump load allocation ratio and a standby capacity reserve ratio.

[0035] The setting of the main pump and auxiliary pump is specifically as follows:

[0036] The total system load sequence is segmented according to the preset load change range to obtain several load intervals. Each interval contains the oil pump output pressure, flow rate and speed data for the corresponding time period.

[0037] The maximum, minimum, and average output pressure of the oil pump in each load range are statistically analyzed to determine the pressure fluctuation range.

[0038] The maximum and minimum deviations of the oil pump speed and the average value were statistically analyzed within each load range to determine the speed stability.

[0039] The maximum, minimum, and average output flow rates of the oil pump in each load range are statistically analyzed to determine the magnitude of flow rate fluctuations.

[0040] Within each load range, the pressure fluctuation range, flow fluctuation range, and speed stability range of each oil pump are compared. The oil pump with the largest number of pumps and the most stable performance is determined as the main pump, and the remaining oil pumps are combined as auxiliary pump groups.

[0041] The initial main pump load allocation ratio and the initial reserve capacity ratio are set as follows:

[0042] The initial load distribution ratio of the main pump is set to enable the main pump to carry as much flow as possible in the preset high stability range, while ensuring that the total flow demand of the system can be met in all load ranges, thus obtaining the initial main pump load distribution ratio.

[0043] Based on the historical total system load demand curve, analyze the instantaneous load changes of the system during sudden load increases and instantaneous depreciation of the main pump, and set the initial standby capacity reserve ratio.

[0044] In this embodiment, the main pump refers to the oil pump that undertakes the core output task in a multi-pump parallel system. Its load distribution ratio directly affects the stability of system pressure and flow. The main pump is usually selected as the oil pump with the smallest output pressure and flow fluctuations in most load ranges to ensure that the system can respond quickly under high demand or sudden load conditions, while reducing the risk of system instability caused by the pump's own output fluctuations.

[0045] In this embodiment, the auxiliary pump group refers to the sum of all oil pumps other than the main pump, serving as an auxiliary output unit of the system. The function of the auxiliary pump group is to share the load not covered by the main pump and handle instantaneous load changes in the system. It provides supplementary flow and pressure when the main pump experiences short-term efficiency degradation or a sudden increase in high load, ensuring the continuous oil supply capability and operational response stability of the entire multi-pump parallel system. The auxiliary pump group is treated as a single unit in the overall control strategy, facilitating the gradual distribution of system load when cyclically setting the main pump.

[0046] In this embodiment, the main pump load distribution ratio refers to the proportion of flow or pressure borne by the main pump in the total system load, and it is the core variable for controlling the operating response of the oil pump. The setting of this ratio must take into account both the full utilization of the high-stability load range and the overall system's needs in each load range. Too high a ratio may cause mechanical wear and increased energy consumption due to long-term high-load operation of the main pump, while too low a ratio may lead to lag in response when the load suddenly increases. Therefore, it is a key control variable for the inter-pump load distribution strategy.

[0047] In this embodiment, the reserve capacity ratio refers to the adjustable output ratio that is reserved in advance to cope with sudden increases in system load or instantaneous degradation of the main pump. This ratio is not a passive loss, but a controllable quantity determined through system design and inter-pump load planning. It is used to respond quickly under high load change scenarios, ensure the continuity and stability of system oil supply, and prevent the main pump and auxiliary pump group from operating under overload in a short period of time.

[0048] It should be noted that during the cyclical setting of the main and auxiliary pumps, the load of the main pump determined in the previous cycle needs to be removed in each cycle, and the remaining pumps are treated as a new auxiliary pump group. This ensures that every pump has a chance to be selected as the main pump, thereby achieving comprehensive coverage and fairness in system load distribution. This cyclical method ensures that the performance characteristics of each pump can be quantitatively evaluated under different load conditions, providing complete data for subsequent load distribution optimization.

[0049] It should be noted that when statistically analyzing the output pressure, flow rate, and speed data of the oil pump within the load range, it is not enough to simply obtain the average value or fluctuation range. It is also necessary to consider the time-series characteristics of the data within each range, observing the pump's instantaneous response capability under sudden load changes, as well as the continuity of fluctuations. This allows for a comprehensive evaluation of the oil pump's stability under all operating conditions, providing a scientific basis for the selection of the main pump.

[0050] It should be noted that the initial main pump load allocation ratio and standby capacity reserve ratio are set not only based on historical data, but also need to consider the distribution characteristics and fluctuation trends of the system's total load under different operating conditions, so as to ensure that the system can maintain stable output under various load scenarios. The scientific setting of the initial values ​​directly affects the effectiveness of subsequent pump load combination generation and machine learning training.

[0051] S2, based on the full factor grid, the initial pump load allocation ratio is divided into several equal parts by Cartesian product to generate several first pump load allocation ratios.

[0052] In this embodiment, the initial inter-pump load allocation ratio is generated by performing a Cartesian product on several equally divided points based on the full-factor grid, specifically as follows:

[0053] Centered on the initial main pump load distribution ratio and the initial standby capacity reserve ratio, the range of change of the main pump load distribution ratio and the range of change of the standby capacity reserve ratio are preset according to percentages respectively.

[0054] The range of changes in the main pump load distribution ratio is divided into several equal parts, and the range of changes in the reserve capacity ratio is divided into several equal parts, resulting in the equal parts sequence of the main pump load distribution ratio and the equal parts sequence of the reserve capacity ratio.

[0055] The main pump load distribution ratio and the reserve capacity ratio are combined by Cartesian product to obtain all combinations of main pump load distribution ratio and reserve capacity ratio. Each combination forms a first inter-pump load distribution ratio.

[0056] In this embodiment, the initial main pump load allocation ratio refers to the basic output ratio allocated to the main pump within the total system load after historical data analysis and main pump selection. This ratio serves as the central point for generating load allocation combinations, determining the possible load range the main pump can bear under different alternative operating conditions. By using this ratio as the center, the system response of the main pump under high or low load conditions can be explored while ensuring high stable output.

[0057] In this embodiment, the initial reserve capacity ratio refers to a certain proportion of output capacity reserved by the system beyond normal load distribution to cope with sudden load surges or main pump efficiency degradation. This ratio serves as a reference center point when generating inter-pump load combinations, allowing for the exploration of the impact of reserve capacity on system stability and responsiveness under different configurations, thereby achieving a more comprehensive load combination analysis.

[0058] In this embodiment, the equal division sequence refers to dividing the range of variation of the main pump load allocation ratio and the reserve capacity ratio into several equal value points, forming a series of continuous candidate ratios. The equal division sequence can cover the fluctuations above and below the center value, ensuring that when generating load allocation combinations, both the high and low limits of the main pump's carrying capacity and the system performance corresponding to different reserve amounts of reserve capacity are considered.

[0059] In this embodiment, the Cartesian product combination refers to combining the main pump load allocation ratio sequence and the reserve capacity ratio sequence according to all possible pairings, with each pair forming a candidate inter-pump load allocation ratio. This method can systematically cover all possible configurations, providing sufficient data samples for subsequent oil pump operating condition response evaluation and machine learning training.

[0060] It should be noted that when generating the first inter-pump load distribution ratio, each combination is not merely a simple pairing of two values, but rather represents a complete load distribution strategy for a given operating condition. The configuration of the main pump and standby capacity under this combination jointly affects the system's output stability under different load variations. Through systematic combination generation, extreme load conditions or potentially high-risk configurations can be avoided, ensuring the comprehensiveness of the analysis.

[0061] It should be noted that the selection of the initial center value has a critical impact on the representativeness of the combined results. If the center value deviates from the historical load characteristics of the system, the generated combination may deviate from the feasible range of actual operating conditions. Therefore, when determining the initial main pump load allocation ratio and the reserve capacity ratio...

[0062] It should be noted that the above combination of equally divided sequences and Cartesian product not only generates all candidate proportions, but can also be used in practical applications to filter combinations that clearly exceed the load capacity or do not meet the control specifications, in conjunction with system response constraints.

[0063] S3, apply the first pump inter-pump load distribution ratio to the oil pump operating condition response control one by one, and extract features during the application process to obtain load distribution features;

[0064] In this embodiment, the first inter-pump load distribution ratio is applied one by one to the oil pump operating condition response control, and feature extraction is performed during the application process to obtain load distribution features, specifically as follows:

[0065] Characteristic quantification is performed to identify excessively high and low loads on the main pump.

[0066] Record the average load percentage of the main pump during the entire operating condition response process, the fluctuation range of the main pump output flow and pressure, and the deviation of the main pump output pressure and flow from the system demand during the high load change phase to calculate the response lag time;

[0067] By statistically analyzing the recorded data over time series, the maximum value, average value, and standard deviation are obtained, thus revealing the load distribution characteristics.

[0068] In this embodiment, the quantification of the main pump's overload and underload characteristics refers to the indexing of the main pump's operating status under different load distribution ratios. By measuring the main pump's output flow, pressure, and the degree of matching between the system response and demand, the performance of the main pump under high or low load conditions is transformed into characteristic data that can be used for evaluation, and then used for subsequent load distribution ratio evaluation.

[0069] In this embodiment, the average load percentage refers to the average ratio of the load actually borne by the main pump to its total rated load during the entire operating condition response process. This indicator reflects the overall load level of the main pump under various load conditions and is an important reference for measuring the rationality of load distribution.

[0070] In this embodiment, the output flow and pressure fluctuation range refers to the fluctuation of the main pump's output flow and pressure relative to the average value during the response process. This indicator can quantify the dynamic stability of the main pump under load changes and sudden changes in system demand.

[0071] In this embodiment, the response lag time refers to the time interval required for the pressure and flow rate output by the main pump to meet the system requirements when the system experiences a sudden high load change or a rapid action phase. This indicator can quantify the main pump's response speed to sudden load changes in the system.

[0072] It should be noted that the above-mentioned load distribution characteristics are not only used to describe the instantaneous operating state of the main pump, but can also be used as input features for machine learning models. By training with historical and simulated data, the rationality evaluation of the main pump load distribution can be obtained, thereby evaluating the advantages and disadvantages of different load ratio combinations across the entire operating range.

[0073] It should be noted that the method for collecting and quantifying load distribution characteristics is explained in detail because existing technologies have problems with pressure fluctuations and response lags that cannot be ignored in actual oil pump systems. This step provides highly operable and quantifiable feature definitions, which gives the subsequent evaluation model training a clear input basis.

[0074] S4. Based on the load distribution characteristics, train a load distribution ratio evaluation model using machine learning, and output the corresponding load distribution ratio evaluation value.

[0075] In this embodiment, the step of training a load allocation ratio evaluation model based on machine learning according to load allocation characteristics and outputting the corresponding load allocation ratio evaluation value specifically involves:

[0076] The load distribution feature vector corresponding to each first pump room load distribution ratio is used as the input sample, and the corresponding system response performance index is used as the model target output value.

[0077] All input samples are divided into training and validation sets proportionally for model training and performance validation.

[0078] Select a supervised learning model, input the training set into the model and iteratively train the proportional parameters corresponding to each feature in the model to minimize the error between the load distribution ratio evaluation value output by the model and the target output value.

[0079] Input the load allocation features corresponding to each first pump room load allocation ratio into the trained model to obtain the corresponding load allocation ratio evaluation value.

[0080] In this embodiment, the input sample refers to the load distribution feature vector corresponding to each first pump load distribution ratio as the input of the machine learning model. These feature vectors include indicators such as the main pump load percentage, output flow and pressure fluctuation amplitude, and response lag time, which are used to reflect the operating performance of the main pump under a specific load combination.

[0081] In this embodiment, the target output value of the model refers to the overall operating performance of the oil pump system quantified by system response performance indicators, such as the degree of satisfaction of the total system flow, pressure stability, and response speed. These indicators serve as the target output of the supervised learning model to guide the model in learning the advantages and disadvantages of different feature combinations.

[0082] In this embodiment, the supervised learning model refers to an algorithm model with trainable parameters, such as a regression model or a neural network, which maps the input load distribution features to the output load distribution ratio evaluation value. Through iterative training, the model parameters are continuously adjusted so that the model can accurately predict the system performance evaluation corresponding to the load ratio combination.

[0083] It should be noted that by mapping the input feature vector to the system response performance index, the model can capture the impact of changes in the main pump load ratio on the overall system response, thereby generating a single load distribution ratio evaluation value. This method transforms the abstract load ratio optimization problem into a quantifiable supervised learning task, improving the engineering operability of the optimization process.

[0084] It should be noted that the impact of load combination on system performance in multi-pump systems is complex and cannot be accurately assessed by experience alone. By clearly defining the division of training and validation sets and conducting supervised learning iterative training, we can ensure that the load allocation ratio evaluation model has reliable predictive capabilities and provide a quantitative basis for subsequent selection of the optimal load allocation ratio.

[0085] In this embodiment, the system response performance index refers to a set of indicators that quantify the actual operating performance of a multi-pump parallel system under a given load combination. These indicators measure the pump set's ability to meet load demands and its operational stability. The indicators include, but are not limited to, whether the main pump's output flow meets the system's total demand, the fluctuation range of the main pump's output pressure under load changes, the main pump's response lag time under high load abrupt changes, and the overall pressure coordination and oil circuit stability of the system. These indicators comprehensively reflect the operating effect of the oil pump system under different load distributions.

[0086] S5. Based on the load allocation ratio of each first pump room and the corresponding load allocation ratio evaluation value, a load allocation ratio display diagram is obtained by mapping and smoothing.

[0087] In this embodiment, the process of mapping and smoothing the load allocation ratio display diagram based on the load allocation ratio of each first pump room and the corresponding load allocation ratio evaluation value specifically involves:

[0088] Each first pump room load distribution ratio is mapped to the corresponding load distribution ratio evaluation value, and all mapping points are plotted in the three-dimensional coordinate system according to the main pump load distribution ratio and the standby capacity reserve ratio coordinates.

[0089] According to the coordinate order of the main pump load distribution ratio and the reserve capacity ratio, the evaluation values ​​of the mapping points are interpolated to form a continuous evaluation value surface, resulting in a smooth and continuous load distribution ratio display diagram.

[0090] In the load allocation ratio display diagram of this embodiment, the load allocation ratio of each group of first pump rooms and the corresponding load allocation ratio evaluation value are visualized in three-dimensional space. The horizontal and vertical axes represent the main pump load allocation ratio and the reserve ratio of standby capacity, respectively, and the depth or color intensity represents the magnitude of the load allocation ratio evaluation value, so as to intuitively show the performance of the system under different combinations.

[0091] In this embodiment, the mapping point refers to a discrete coordinate point in three-dimensional space that maps each group of pump load distribution ratios and their corresponding evaluation values. Each coordinate point fully reflects the quantitative effect of the main pump allocation and standby capacity reservation ratio on the system performance under this combination. These discrete points form the preliminary spatial distribution of the evaluation values, providing basic data for the subsequent generation of smooth surfaces.

[0092] It should be noted that interpolating discrete mapping points to form a continuous surface can eliminate the discreteness of single-point measurement or simulation, so that the load distribution ratio display diagram not only reflects the performance of existing combinations, but also predicts the performance trend of adjacent unmeasured combinations, thus providing a more complete optimization reference.

[0093] It should be noted that existing technologies in multi-pump load optimization typically rely only on discrete experimental or simulation results, lacking continuous visualization and making it impossible to intuitively compare the advantages and disadvantages of different combinations. This embodiment, through interpolation and three-dimensional visualization, not only displays the distribution of evaluation values ​​but also allows engineers to directly select the optimal load allocation ratio based on trends, achieving more efficient operating condition optimization.

[0094] S6 analyzes and selects the optimal pump load distribution ratio from the load distribution ratio display chart based on demand data to control the oil pump.

[0095] In this embodiment, the step of analyzing and selecting the optimal inter-pump load distribution ratio control oil pump based on the demand data load distribution ratio display chart specifically involves:

[0096] Contour analysis was performed on the smooth and continuous load distribution ratio display diagram along the direction of the main pump load distribution ratio and the reserve capacity ratio to identify the peak region of the evaluation value in the local continuous region.

[0097] Based on the total load demand and load fluctuation range of the actual system, the pump load distribution ratio in the peak area is applied one by one to the historical load curve to simulate the oil pump response, and the output pressure, flow rate and speed data of the main pump are statistically analyzed.

[0098] The method of selecting the main pump is used to select the optimal load distribution ratio between pumps.

[0099] In the contour analysis of this embodiment, contour lines are drawn along the two directions of the main pump load distribution ratio and the reserve capacity ratio on the load distribution ratio display map. Combinations with the same load distribution ratio evaluation value are connected into curves, which makes it easier to identify local peak areas with higher evaluation values ​​in continuous areas and provides a reference for selecting the optimal pump load distribution ratio.

[0100] In this embodiment, the peak region refers to the continuous region formed in the contour analysis, where the load allocation ratio evaluation value reaches a local maximum level. This means that within this combination range, the main pump load allocation ratio and the reserve capacity ratio enable the system to exhibit better pressure, flow, and speed response performance under historical load conditions. These regions contain multiple feasible load allocation ratio combinations.

[0101] It should be noted that applying the peak region pump load distribution ratio to the historical load curve to simulate the pump response not only verifies its feasibility under actual working conditions, but also further screens the combination with the best stability under different load fluctuation conditions, ensuring that the selected ratio has reliability and applicability under dynamic working conditions.

[0102] It should be noted that the method of selecting the main pump is used to finally determine the optimal load distribution ratio between pumps. The main pump load ratio directly affects the system response speed, stability and energy consumption performance. Existing technologies do not perform dynamic verification and main pump selection for local peak regions on continuous surfaces. This embodiment supplements the complete technical process of load distribution ratio optimization by combining historical load curve simulation and main pump selection method, thereby improving the system control accuracy and reliability.

[0103] Example 2, Figure 2 This invention presents a machine learning-based oil pump operating condition response control optimization system, characterized by comprising an initial inter-pump load allocation ratio module, a full-factor grid module, a feature extraction module, a load allocation ratio evaluation module, a graph quantization module, and a screening module. The initial inter-pump load allocation ratio module is used to acquire historical data of the multi-pump parallel system under target operating conditions, analyze the impact of load changes on the output stability of each oil pump under the target operating conditions, and set the initial inter-pump load allocation ratio. The full-factor grid module is used to generate several first inter-pump loads by performing Cartesian products on the initial inter-pump load allocation ratio at several equal division points based on the full-factor grid. The system comprises the following modules: a load distribution ratio module and a feature extraction module. The first pump inter-station load distribution ratio is applied to the oil pump operating condition response control one by one, and feature extraction is performed during the application process to obtain load distribution features. Based on the load distribution features, a load distribution ratio evaluation model is trained using machine learning, and the corresponding load distribution ratio evaluation value is output. The load distribution ratio evaluation module is used to map and smooth the load distribution ratio display diagram based on each first pump inter-station load distribution ratio and the corresponding load distribution ratio evaluation value to obtain the load distribution ratio display diagram. The graph quantization module and the filtering module are used to analyze the load distribution ratio display diagram based on the demand data and filter out the optimal pump inter-station load distribution ratio to control the oil pump.

[0104] In the embodiments provided by this invention, it should be understood that the disclosed system can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0105] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0106] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the invention. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0107] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0108] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0109] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above.

[0110] In the embodiments provided in this disclosure, it should be understood that the disclosed systems and methods can also be implemented in other ways. The system embodiments described above are merely illustrative; for example, the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those marked in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in a block diagram and / or flowchart, and combinations of blocks in block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0111] It should be noted that, in this disclosure, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element limited by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0112] The above-described embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should all be included within the protection scope of the present invention.

Claims

1. A machine learning-based method for optimizing oil pump operating condition response control, characterized in that, Includes the following steps: Acquire historical data of the multi-pump parallel system under target operating conditions, analyze the impact of load changes on the output stability of each oil pump under target operating conditions, and set the initial load distribution ratio between pumps. Based on the full factor grid, the initial pump load allocation ratio is divided into several equal parts by Cartesian products to generate several first pump load allocation ratios. The first pump inter-pump load distribution ratio is applied one by one to the oil pump operating condition response control, and feature extraction is performed during the application process to obtain load distribution features. Based on the load distribution characteristics, a load distribution ratio evaluation model is trained using machine learning, and the corresponding load distribution ratio evaluation value is output. The load allocation ratio display diagram is obtained by mapping and smoothing the load allocation ratio of each first pump room and the corresponding load allocation ratio evaluation value. Based on the demand data, the load distribution ratio display diagram is analyzed and the optimal pump load distribution ratio is selected to control the oil pump.

2. The machine learning-based oil pump operating condition response control optimization method according to claim 1, characterized in that, The process involves acquiring historical data of the multi-pump parallel system under target operating conditions, analyzing the impact of load changes on the output stability of each pump under the target operating conditions, and setting an initial inter-pump load distribution ratio. Specifically: The setting of the initial pump load distribution ratio includes setting the main pump, auxiliary pump, and setting the initial main pump load distribution ratio and the initial standby capacity reserve ratio. The pump load allocation ratio includes a main pump load allocation ratio and a standby capacity reserve ratio. The auxiliary pump is not a single pump, but rather the sum of all pumps except the main pump; After determining the allocation ratio of a main pump, the next cycle begins. The main pump from the previous cycle is removed, and the auxiliary pump from the previous cycle is set as 1. The main pump, auxiliary pump, and the initial main pump load allocation ratio and the initial standby capacity reserve ratio are set again. The cycle continues until all pumps are allocated. The total load of the auxiliary pump in the next cycle is the remaining load after the main pump shares the load and the reserve capacity is reserved for the load in the previous cycle.

3. The machine learning-based oil pump operating condition response control optimization method according to claim 2, characterized in that, The setting of the main pump and auxiliary pump is specifically as follows: The total system load sequence is segmented according to the preset load change range to obtain several load intervals. Each interval contains the oil pump output pressure, flow rate and speed data for the corresponding time period. The maximum, minimum, and average output pressure of the oil pump in each load range are statistically analyzed to determine the pressure fluctuation range. The maximum and minimum deviations of the oil pump speed and the average value were statistically analyzed within each load range to determine the speed stability. The maximum, minimum, and average output flow rates of the oil pump in each load range are statistically analyzed to determine the magnitude of flow rate fluctuations. Within each load range, the pressure fluctuation range, flow fluctuation range, and speed stability range of each oil pump are compared. The oil pump with the largest number of pumps and the most stable performance is determined as the main pump, and the remaining oil pumps are combined as auxiliary pump groups.

4. The machine learning-based oil pump operating condition response control optimization method according to claim 3, characterized in that, The initial main pump load allocation ratio and the initial reserve capacity ratio are set as follows: The initial load distribution ratio of the main pump is set to enable the main pump to carry as much flow as possible in the preset high stability range, while ensuring that the total flow demand of the system can be met in all load ranges, thus obtaining the initial main pump load distribution ratio. Based on the historical total system load demand curve, analyze the instantaneous load changes of the system during sudden load increases and instantaneous depreciation of the main pump, and set the initial standby capacity reserve ratio.

5. The machine learning-based oil pump operating condition response control optimization method according to claim 4, characterized in that, The method of generating several first pump load allocation ratios by performing Cartesian products on the initial pump load allocation ratios based on the full-factor grid at several equally divided points is as follows: Centered on the initial main pump load distribution ratio and the initial standby capacity reserve ratio, the range of change of the main pump load distribution ratio and the range of change of the standby capacity reserve ratio are preset according to percentages respectively. The range of changes in the main pump load distribution ratio is divided into several equal parts, and the range of changes in the reserve capacity ratio is divided into several equal parts, resulting in the equal parts sequence of the main pump load distribution ratio and the equal parts sequence of the reserve capacity ratio. The main pump load distribution ratio and the reserve capacity ratio are combined by Cartesian product to obtain all combinations of main pump load distribution ratio and reserve capacity ratio. Each combination forms a first inter-pump load distribution ratio.

6. The machine learning-based oil pump operating condition response control optimization method according to claim 5, characterized in that, The first method involves applying the load distribution ratio between pumps one by one to the oil pump operating condition response control, and extracting features during the application process to obtain load distribution features, specifically: Characteristic quantification is performed to identify excessively high and low loads on the main pump. Record the average load percentage of the main pump during the entire operating condition response process, the fluctuation range of the main pump output flow and pressure, and the deviation of the main pump output pressure and flow from the system demand during the high load change phase to calculate the response lag time; By statistically analyzing the recorded data over time series, the maximum value, average value, and standard deviation are obtained, thus revealing the load distribution characteristics.

7. The machine learning-based oil pump operating condition response control optimization method according to claim 6, characterized in that, The process of training a load allocation ratio evaluation model based on machine learning according to load allocation characteristics and outputting the corresponding load allocation ratio evaluation value is as follows: The load distribution feature vector corresponding to each first pump room load distribution ratio is used as the input sample, and the corresponding system response performance index is used as the model target output value. All input samples are divided into training and validation sets proportionally for model training and performance validation. Select a supervised learning model, input the training set into the model and iteratively train the proportional parameters corresponding to each feature in the model to minimize the error between the load distribution ratio evaluation value output by the model and the target output value. Input the load allocation features corresponding to each first pump room load allocation ratio into the trained model to obtain the corresponding load allocation ratio evaluation value.

8. The machine learning-based oil pump operating condition response control optimization method according to claim 7, characterized in that, The process of mapping and smoothing the load allocation ratio based on the load allocation ratio of each first pump room and its corresponding load allocation ratio evaluation value to obtain the load allocation ratio display diagram is as follows: Each first pump room load distribution ratio is mapped to the corresponding load distribution ratio evaluation value, and all mapping points are plotted in the three-dimensional coordinate system according to the main pump load distribution ratio and the standby capacity reserve ratio coordinates. According to the coordinate order of the main pump load distribution ratio and the reserve capacity ratio, the evaluation values ​​of the mapping points are interpolated to form a continuous evaluation value surface, resulting in a smooth and continuous load distribution ratio display diagram.

9. The machine learning-based oil pump operating condition response control optimization method according to claim 8, characterized in that, The process of analyzing and selecting the optimal inter-pump load distribution ratio based on demand data to control the oil pump involves: Contour analysis was performed on the smooth and continuous load distribution ratio display diagram along the direction of the main pump load distribution ratio and the reserve capacity ratio to identify the peak region of the evaluation value in the local continuous region. Based on the total load demand and load fluctuation range of the actual system, the pump load distribution ratio in the peak area is applied one by one to the historical load curve to simulate the oil pump response, and the output pressure, flow rate and speed data of the main pump are statistically analyzed. The method of selecting the main pump is used to select the optimal load distribution ratio between pumps.

10. A system using the machine learning-based oil pump operating condition response control optimization method as described in any one of claims 1-9, characterized in that, It includes an initial pump load allocation ratio module, a full-factor grid module, a feature extraction module, a load allocation ratio evaluation module, a graph quantization module, and a screening module; The initial inter-pump load distribution ratio module is used to acquire historical data of the multi-pump parallel system under target operating conditions, analyze the impact of load changes on the output stability of each oil pump under target operating conditions, and set the initial inter-pump load distribution ratio. The full-factor grid module is used to perform Cartesian products on the initial pump load allocation ratios based on the full-factor grid, dividing them into several equal parts, to generate several first pump load allocation ratios. The feature extraction module is used to apply the first inter-pump load distribution ratio to the oil pump operating condition response control one by one, and to extract features during the application process to obtain load distribution features. Based on the load distribution characteristics, a load distribution ratio evaluation model is trained using machine learning, and the corresponding load distribution ratio evaluation value is output. The load allocation ratio evaluation module is used to map and smooth the load allocation ratio display diagram based on the load allocation ratio of each first pump room and the corresponding load allocation ratio evaluation value. The graph quantification module and the filtering module are used to analyze and filter the load distribution ratio display graph based on the demand data to select the optimal inter-pump load distribution ratio to control the oil pump.

Images