Pump set power distribution method and system, engine and operation machine

Abstract

The invention relates to the technical field of engines, in particular to a pump unit power distribution method and system, an engine and an operation machine. The static torque of a master pump set corresponding to the target engine and the quantitative torque corresponding to a quantitative pump set in the target engine are obtained; based on the static torque and the quantitative torque of the master pump set, the distributable torque corresponding to the variable pump set of the target engine is obtained; acquiring current speed data corresponding to each variable pump in the variable pump set; and determining a target torque corresponding to each variable pump based on the distributable torque and the current speed data corresponding to each variable pump. Meanwhile, accurate torque distribution is also beneficial to reducing abrasion and energy consumption of equipment, prolonging the service life of the equipment and reducing the operation cost. In the prior art, when a plurality of pump sets operate at a high load at the same time, an engine is prone to overtorque, speed reduction and flameout; when a single pump set operates in a high-load mode, the torque of an engine cannot be brought into full play; and torque distribution cannot be carried out on a plurality of pump sets.

Description

Technical Field

[0001] The present invention relates to the technical field of engines, and particularly to a method and system for power distribution of pump sets, an engine, and a working machine. Background Art

[0002] Currently, in the hydraulic systems equipped in many mechanical devices such as large engineering vehicles and heavy industrial production equipment, there are generally some problems that need to be solved urgently. When multiple pump sets are operating at high load simultaneously, the engine is extremely likely to experience over-torque speed drop or even flameout. Taking a large excavator as an example, during compound operations such as excavation and loading, the pump sets corresponding to multiple working devices such as the boom, arm, and bucket will operate at high load simultaneously. At this time, the torque borne by the engine increases significantly instantaneously, exceeding its rated torque range, resulting in a sharp drop in the engine speed and ultimately possible flameout and shutdown. This not only seriously affects the normal operation progress of the equipment, but also the frequent flameout and restart will accelerate the wear of the engine, greatly shorten its service life, and increase the maintenance cost and operation risk of the equipment.

[0003] On the other hand, when a single pump set operates at high load, the engine torque cannot be fully utilized. In some equipment with multiple pump set configurations but relatively single working conditions, such as some special material conveying equipment, only one pump set may need to work at full capacity to meet the pressure and flow requirements of material conveying. However, due to the lack of an effective torque adjustment mechanism in the target engine, the torque output by the engine cannot accurately match the high load requirements of a single pump set, and a large amount of torque is wasted, resulting in low working efficiency of the engine, increased energy consumption, and unnecessary energy waste.

[0004] In addition, the existing target engines often cannot perform reasonable and effective torque distribution for multiple pump sets. Therefore, providing a method for reasonable and effective torque distribution for multiple pump sets has become an urgent problem to be solved. Summary of the Invention

[0005] In view of this, the present invention provides a method and system for power distribution of pump sets, an engine, and a working machine to solve the problem of how to perform reasonable and effective torque distribution for pump sets.

[0006] In the first aspect, the present invention provides a method for power distribution of pump sets, the method comprising:

[0007] Obtaining the total static torque of the total pump set corresponding to the target engine and the fixed torque corresponding to the fixed pump set in the target engine;

[0008] Based on the total static torque of the total pump set and the fixed torque, obtaining the distributable torque corresponding to the variable pump set of the target engine;

[0009] Obtaining the current speed data of each variable pump in the variable pump set;

[0010] Based on the distributable torque and the current speed data corresponding to each variable pump, determine the target torque corresponding to each variable pump.

[0011] The pump group power distribution method provided by the embodiment of the present application obtains the total pump group static torque corresponding to the target engine and the fixed torque corresponding to the fixed pump group in the target engine. Based on the total pump group static torque and the fixed torque, obtain the distributable torque corresponding to the variable pump group of the target engine, which ensures the accuracy of the distributable torque corresponding to the variable pump group of the obtained target engine. Based on this exact distributable torque value, subsequent scientific and reasonable torque distribution can be carried out according to the actual requirements and operating conditions of each variable pump. This avoids the unreasonable torque distribution phenomenon caused by unclear adjustable resources, such as excessive torque distribution in a certain variable pump group, exceeding its actual requirements, while other pump groups cannot work properly due to insufficient torque. Through precise distributable torque calculation, the torque resource configuration of the target engine is optimized, and the working efficiency and reliability of the entire hydraulic pump group are improved. Then, obtain the current speed data corresponding to each variable pump in the variable pump group, and based on the distributable torque and the current speed data corresponding to each variable pump, determine the target torque corresponding to each variable pump. Realize precise control based on the resources of the target engine and real-time working conditions. According to the working state of the variable pump group reflected by the speed data, reasonably distribute the distributable torque, so that each variable pump group can obtain the torque most suitable for its current working condition. This significantly improves the adaptability of the hydraulic pump group to complex working conditions, ensuring that the target engine can operate efficiently and stably in various situations. At the same time, precise torque distribution also helps to reduce equipment wear and energy consumption, extend the service life of the equipment, and reduce operating costs. Solve the problems in the prior art that when multiple pump groups operate at high load simultaneously, the engine is prone to over-torque, speed drop and flameout; when a single pump group operates at high load, the engine torque cannot be fully utilized; and torque distribution cannot be performed on multiple pump groups.

[0012] In an optional implementation manner, the current speed data is a current speed sequence composed of the current moment speed and the historical moment speeds corresponding to a preset number of historical moments before the current moment; based on the distributable torque and the current speed corresponding to each variable pump, determining the target torque corresponding to each variable pump includes:

[0013] Input the total pump group static torque, the distributable torque, and the current speed sequence corresponding to each variable pump into a preset torque distribution model;

[0014] The preset torque distribution model extracts features from the total pump group static torque, the distributable torque, and the current speed sequence corresponding to each variable pump, and outputs the target torque corresponding to each variable pump.

[0015] The pump group power distribution method provided by the embodiments of the present application inputs the total pump group static torque, the distributable torque, and the current speed sequences corresponding to each variable pump into a preset torque distribution model. The preset torque distribution model extracts features from the total pump group static torque, the distributable torque, and the current speed sequences corresponding to each variable pump, and outputs the target torque corresponding to each variable pump. The total pump group static torque reflects the basic power required for the stable operation of the entire pump group, the distributable torque clarifies the total amount of resources that can be flexibly allocated, and the current speed sequences of each variable pump reflect the real-time operating status of each pump group. The preset torque distribution model integrates and analyzes these multi-dimensional information, can comprehensively and accurately grasp the working conditions of the variable pump group in the target engine, and thus ensures the target torque corresponding to each variable pump output. Thereby, the energy is optimally configured among each pump group. For example, for some pump groups with higher efficiency, more torque can be appropriately allocated to give full play to their performance advantages; while for pump groups with lower efficiency, the torque allocation is reduced to avoid inefficient use of energy. By this way of optimizing energy distribution, the energy utilization efficiency of the entire hydraulic system is improved and the operating cost is reduced.

[0016] In an alternative embodiment, inputting the total pump group static torque, the distributable torque, and the current speed sequences corresponding to each variable pump into the preset torque distribution model includes:

[0017] Obtain the current pressure data and the current flow data corresponding to each variable pump;

[0018] Construct a spatial feature matrix based on the current pressure data and the current flow data;

[0019] Input the spatial feature matrix into a preset graph neural network, and construct a graph structure with each variable pump as a node and the relationship between each variable pump as an edge;

[0020] Based on the graph structure, output node embedding vectors;

[0021] Input the total pump group static torque, the distributable torque, the current speed sequences corresponding to each variable pump, and the node embedding vectors into the preset torque distribution model;

[0022] Correspondingly, the preset torque distribution model extracts features from the total pump group static torque, the distributable torque, the current speed sequences corresponding to each variable pump, and the node embedding vectors, and outputs the target torque corresponding to each variable pump.

[0023] The pump group power distribution method provided by the embodiment of the present application obtains the current pressure data and current flow data corresponding to each variable pump; constructs a spatial feature matrix based on the current pressure data and current flow data, and presents the state information of the variable pump group in a structured form. This matrix structure can clearly display the state relationship between variable pumps, providing a good data basis for subsequent graph neural network processing. Input the spatial feature matrix into a preset graph neural network, construct a graph structure with each variable pump as a node and the relationship between variable pumps as an edge, which can intuitively represent the physical connection and interaction between pump groups. In a hydraulic system, the pressure or flow change of one pump group may affect other connected pump groups, and the graph structure can accurately capture this coupling relationship. Process the graph structure through the graph neural network to deeply explore the complex non-linear relationship between pump groups. Based on the graph structure, output node embedding vectors. The node embedding vectors output by the graph neural network are a low-dimensional vector representation of each variable pump group and its relationship with other pump groups. These vectors integrate the pressure, flow information of the pump group itself and the coupling relationship information with other pump groups, and can more effectively reflect the state and role of the pump group in the entire target engine. Compared with the traditional analysis method based on the data of a single pump group, the node embedding vectors can better capture the global relationship between pump groups and provide richer and more accurate information for torque distribution. Input the total pump group static torque, distributable torque, the current speed sequence corresponding to each variable pump, and the node embedding vectors into a preset torque distribution model. The preset torque distribution model extracts features from the total pump group static torque, distributable torque, the current speed sequence corresponding to each variable pump, and the node embedding vectors, and outputs the target torque corresponding to each variable pump. The preset torque distribution model can comprehensively consider the self-state of the variable pump group, the mutual influence between variable pump groups, and the overall power resources of the target engine according to the total pump group static torque, distributable torque, the current speed sequence corresponding to each variable pump, and the node embedding vectors, so as to allocate torque more accurately. For example, when a certain variable pump group undergoes pressure or flow changes due to the influence of other pump groups, the preset torque distribution model can timely adjust the target torque of the variable pump group according to the coupling relationship reflected in the node embedding vectors to ensure the stable operation of the target engine.

[0024] In an alternative embodiment, the preset torque distribution model extracts features from the total pump group static torque, distributable torque, the current speed sequence corresponding to each variable pump, and the node embedding vectors, and outputs the target torque corresponding to each variable pump, including:

[0025] The preset torque distribution model extracts features from each current speed sequence based on convolutional kernels of different sizes, and outputs speed sequence feature vectors;

[0026] Fuse each speed sequence feature vector, the static torque of the main pump group, the distributable torque, and the node embedding vector to generate an initial fused feature;

[0027] Transform the initial fused feature to generate a transformed feature;

[0028] Fuse the initial fused feature and the transformed feature to generate a target feature;

[0029] Based on the target feature, output the target torque corresponding to each variable pump.

[0030] The pump group power distribution method provided by the embodiments of the present application. The preset torque distribution model extracts features from each current speed sequence based on convolution kernels of different sizes, and outputs speed sequence feature vectors. Convolution kernels of different sizes can capture features at different time scales in the current speed sequence. Smaller convolution kernels can focus on local detail changes in the speed sequence, such as short-term speed fluctuations, which is very important for promptly responding to rapid changes in the pump group speed. Larger convolution kernels, on the other hand, can extract features of the speed sequence from a more macroscopic perspective, such as long-term speed trends, which helps to grasp the overall change pattern of the pump group speed. Through this multi-scale feature extraction method, the characteristics of the speed sequence can be described more comprehensively, providing richer and more accurate speed information for subsequent torque distribution. Fuse the speed sequence feature vectors, total pump group static torque, distributable torque, and node embedding vectors to generate an initial fusion feature. It can make full use of information from different aspects. The speed sequence feature reflects the dynamic operating state of the pump group, the total pump group static torque and distributable torque represent the power resource status of the target engine, and the node embedding vector contains information about the mutual relationship between pump groups. By fusing this information, the preset torque distribution model can comprehensively consider various aspects of the target engine, avoiding the limitations of relying solely on a single information source for torque distribution, and thus being able to make more comprehensive and reasonable torque distribution decisions. Transform the initial fusion feature to generate a transformed feature, which can further increase the expressive ability of the feature. Through operations such as non-linear transformation, the initial fusion feature can be mapped to a feature space that is more conducive to the model making torque distribution decisions. Fuse the initial fusion feature and the transformed feature to generate a target feature. Thus, it can fully combine the intuitive information of the original feature and the enhanced expressive ability of the transformed feature. The initial fusion feature retains the basic information and direct relationship of the original data, while the transformed feature has undergone further processing and optimization, highlighting the features that have an important impact on torque distribution. By fusing the two, the preset torque distribution model can utilize the advantages of both aspects at the same time. It can not only have a comprehensive understanding of the overall state of the target engine based on the original information, but also more accurately capture the key information related to the target torque with the help of the transformed feature, thereby improving the accuracy and reliability of torque distribution. Based on the target feature, output the target torque corresponding to each variable pump, which can ensure that the torque distribution is more accurate and reasonable.

[0031] In an alternative embodiment, transforming the initial fusion feature to generate a transformed feature includes:

[0032] Perform downsampling operations of different scales on the initial fusion feature to obtain multiple sub-fusion features;

[0033] Based on the multi-hidden layer perceptrons corresponding to each sub-fusion feature, perform transformation processing on each sub-fusion feature to generate sub-transformed features corresponding to each sub-fusion feature;

[0034] Fuse each sub - transformation feature to generate a transformation feature.

[0035] The pump - group power distribution method provided by the embodiments of this application performs down - sampling operations on the initial fusion feature at different scales to obtain multiple sub - fusion features. The down - sampling operations at different scales enable the preset torque distribution model to examine the initial fusion feature from multiple granularities. This multi - scale feature extraction method comprehensively and meticulously depicts the initial fusion feature, providing a rich and diverse information basis for the subsequent processing of the preset torque distribution model. Based on the multi - layer perceptrons corresponding to each sub - fusion feature, perform transformation processing on each sub - fusion feature to generate sub - transformation features corresponding to each sub - fusion feature. Each sub - fusion feature has its unique feature pattern and information distribution, and the corresponding multi - layer perceptron can perform targeted transformations according to the characteristics of the sub - fusion feature. The sub - fusion features at different scales contain different levels of information, and the exclusive multi - layer perceptron can learn the specific relationship between the feature at this scale and the target (such as the target torque) by adjusting the weights and biases. Fusing each sub - transformation feature to generate a transformation feature can integrate the sub - transformation features that have been targeted - transformed at different scales, ensuring the comprehensiveness and integrity of the generated transformation feature.

[0036] In an alternative embodiment, the preset torque distribution model extracts features from the total pump - group static torque, the distributable torque, and the current speed sequences corresponding to each variable pump, and outputs the target torque corresponding to each variable pump, including:

[0037] The preset torque distribution model extracts features from the total pump - group static torque, the distributable torque, and the current speed sequences corresponding to each variable pump, and outputs the candidate torque corresponding to each variable pump;

[0038] Based on each candidate torque, determine the variable - pump state information corresponding to each variable pump;

[0039] According to the variable - pump state information corresponding to each variable pump, determine the engine - state information corresponding to the target engine;

[0040] Adjust each candidate torque according to the engine state, and output the target torque corresponding to each variable pump.

[0041] The pump group power distribution method provided by the embodiment of the present application is characterized by a preset torque distribution model for extracting features of the static torque of the total pump group, the distributable torque, and the current speed sequence corresponding to each variable pump, and outputting the candidate torque corresponding to each variable pump. Through feature extraction, the preset torque distribution model can deeply explore the potential laws behind these information and provide an accurate basis for subsequent torque distribution. Based on the extracted features, the candidate torque corresponding to each variable pump is output, and the initial reasonable distribution of torque is realized, so that the torque distribution is initially matched with the actual operation requirements of the variable pump group, and the inefficiency or equipment damage caused by blindly distributing torque is avoided. Based on each candidate torque, the variable pump state information corresponding to each variable pump is determined, and the working state of the variable pump group can be comprehensively evaluated, and whether the variable pump group has problems such as overload, inefficiency or unstable operation is timely discovered. According to the variable pump state information corresponding to each variable pump, the engine state information corresponding to the target engine is determined. According to the state information of each variable pump, the engine state information corresponding to the target engine is determined, and a close connection between the variable pump group and the target engine can be established. By monitoring the state information of the target engine, potential problems can be warned in advance. According to the engine state, each candidate torque is adjusted and the target torque corresponding to each variable pump is output. Adjusting the candidate torque according to the engine status can further optimize the torque distribution. If the engine load rate is too high, in order to avoid engine overload, the candidate torque of some variable pump groups can be appropriately reduced; if the engine speed stability is poor, the torque distribution of the variable pump group can be optimized to reduce sudden changes in load. Through this adjustment, the torque distribution is more in line with the actual working capacity of the engine, improving the operating efficiency and stability of the entire target engine.

[0042] In an optional implementation, obtaining the static torque of the total pump group corresponding to the target engine includes:

[0043] Get the current engine speed corresponding to the target engine;

[0044] Determine the rated torque of the target engine at the current engine speed according to the engine external characteristic curve corresponding to the target engine;

[0045] According to the engine torque absorption ratio corresponding to the target engine, the total pump group static torque corresponding to the target engine is calculated.

[0046] The pump group power distribution method provided by the embodiment of the present application obtains the current engine speed corresponding to the target engine. According to the engine external characteristic curve corresponding to the target engine, the rated torque of the target engine at the current engine speed is determined; the rated torque that the target engine can provide at the current engine speed can be accurately determined. This enables the operator or the controlled target engine to clearly understand the output capacity of the engine, providing an important basis for reasonably distributing torque and arranging work tasks. For example, if it is known that the rated torque of the engine at the current speed is low, it is necessary to avoid assigning too large a load to the target engine to prevent the target engine from being overloaded. According to the engine torque absorption ratio corresponding to the target engine, the total pump group static torque corresponding to the target engine is calculated. Calculating the total pump group static torque through the engine torque absorption ratio can fully consider the actual working requirements and operating characteristics of the target engine. Different hydraulic systems have different degrees of absorption of engine torque under different working conditions, and the torque absorption ratio can accurately reflect this difference. Therefore, calculating the total pump group static torque according to the torque absorption ratio can make the calculation result more in line with the actual situation of the target engine, providing more accurate basic data for subsequent torque distribution and target engine control. In addition, by considering the engine torque absorption ratio to calculate the total pump group static torque, torque distribution can be more precisely carried out according to the actual output capacity of the engine and the requirements of the target engine, avoiding problems such as target engine failures or low efficiency caused by unreasonable torque distribution, and improving the control accuracy and stability of the entire hydraulic system.

[0047] In a second aspect, the present invention provides a pump group power distribution system, the system includes:

[0048] A first acquisition module, configured to acquire the total pump group static torque corresponding to the target engine and the fixed torque corresponding to the fixed displacement pump group in the target engine;

[0049] A subtraction module, configured to obtain the distributable torque corresponding to the variable displacement pump group of the target engine based on the total pump group static torque and the fixed torque;

[0050] A second acquisition module, configured to acquire the current speed data corresponding to each variable displacement pump in the variable displacement pump group;

[0051] A determination module, configured to determine the target torque corresponding to each variable displacement pump based on the distributable torque and the current speed data corresponding to each variable displacement pump.

[0052] The pump group power distribution system provided by the embodiment of the present application obtains the total pump group static torque corresponding to the target engine and the fixed torque corresponding to the fixed displacement pump group in the target engine. Based on the total pump group static torque and the fixed torque, the distributable torque corresponding to the variable displacement pump group of the target engine is obtained, ensuring the accuracy of the distributable torque corresponding to the variable displacement pump group of the obtained target engine. Based on this exact distributable torque value, subsequent scientific and reasonable torque distribution can be carried out according to the actual requirements and operating states of each variable displacement pump. This avoids the unreasonable torque distribution phenomenon caused by unclear adjustable resources, such as excessive torque distribution in a certain variable displacement pump group, exceeding its actual requirements, while other pump groups cannot work properly due to insufficient torque. Through precise calculation of the distributable torque, the torque resource configuration of the target engine is optimized, and the working efficiency and reliability of the entire hydraulic pump group are improved. Then, the current speed data corresponding to each variable displacement pump in the variable displacement pump group is obtained, and based on the distributable torque and the current speed data corresponding to each variable displacement pump, the target torque corresponding to each variable displacement pump is determined. Precise control based on the resources of the target engine and real-time working conditions is achieved. According to the working state of the variable displacement pump group reflected by the speed data, the distributable torque is reasonably distributed, enabling each variable displacement pump group to obtain the torque most suitable for its current working condition. This significantly improves the adaptability of the hydraulic pump group to complex working conditions, ensuring that the target engine can operate efficiently and stably under various circumstances. At the same time, precise torque distribution also helps to reduce equipment wear and energy consumption, extend the service life of the equipment, and reduce operating costs. It solves the problems in the prior art that when multiple pump groups operate at high load simultaneously, the engine is prone to over-torque, speed drop, and flameout; when a single pump group operates at high load, the engine torque cannot be fully utilized; and torque distribution cannot be performed on multiple pump groups.

[0053] In a third aspect, the present invention provides an engine, including: an engine body and an electronic device, where the electronic device includes: a memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to execute the pump group power distribution method according to the first aspect or any corresponding implementation manner thereof.

[0054] In a fourth aspect, the present invention provides a work machine, and the work machine includes the engine according to the third aspect above.

[0055] In a fifth aspect, the present invention provides a computer program product, including computer instructions for causing a computer to execute the pump group power distribution method according to the first aspect or any corresponding implementation manner thereof. Description of the Drawings

[0056] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0057] Figure 1 is a left view of a work machine according to an embodiment of the present invention;

[0058] Figure 2 is a schematic flowchart of a pump group power distribution method according to an embodiment of the present invention;

[0059] Figure 3 is a schematic flowchart of another pump group power distribution method according to an embodiment of the present invention;

[0060] Figure 4 is a schematic flowchart of yet another pump group power distribution method according to an embodiment of the present invention;

[0061] Figure 5 is a schematic flowchart of still another pump group power distribution method according to an embodiment of the present invention;

[0062] Figure 6 is a structural block diagram of a pump group power distribution system according to an embodiment of the present invention;

[0063] Figure 7 is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. Specific Embodiments

[0064] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope of protection of the present invention.

[0065] At present, there are some common problems that need to be solved in the hydraulic systems of many mechanical equipment such as large engineering vehicles and heavy industrial production equipment. When multiple pump groups are in high-load operation at the same time, the engine is very likely to over-torque and even stall. Taking operating machinery as an example, when performing complex actions such as luffing, rotation, walking, and operation, the driving pump groups corresponding to multiple working devices such as the luffing winch, slewing motor, and walking motor will operate at high load at the same time. At this time, the torque borne by the engine increases instantly and exceeds its rated torque range, causing the engine speed to drop sharply, and may eventually stall. This not only seriously affects the normal operation progress of the equipment, but frequent shutdown and restart will also accelerate the wear of the engine, greatly shorten its service life, and increase the maintenance cost and operation risk of the equipment.

[0066] On the other hand, when a single pump group is running at high load, the engine torque cannot be fully utilized. In some equipment with multiple pump groups but relatively single working conditions, such as some special operating machinery, only one pump group may need to work at full capacity to meet the material delivery pressure and flow requirements. Figure 1 However, due to the lack of an effective torque regulation mechanism in the target engine, the torque output by the engine cannot accurately adapt to the high load requirements of a single pump group, and a large amount of torque is wasted, resulting in low engine efficiency, increased energy consumption, and unnecessary energy waste.

[0067] In addition, existing target engines often cannot reasonably and effectively distribute torque to multiple pump groups. Therefore, how to reasonably and effectively distribute torque to multiple pump groups has become an urgent problem to be solved.

[0068] It should be noted that the method for distributing power to a pump group provided in the embodiment of the present application may be implemented by a device for distributing power to a pump group, and the device for distributing power to a pump group may be implemented as part or all of an electronic device through software, hardware, or a combination of software and hardware, wherein the electronic device may be a control device in an engine. In the following method embodiments, the implementation subject is an electronic device as an example for explanation.

[0069] According to an embodiment of the present invention, an embodiment of a pump group power distribution method is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer target engine such as a set of computer executable instructions, and although the logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0070] In this embodiment, a pump group power distribution method is provided, which can be used for the above-mentioned electronic equipment. Figure 2 : is a flow chart of a method for distributing power to a pump group according to an embodiment of the present invention.Figure 2 As shown, the process includes the following steps:

[0071] Step S101: Obtain the static torque of the master pump group corresponding to the target engine and the fixed torque corresponding to the fixed pump group in the target engine.

[0072] Specifically, the electronic device can calculate the static torque of the master pump group corresponding to the target engine according to the rotational speed of the target engine, or receive the static torque of the master pump group sent by other devices, or receive the static torque of the master pump group input by the user.

[0073] Then, the electronic device can calculate the fixed torque corresponding to each fixed pump according to the pump rotational speed of each fixed pump. Then, add up the fixed torques corresponding to each fixed pump to obtain the fixed torque corresponding to the fixed pump group in the target engine.

[0074] This step will be introduced in detail below.

[0075] Step S102: Based on the static torque of the master pump group and the fixed torque, obtain the distributable torque corresponding to the variable pump group of the target engine.

[0076] Specifically, the electronic device can subtract the fixed torque from the static torque of the master pump group to obtain the distributable torque corresponding to the variable pump group of the target engine.

[0077] Step S103: Obtain the current speed data corresponding to each variable pump in the variable pump group.

[0078] Specifically, the electronic device detects the current speed data corresponding to each variable pump based on the speed sensors installed on each variable pump.

[0079] Among them, the current speed data can be the speed at the current moment, or the current speed sequence composed of the speed at the current moment and the speed at the previous historical moment. The embodiments of the present application do not make specific limitations on the current speed data.

[0080] Step S104: Based on the distributable torque and the current speed data corresponding to each variable pump, determine the target torque corresponding to each variable pump.

[0081] Specifically, the electronic device constructs a set of complex mathematical models. This mathematical model integrates multidisciplinary knowledge such as fluid mechanics and mechanical dynamics, and comprehensively considers the displacement characteristics of the variable pump, the relationship between the pressure and flow rate of the target engine, and the coupling effects between the variable pumps.

[0082] Then, the electronic device can substitute the current speed data corresponding to each variable pump collected and the target operating parameters corresponding to the target engine into the mathematical model.

[0083] Through a rigorous logical judgment process, analyze the matching degree between the current speed data corresponding to each variable pump and the target engine demand. For example, if the real-time speed of a certain variable pump is lower than the speed expected according to the target parameters of the target engine, the electronic device can determine that the output power of this variable pump is insufficient.

[0084] Based on this judgment, the electronic device further uses a mathematical model to calculate the specific values of the setting ratios of each variable pump in the variable pump group, so as to allocate more available torque to it and increase the output power. Conversely, for a variable pump with too high a speed, the algorithm also calculates through the model the amplitude by which its setting ratio should be reduced to avoid excessive output.

[0085] Among them, each variable pump is equipped with a potentiometer group, and the number of potentiometers is one less than that of the variable pump group. The electronic device calculates the specific values of the setting ratios of each variable pump in the variable pump group. Then, use the potentiometer group voltage to control the torque setting ratio of each variable pump in the electro-hydraulic proportional displacement pump group (i.e., the variable pump group). Exemplarily, first run the first variable pump group in the electro-hydraulic proportional displacement pump group, and multiply the available torque of the electro-hydraulic proportional variable pump group by the first variable pump setting ratio X1 in the potentiometer group to obtain the limit torque of the first variable pump. When the first variable pump outputs a current value, the current displacement of the first variable pump can be calculated using the I-Q characteristic curve of the first variable pump. Combining with the pressure value collected by the pressure sensor at the first variable pump port, the target torque of the first variable pump can be calculated.

[0086] Then, subtract the actual output torque of the first variable pump from the available torque of the variable pump group, and then multiply by the setting ratio X2 of the potentiometer group of the second variable pump, and repeat the operation steps. Run variable pumps 2, 3... K-1 in the same way in turn, and finally assign all the remaining torque to the Kth variable pump. Finally, display the torque set value and the actual output value of the variable pump group on the power display.

[0087] The pump group power distribution method provided by the embodiment of the present application obtains the total pump group static torque corresponding to the target engine and the fixed torque corresponding to the fixed displacement pump group in the target engine. Based on the total pump group static torque and the fixed torque, the distributable torque corresponding to the variable displacement pump group of the target engine is obtained, ensuring the accuracy of the distributable torque corresponding to the variable displacement pump group of the obtained target engine. Based on this exact distributable torque value, subsequent scientific and reasonable torque distribution can be carried out according to the actual requirements and operating states of each variable displacement pump. This avoids unreasonable torque distribution caused by unclear adjustable resources, such as excessive torque distribution to a certain variable displacement pump group, exceeding its actual requirements, while other pump groups cannot work properly due to insufficient torque. Through precise calculation of the distributable torque, the torque resource configuration of the target engine is optimized, and the working efficiency and reliability of the entire hydraulic pump group are improved. Then, the current speed data corresponding to each variable displacement pump in the variable displacement pump group is obtained, and based on the distributable torque and the current speed data corresponding to each variable displacement pump, the target torque corresponding to each variable displacement pump is determined. Precise control based on the resources of the target engine and real-time working conditions is achieved. According to the working state of the variable displacement pump group reflected by the speed data, the distributable torque is reasonably distributed, enabling each variable displacement pump group to obtain the torque most suitable for its current working condition. This significantly improves the adaptability of the hydraulic pump group to complex working conditions, ensuring that the target engine can operate efficiently and stably in various situations. For example, in a hydraulic system with multi-task collaborative work, the torque requirements of variable displacement pumps for different working links change dynamically over time. Through this method of determining the target torque based on real-time data, these changes can be quickly responded to, ensuring that each working link can obtain appropriate power support, thereby improving the efficiency and quality of the entire production process. At the same time, precise torque distribution also helps to reduce equipment wear and energy consumption, extend the service life of the equipment, and reduce operating costs. It solves the problems in the prior art that when multiple pump groups operate at high load simultaneously, the engine is prone to over-torque, speed drop, and flameout; when a single pump group operates at high load, the engine torque cannot be fully utilized; and torque distribution cannot be performed on multiple pump groups.

[0088] In this embodiment, a pump group power distribution method is provided, which can be used for the above-mentioned electronic device. Figure 3 It is a flowchart of the pump group power distribution method according to an embodiment of the present invention, as Figure 3 shown, and this process includes the following steps:

[0089] Step S201, obtain the total pump group static torque corresponding to the target engine and the fixed torque corresponding to the fixed displacement pump group in the target engine.

[0090] For this step, please refer to the above introduction to step S101.

[0091] Step S202: Obtain the distributable torque corresponding to the variable pump set of the target engine based on the static torque of the master pump set and the quantitative torque.

[0092] For the introduction of this step, please refer to the above introduction of step S102.

[0093] Step S203: Obtain the current speed data corresponding to each variable pump in the variable pump set.

[0094] For the introduction of this step, please refer to the above introduction of step S103.

[0095] Step S204: Determine the target torque corresponding to each variable pump based on the distributable torque and the current speed data corresponding to each variable pump.

[0096] Specifically, the current speed data is a current speed sequence composed of the speed at the current moment and the historical moment speeds corresponding to a preset number of historical moments before the current moment; the above step S204 may include the following steps:

[0097] Step S2041: Input the static torque of the master pump set, the distributable torque, and the current speed sequence corresponding to each variable pump into a preset torque distribution model.

[0098] Specifically, the above step S2041 may include the following steps:

[0099] Step a1: Obtain the current pressure data and the current flow data corresponding to each variable pump.

[0100] Specifically, the electronic device can obtain the current pressure data and the current flow data corresponding to each variable pump based on the pressure sensors and flow sensors corresponding to each variable pump.

[0101] Step a2: Construct a spatial feature matrix based on the current pressure data and the current flow data.

[0102] Specifically, the electronic device can form a spatial feature matrix S ∈ R with the current pressure data and the current flow data corresponding to each variable pump at the current moment. m×n

[0103] Step a3: Input the spatial feature matrix into a preset graph neural network, and construct a graph structure with each variable pump as a node and the relationship between each variable pump as an edge.

[0104] Specifically, the electronic device inputs the spatial feature matrix into a preset graph neural network, where nodes correspond to each variable pump, and edges are determined according to the physical connection or coupling relationship between variable pumps. In the message passing mechanism of the preset graph neural network, in order to better distinguish the importance of the coupling relationship between different variable pumps, the dynamic weight of the edge is introduced. For edge (i, j), its weight ωij is dynamically adjusted according to the flow interaction intensity and the absolute value of the pressure difference between variable pump i and variable pump j.

[0105] For example,

[0106] where β is a hyperparameter, ΔP ij is the pressure difference between variable pump i and variable pump j, and ΔQij is the flow difference between variable pump i and variable pump j.

[0107] Step a4, based on the graph structure, output the node embedding vector.

[0108] Specifically, for each node in the graph structure, the corresponding current pressure data and current flow data are used as the initial features. For example, the initial feature vector h i0 of node i is measured by the pressure sensor P i related to the pump and the flow sensor measurement Q i and can be expressed as h i0 = [P i ; Q i . If multiple pressure or flow sensors measure different parameters of the same pump, these data will be integrated into the initial feature vector.

[0109] In the preset graph neural network, the node features are updated through the message passing mechanism. During each layer of message passing, a node collects information from its neighbor nodes and updates it in combination with its own features. Taking the l-th layer of message passing as an example, for node i, its set of neighbor nodes is N(i). The message m ij l transmitted by neighbor node j ∈ N(i) to node i is determined by the feature h j l-1 of neighbor node j and the weight ω ij of edge (i, j), and can be dynamically adjusted according to the flow interaction intensity and the absolute value of the pressure difference between variable pump i and variable pump j.

[0110] Among them, a common way to calculate the message is where g is a learnable function, for example, it can be a simple multi-layer perceptron (MLP).

[0111] Node i is updated at the l-th layer according to all the neighbor messages received and its own previous layer features, and the update formula is Among them, f is also a learnable function, which can also be implemented by an MLP. This update process enables the node features to continuously fuse the information from neighboring nodes, gradually capturing the coupling effects between the pump groups. For example, if the flow rate change of variable pump j affects variable pump i through pipeline connection, after multiple message passing and node feature updates, the feature vector h of variable pump i i l will reflect this influence.

[0112] After multiple layers (assumed to be L layers) of message passing and node feature updates, the finally obtained node feature vector h i L is the node embedding vector output by the preset graph neural network. These node embedding vectors fuse the pressure and flow rate information of the variable pump itself and the coupling relationship information learned from other variable pumps through message passing. For example, in a hydraulic system containing multiple variable pumps, the node embedding vector corresponding to a certain variable pump not only contains the pressure and flow rate data characteristics of itself at the current moment, but also contains the influence characteristics of the pressure and flow rate changes of other variable pumps connected to it, as well as the complex coupling effect characteristics between these pumps. These node embedding vectors can comprehensively reflect the state and mutual relationship of the pump in the entire variable pump group, providing key information for subsequent fusion with other features (such as velocity sequence feature vectors, total pump group static torque, allocable torque, etc.) and the model's prediction of the target torque of the variable pump.

[0113] Step a5, input the total pump group static torque, allocable torque, the current velocity sequence corresponding to each variable pump, and the node embedding vector into the preset torque distribution model.

[0114] Specifically, the electronic device can input the total pump group static torque, allocable torque, the current velocity sequence corresponding to each variable pump, and the node embedding vector into the preset torque distribution model.

[0115] Step S2042, the preset torque distribution model extracts features from the total pump group static torque, allocable torque, and the current velocity sequence corresponding to each variable pump, and outputs the target torque corresponding to each variable pump.

[0116] Correspondingly, the above step S2042 may include the following steps:

[0117] Step b1, the preset torque distribution model extracts features from the total pump group static torque, allocable torque, the current velocity sequence corresponding to each variable pump, and the node embedding vector, and outputs the target torque corresponding to each variable pump.

[0118] Specifically, the above step b1 may include the following steps:

[0119] Step b11, the preset torque distribution model extracts features from each current speed sequence based on convolutional kernels of different sizes, and outputs a speed sequence feature vector.

[0120] Specifically, the preset torque distribution model can use a one-dimensional convolutional layer to extract features from each current speed sequence based on convolutional kernels of different sizes, and output a speed sequence feature vector.

[0121] Exemplarily, the current speed sequence is the speed sequence {vi(t - 4), …, vi(t)} of the first 5 time steps of each variable pump, and a one-dimensional convolutional layer is used to extract features from each current speed sequence based on convolutional kernels of different sizes.

[0122] The convolutional kernel is the core tool for feature extraction in a convolutional neural network. Convolutional kernels of different sizes can capture feature information at different scales. In the preset torque distribution model, multiple convolutional kernels of different sizes are usually selected, such as k1 = 3, k2 = 5, k3 = 7, etc. These convolutional kernels of different sizes can extract local and global features from the speed sequence.

[0123] Taking a convolutional kernel w = [w1, w2, …, wk] of size k as an example, a convolution operation is performed on the current speed sequence vi. The convolution operation is to slide the convolutional kernel on the speed sequence and perform a weighted sum on the local window at each position to obtain a new value. Specifically, the convolution output yit at the t-th position can be calculated by the following formula:

[0124]

[0125] where b is the bias term. This process slides the convolutional kernel on the current speed sequence until the entire sequence is covered, and finally a new feature sequence yi is obtained.

[0126] To extract richer feature information, the preset torque distribution model uses multiple convolutional kernels of different sizes to perform convolution operations on the current speed sequence simultaneously. Assuming m convolutional kernels of different sizes are used, each convolutional kernel will obtain a corresponding feature sequence, denoted as yi1, yi2, …, yim.

[0127] After the convolution operation, an activation function is usually applied to introduce non-linearity and enhance the expressive power of the model. A commonly used activation function is ReLU (Rectified Linear Unit), which is defined as: ReLU(x) = max(0, x).

[0128] Apply the ReLU activation function to the feature sequence obtained by each convolutional kernel to obtain the activated feature sequence z i1 ,z i2 ,…,z im. To reduce the dimension of features while retaining important feature information, a pooling operation can be performed on the activated feature sequence. Commonly used pooling operations include Max Pooling and Average Pooling. Taking Max Pooling as an example, assuming the size of the pooling window is p, the maximum value within each local window of length p is taken as the output after pooling. Perform the pooling operation on each activated feature sequence to obtain the pooled feature sequence p i1 , p i2 , …, p im .

[0129] In addition, to enhance the model's ability to capture key features in the current speed sequence, an attention mechanism is introduced. After the convolutional layer, calculate the attention weight α of each convolutional feature j , and then, for each feature sequence p i1 , p i2 , …, p im pooled by each convolutional kernel, multiply it by the corresponding attention weight α j , and then output the speed sequence feature vector.

[0130] Step b12, fuse the speed sequence feature vectors, the static torque of the total pump group, the distributable torque, and the node embedding vectors to generate the initial fused feature.

[0131] Optionally, the electronic device can directly concatenate the speed sequence feature vector, the static torque of the total pump group, the distributable torque, and the node embedding vector in a certain order to form a longer vector as the initial fused feature. For example, assuming the dimension of the speed sequence feature vector is d1, the static torque of the total pump group and the distributable torque are both 1 - dimensional, and the dimension of the node embedding vector is d2, then the dimension of the concatenated initial fused feature vector is d1 + 2 + d2.

[0132] Optionally, the electronic device can also assign a weight to each input data, and then perform a weighted sum of them to obtain the initial fused feature. The size of the weight can be determined according to the importance of different data for the final torque distribution, and usually, these weights can be automatically learned by training the model. For example, for a certain input data x i , its weight is w i , then the initial fused feature F can be expressed as where n is the number of input data.

[0133] Step b13, transform the initial fused feature to generate the transformed feature.

[0134] Specifically, the above - mentioned step b13 can include the following steps:

[0135] Step b131, perform downsampling operations on the initial fusion feature at different scales to obtain multiple sub-fusion features.

[0136] Specifically, the electronic device can adopt a preset downsampling method to perform downsampling operations on the initial fusion feature at different scales. Among them, the preset downsampling method can be the average pooling or the maximum pooling method, and the present application does not make specific limitations on the preset downsampling method.

[0137] Taking average pooling as an example, assume that the initial fusion feature X is a one-dimensional vector of length N. When using an average pooling window of size k and stride s, starting from the starting position of the vector, take the average of the k elements within the window to obtain the first element of the downsampled vector; then slide the window backward by the stride s and repeat the averaging operation until the entire original vector is covered, and finally obtain a sub-fusion feature of length . The maximum pooling operation is to select the maximum value within each pooling window as the output, compressing and extracting the original features from different perspectives. By setting different pooling window sizes and strides, multiple sub-fusion features X s1 、X s2 etc. can be obtained. These sub-fusion features respectively retain the key information of the original features at different scales.

[0138] Step b132, based on the multi-hidden-layer perceptrons corresponding to each sub-fusion feature, perform transformation processing on each sub-fusion feature to generate sub-transformation features corresponding to each sub-fusion feature.

[0139] Specifically, for each sub-fusion feature at a different scale, there is a corresponding small multi-layer perceptron for feature transformation. Each small MLP has an independent network structure and learnable parameters, and can learn and extract the most representative features at that scale according to the scale characteristics of the sub-fusion feature. Taking the sub-fusion feature X s1 as an example, it is input into the corresponding MLP1(X s1 ). After the weighted summation of the hidden layer neurons and the non-linear transformation of the activation function (such as the ReLU function: ReLU(x) = max(0, x)), the original features are mapped to a new feature space, thereby generating sub-transformation features corresponding to the sub-fusion feature. In this process, the MLP continuously adjusts the weights and biases to highlight the important features related to the torque distribution of the hydraulic pump group at that scale, suppress noise and irrelevant information, so that the output sub-transformation features can better serve the subsequent decision-making process.

[0140] Step b133, perform fusion processing on each sub-transformation feature to generate transformation features.

[0141] Optionally, the electronic device may directly splice each sub-transformation feature together in a certain order to form a longer vector as the transformation feature.

[0142] Optionally, the electronic device may also assign a weight to each sub-transformation feature and then perform weighted summation on them to obtain the transformation feature.

[0143] Step b14: Fuse the initial fusion feature with the transformation feature to generate the target feature.

[0144] Specifically, the electronic device may introduce an attention mechanism to fuse the initial fusion feature with the transformation feature to generate the target feature.

[0145] Exemplarily, for each dimension j of the initial fusion feature X and the transformation feature MLP(X), first calculate the score s of the eigenvalue of the two vectors in dimension j through a simple linear layer j , and the formula is s j = W j [X j ; MLP(X) j + b j . Among them, W j and bj are learnable parameters, which are continuously adjusted by training the model to optimize the score calculation; [X j ; MLP(X) j represents splicing the values of X and MLP(X) in dimension j. After obtaining the score, calculate the attention weight α j through the Softmax function, and the formula is The Softmax function converts the score into a probability form, making the sum of the attention weights of all dimensions equal to 1, and the weight size reflects the relative importance of the feature of that dimension. The higher the score of a dimension, the greater the corresponding attention weight.

[0146] Then, according to the calculated attention weights, perform weighted fusion on the initial fusion feature X and the transformation feature MLP(X). The calculation formula for the j-th dimension value of X fused is α j X j + (1 - α j )MLP(X) j .

[0147] In this way, the attention mechanism can perform weighted fusion on features from different sources according to the importance of the features, highlight key features, suppress unimportant features, and further improve the quality of feature fusion.

[0148] Step b15: Based on the target feature, output the target torque corresponding to each variable pump.

[0149] Specifically, when the target feature is input into the input layer, the number of neurons in the input layer is the same as the dimension of the target feature. Each neuron corresponds to an element in the target feature and directly receives the value of that element as input. These neurons do not perform complex calculations but simply pass the received data unchanged to the next layer, which is the first hidden layer. They act as a data interface, introducing the externally input data into the calculation system of the preset torque distribution model.

[0150] Calculation of neurons in the hidden layer:

[0151] Taking the first hidden layer as an example, assume that the first hidden layer has h1 neurons. For the k-th neuron (1 ≤ k ≤ h1) in the first hidden layer, it receives inputs from all neurons in the input layer. Let the weight matrix from the input layer to the first hidden layer be W 1 , and the bias vector be b 1 . The input z k1 of the k-th neuron is calculated through weighted summation: where W kj 1 is the weight matrix, the element in the k-th row and j-th column of W 1 , X j is the j-th element in the target feature X, and a + b + c is the dimension of the target feature. This weighted summation process is essentially a linear combination of the input data. The weight W kj 1 determines the influence degree of the j-th neuron in the input layer on the k-th neuron in the hidden layer, and the bias b k 1 provides a basic value for the activation of the neuron, so that the neuron may be activated even if all inputs are 0. After weighted summation, the obtained value z k 1 will undergo a non-linear transformation through the activation function σ to obtain the output y k 1 = σ(z k 1 ). Commonly used activation functions such as the ReLU function, whose expression is σ(x) = max(0, x). The ReLU function can introduce non-linearity into the model, enabling the MLP to learn complex non-linear relationships. Without an activation function, no matter how many hidden layers the MLP has, it can only perform linear transformations, and its expressive power will be very limited.

[0152] Output of the first hidden layer will be used as the input to the second hidden layer. Similarly, for the neurons in the second hidden layer, they receive the outputs from all neurons in the first hidden layer and perform similar weighted summation and non-linear transformation operations.

[0153] Let the second hidden layer have h2 neurons, and the weight matrix of the second hidden layer be W 2 ×h 1 , and the bias vector be b 2 . For the m-th neuron (1 ≤ m ≤ h2) in the second hidden layer, its input output y m 2 = σ(z m 2 ).

[0154] As the number of hidden layers increases, each layer further extracts and transforms data based on the previous layer, and finally outputs the target torque corresponding to each variable pump. Shallow hidden layers may learn some basic and local features, such as short-term change trends in speed sequences, simple relationships between pressure and flow of a single variable pump, etc. Deeper hidden layers can combine and abstract these basic features to learn more complex and global features, such as the overall coupling mode formed by the connection of multiple pumps through complex pipelines, the comprehensive response characteristics of the target engine under different target operating parameters, etc.

[0155] The pump group power distribution method provided in the embodiment of the present application obtains the current pressure data and current flow data corresponding to each variable pump; constructs a spatial feature matrix based on the current pressure data and the current flow data, and presents the state information of the variable pump group in a structured form. This matrix structure can clearly show the state relationship between the variable pumps, and provides a good data basis for subsequent graph neural network processing. The spatial feature matrix is input into the preset graph neural network, and each variable pump is used as a node, and the relationship between each variable pump is used as an edge to construct a graph structure, which can intuitively represent the physical connection and interaction between the pump groups. In a hydraulic system, the pressure or flow change of a pump group may affect other pump groups connected to it, and the graph structure can accurately capture this coupling relationship. By processing the graph structure through the graph neural network, the complex nonlinear relationship between the pump groups can be deeply excavated. Based on the graph structure, the node embedding vector is output. The node embedding vector output by the graph neural network is a low-dimensional vector representation of each variable pump group and its relationship with other pump groups. These vectors integrate the pressure and flow information of the pump group itself and the coupling relationship information with other pump groups, which can more effectively reflect the state and role of the pump group in the entire target engine. Compared with the traditional analysis method based on single pump group data, the node embedding vector can better capture the global relationship between pump groups and provide richer and more accurate information for torque distribution. The static torque of the total pump group, the distributable torque, the current speed sequence corresponding to each variable pump and the node embedding vector are input into the preset torque distribution model. The preset torque distribution model extracts features from the static torque of the total pump group, the distributable torque, the current speed sequence corresponding to each variable pump and the node embedding vector, and outputs the target torque corresponding to each variable pump. The preset torque distribution model can comprehensively consider the state of the variable pump group, the mutual influence between the variable pump groups and the overall power resources of the target engine according to the static torque of the total pump group, the distributable torque, the current speed sequence corresponding to each variable pump and the node embedding vector, so as to distribute the torque more accurately. For example, when a variable pump group is affected by other pump groups and the pressure or flow changes, the preset torque distribution model can adjust the target torque of the variable pump group in time according to the coupling relationship reflected in the node embedding vector to ensure the stable operation of the target engine.

[0156] The preset torque distribution model extracts features from each current speed sequence based on convolutional kernels of different sizes and outputs speed sequence feature vectors. Convolutional kernels of different sizes can capture features at different time scales in the current speed sequence. Smaller convolutional kernels can focus on local detail changes in the speed sequence, such as short-term speed fluctuations, which are very important for timely responding to rapid changes in the pump group speed. Larger convolutional kernels, on the other hand, can extract features of the speed sequence from a more macroscopic perspective, such as long-term speed trends, which helps to grasp the overall change pattern of the pump group speed. Through this multi-scale feature extraction method, the characteristics of the speed sequence can be more comprehensively described, providing richer and more accurate speed information for subsequent torque distribution. The speed sequence feature vectors, total pump group static torque, distributable torque, and node embedding vectors are fused to generate initial fusion features. It can make full use of information from different aspects. The speed sequence features reflect the dynamic operating state of the pump group, the total pump group static torque and distributable torque represent the power resource status of the target engine, and the node embedding vectors contain information about the interrelationships between pump groups. By fusing this information, the preset torque distribution model can comprehensively consider all aspects of the target engine, avoiding the limitations of relying on a single information source for torque distribution, and thus being able to make more comprehensive and reasonable torque distribution decisions. Then, downsampling operations of different scales are performed on the initial fusion features to obtain multiple sub-fusion features. Downsampling operations of different scales enable the preset torque distribution model to examine the initial fusion features from multiple granularities. This multi-scale feature extraction method comprehensively and meticulously depicts the initial fusion features, providing a rich and diverse information basis for subsequent processing by the preset torque distribution model. Based on the multi-hyper layer perceptrons corresponding to each sub-fusion feature, the sub-fusion features are transformed to generate sub-transformed features corresponding to each sub-fusion feature. Each sub-fusion feature has its unique feature pattern and information distribution, and the corresponding multi-layer perceptron can perform targeted transformations according to the characteristics of the sub-fusion feature. Sub-fusion features of different scales contain different levels of information, and the exclusive multi-layer perceptron can learn the specific relationship between the features at this scale and the target (such as the target torque) by adjusting weights and biases. The sub-transformed features are fused to generate transformed features, which can integrate the sub-transformed features that have been targeted transformed at different scales, ensuring the comprehensiveness and integration of the generated transformed features.

[0157] Next, the initial fusion feature and the transformed feature are fused to generate the target feature, enabling the full combination of the intuitive information of the original feature and the enhanced expression ability of the transformed feature. The initial fusion feature retains the basic information and direct relationships of the original data, while the transformed feature has undergone further processing and optimization, highlighting the features that have an important impact on torque distribution. By fusing the two, the preset torque distribution model can utilize the advantages of both aspects simultaneously. It can not only have a comprehensive understanding of the overall state of the target engine based on the original information but also more accurately capture the key information related to the target torque with the help of the transformed feature, thereby improving the accuracy and reliability of torque distribution. Based on the target feature, the target torque corresponding to each variable pump is output, ensuring more accurate and reasonable torque distribution.

[0158] In this embodiment, a pump group power distribution method is provided, which can be used in the above-mentioned electronic device. Figure 4 It is a flowchart of the pump group power distribution method according to an embodiment of the present invention, as Figure 4 shown, and this process includes the following steps:

[0159] Step S301, obtain the total pump group static torque corresponding to the target engine and the fixed torque corresponding to the fixed pump group in the target engine.

[0160] Specifically, the above step S301 includes the following steps:

[0161] Step S3011, obtain the current engine speed corresponding to the target engine.

[0162] Specifically, the electronic device can measure the current engine speed corresponding to the target engine based on a speed sensor.

[0163] Step S3012, according to the engine external characteristic curve corresponding to the target engine, determine the rated torque of the target engine at the current engine speed.

[0164] Specifically, the electronic device can receive the engine external characteristic curve corresponding to the target engine input by the user, or receive the engine external characteristic curve corresponding to the target engine sent by other devices, or search for the engine external characteristic curve corresponding to the target engine in the storage space.

[0165] Then, the electronic device determines the rated torque of the target engine at the current engine speed according to the engine external characteristic curve corresponding to the target engine.

[0166] Step S3013, calculate the total pump group static torque corresponding to the target engine according to the engine torque absorption ratio corresponding to the target engine.

[0167] Specifically, the electronic device can receive the engine torque absorption ratio corresponding to the target engine input by the user, can also receive the engine torque absorption ratio corresponding to the target engine sent by other devices, or can determine the engine torque absorption ratio corresponding to the target engine based on the attribute information of the target engine. Then, the electronic device multiplies the engine torque absorption ratio corresponding to the target engine by the rated torque at the current engine speed to obtain the static torque of the main pump group corresponding to the target engine.

[0168] Step S3014: Obtain the fixed torque corresponding to the fixed displacement pump group in the target engine.

[0169] Specifically, the electronic device can obtain the theoretical displacement of each fixed displacement pump in the fixed displacement pump group. Then, it multiplies the theoretical displacement by the speed of the fixed displacement pump to obtain the theoretical flow rate of the fixed displacement pump. The electronic device calculates the actual flow rate of the fixed displacement pump by multiplying the theoretical flow rate by the volumetric efficiency.

[0170] Next, the electronic device obtains the outlet pressure of the fixed displacement pump corresponding to the fixed displacement pump based on the pressure sensor at the fixed displacement pump port. Then, it calculates the hydraulic power of the fixed displacement pump based on the outlet pressure of the fixed displacement pump multiplied by the actual flow rate of the fixed displacement pump. Finally, based on the relationship between hydraulic power, torque, and speed P = T×n×2π÷60 (where P is hydraulic power in kW, T is torque in N·m, and n is speed in r / min), and since the mechanical power consumed by the fixed displacement pump Pm = Ph÷ηm (Pm is mechanical power), by combining these, the calculation formula for the torque consumed by the fixed displacement pump can be obtained: T = 9550×Ph÷(n×ηm). Substitute the previously calculated hydraulic power Ph, the known speed n, and the mechanical efficiency ηm into this formula to calculate the torque consumed by the fixed displacement pump.

[0171] Finally, the electronic device adds up the torques consumed by each fixed displacement pump to calculate the fixed torque corresponding to the fixed displacement pump group in the target engine.

[0172] Step S302: Obtain the distributable torque corresponding to the variable displacement pump group of the target engine based on the static torque of the main pump group and the fixed torque.

[0173] For this step, please refer to the above introduction to step S202 and will not be elaborated here.

[0174] Step S303: Obtain the current speed data of each variable displacement pump in the variable displacement pump group.

[0175] For this step, please refer to the above introduction to step S203 and will not be elaborated here.

[0176] Step S304: Determine the target torque corresponding to each variable displacement pump based on the distributable torque and the current speed data of each variable displacement pump.

[0177] Specifically, the above-mentioned step S304 may include the following steps:

[0178] Step S3041: Input the static torque of the main pump group, the distributable torque, and the current speed sequences corresponding to the variable pumps into a preset torque distribution model.

[0179] For this step, please refer to the above introduction to step S2041, and details will not be elaborated here.

[0180] Step S3042: The preset torque distribution model extracts features from the static torque of the main pump group, the distributable torque, and the current speed sequences corresponding to the variable pumps, and outputs the target torque corresponding to each variable pump.

[0181] Specifically, the above-mentioned step S3042 may include the following steps:

[0182] Step c1: The preset torque distribution model extracts features from the static torque of the main pump group, the distributable torque, and the current speed sequences corresponding to the variable pumps, and outputs the candidate torque corresponding to each variable pump.

[0183] Specifically, for the specific process of the preset torque distribution model extracting features from the static torque of the main pump group, the distributable torque, and the current speed sequences corresponding to the variable pumps and outputting the candidate torque corresponding to each variable pump, please refer to the above, and details will not be elaborated here.

[0184] Step c2: Based on each candidate torque, determine the variable pump status information corresponding to each variable pump.

[0185] Specifically, the electronic device may calculate the variable pump status information according to the candidate torque corresponding to each variable pump and other known parameters (such as the rated parameters of the variable pump, the current rotation speed, etc.). For example, for the i-th variable pump, its load rate is where T rated,i is the rated torque of the i-th variable pump. The efficiency of the variable pump can be calculated according to the power calculation formula of the hydraulic system and the actually measured parameters such as flow rate and pressure. The stability index can be obtained by statistically analyzing the speed sequence. For example, calculate the standard deviation of the speed sequence where, is the average value of the speed sequence. Combining these indexes, obtain the status information vector S of each variable pump group pump,i = [L i , E i , σ i , …], where E i represents the efficiency of the i-th variable pump.

[0186] Step c3: According to the variable pump status information corresponding to each variable pump, determine the engine status information corresponding to the target engine.

[0187] Specifically, the electronic device can calculate the total candidate torque of all variable pumps Combining information such as the static torque of the total pump group and the distributable torque, analyze the load condition of the target engine. The load rate of the target engine can be expressed as where T engine-rated is the rated torque of the target engine, and T static is the static torque of the total pump group. The rotational speed stability of the target engine can be evaluated by monitoring the fluctuation of the target engine speed. If the load of the variable pump group changes drastically, it may cause an increase in the fluctuation of the target engine speed. At the same time, according to the efficiency of the variable pump group, evaluate the energy utilization efficiency of the target engine. If the efficiency of the variable pump group is low, the target engine needs to output more power to maintain the operation of the pump group. Considering these factors comprehensively, the state information vector of the target engine is obtained as Sengine = [L engine , σ engine-speed , E engine-utilization , …], where σ engine-speed is the standard deviation of the target engine speed, and E engine-utilization is the energy utilization efficiency of the target engine.

[0188] Step c4, adjust each candidate torque according to the engine state, and output the target torque corresponding to each variable pump.

[0189] Specifically, the electronic device can determine the corresponding candidate torque adjustment strategy according to the engine state corresponding to the target engine. If the engine load rate is too high, in order to avoid engine overload, it is necessary to appropriately reduce the candidate torque of some variable pumps; if the rotational speed stability of the engine is poor, it may be necessary to optimize the torque distribution of the variable pump group to reduce sudden changes in the load. For example, when the engine load rate exceeds a set threshold (such as 80%), reduce the candidate torque of some variable pumps with a higher load rate by a certain proportion (such as 10%).

[0190] Specifically, adjust the candidate torque of each variable pump. Let the adjustment coefficient vector be α = [α1, α2, …, α n , and determine the adjustment coefficient of each variable pump according to the engine state and the adjustment strategy. For example, for the pump group with too high load rate and a greater impact on the engine speed, its adjustment coefficient α i is less than 1; for the variable pump with a lower load rate and a positive effect on the stability of the target engine, its adjustment coefficient α i can be appropriately greater than 1. The adjusted target torque vector is T target = [α1T c1 , α2T c2 , …, α n T cn , where Ttarget,i = α i T ci represents the target torque of the i-th variable pump.

[0191] The pump group power distribution method provided by the embodiments of the present application obtains the current engine speed corresponding to the target engine. According to the engine external characteristic curve corresponding to the target engine, the rated torque of the target engine at the current engine speed is determined; the rated torque that the target engine can provide at the current engine speed can be accurately determined. This enables the operator or the control of the target engine to clearly understand the output capacity of the engine, providing an important basis for reasonably distributing torque and arranging work tasks. For example, if it is known that the rated torque of the engine at the current speed is low, it is necessary to avoid assigning an excessive load to the target engine to prevent the target engine from being overloaded. According to the engine torque absorption ratio corresponding to the target engine, the total pump group static torque corresponding to the target engine is calculated. Calculating the total pump group static torque through the engine torque absorption ratio can fully consider the actual working requirements and operating characteristics of the target engine. Different hydraulic systems have different degrees of torque absorption from the engine under different working conditions, and the torque absorption ratio can accurately reflect this difference. Therefore, calculating the total pump group static torque according to the torque absorption ratio can make the calculation result more in line with the actual situation of the target engine, providing more accurate basic data for subsequent torque distribution and target engine control. In addition, by considering the engine torque absorption ratio to calculate the total pump group static torque, torque distribution can be more precisely carried out according to the actual output capacity of the engine and the requirements of the target engine, avoiding problems such as target engine failures or low efficiency caused by unreasonable torque distribution, and improving the control accuracy and stability of the entire hydraulic system.

[0192] In addition, the preset torque distribution model extracts features of the static torque of the total pump group, the distributable torque, and the current speed sequence corresponding to each variable pump, and outputs the candidate torque corresponding to each variable pump. Through feature extraction, the preset torque distribution model can deeply explore the potential rules behind this information and provide an accurate basis for subsequent torque distribution. Based on the extracted features, the candidate torque corresponding to each variable pump is output, which realizes the preliminary and reasonable distribution of torque, so that the torque distribution is preliminarily matched with the actual operation requirements of the variable pump group, avoiding the problem of inefficiency or equipment damage caused by blind torque distribution. Based on each candidate torque, the variable pump status information corresponding to each variable pump is determined, which can comprehensively evaluate the working status of the variable pump group and timely discover whether the variable pump group has problems such as overload, inefficiency or unstable operation. According to the variable pump status information corresponding to each variable pump, the engine status information corresponding to the target engine is determined. According to the status information of each variable pump, the engine status information corresponding to the target engine is determined, which can establish a close connection between the variable pump group and the target engine. By monitoring the target engine status information, potential problems can be warned in advance. According to the engine status, each candidate torque is adjusted, and the target torque corresponding to each variable pump is output. According to the engine status, the candidate torque is adjusted to further optimize the torque distribution. If the engine load rate is too high, in order to avoid engine overload, the candidate torque of some variable pump groups can be appropriately reduced; if the engine speed stability is poor, the torque distribution of the variable pump group can be optimized to reduce sudden changes in load. Through this adjustment, the torque distribution is more in line with the actual working capacity of the engine, improving the operating efficiency and stability of the entire target engine.

[0193] In an optional embodiment of the present application, Figure 5 As shown, an implementation method of a pump group power distribution method is provided. Specifically, the following steps are included:

[0194] S1. Pressure detection of the metering pump port and calculation of the metering pump torque.

[0195] Pressure detection: The outlet pressure of each metering pump (metering pump 1 to metering pump m) is detected by a metering pump port pressure sensor, which are recorded as p1, p2, ..., pm respectively.

[0196] Torque calculation: Calculate the torque of each metering pump according to the formula qm×pm / 2 / π, where qm is the displacement of the metering pump. Then add the torques of all metering pumps to obtain the torque of the metering pump group Td=∑i=1mqm×pm / 2 / π.

[0197] S2. Engine parameter detection and pump group static torque calculation.

[0198] Parameter detection: Detect the engine speed n, obtain the engine torque Tn according to the engine external characteristic curve, and determine the torque absorption ratio pot at the same time.

[0199] Torque calculation: Calculate the static torque of the pump group according to the formula Tps = Tn × pot.

[0200] S3. Torque calculation of the electro-hydraulic proportional displacement variable pump group.

[0201] Subtract the torque Td of the fixed-displacement pump group from the static torque Tps of the pump group to obtain the torque Tb of the electro-hydraulic proportional displacement variable pump group: Tb = Tps - Td.

[0202] S4. Torque distribution of the variable pump (potentiometer setting).

[0203] For variable pumps 1 to k - 1, set the percentages X1, X2,..., X(k - 1) (range: 5% - 95%) through the potentiometer. Calculate the limiting torque of each variable pump in sequence, such as Tb1 = Tb × X1, Tb2 = Tb - Tb1, Tb(k - 1) = Tb - Tb1' - Tb2' -... - Tb(k - 2)', Tb(k) = Tb - Tb1' - Tb2' -... - Tb(k - 1)'.

[0204] S5. Limiting torque equivalent calculation.

[0205] Calculate the correlation coefficient according to the formula yi = (QNimax - QNimin) × 2 × π / qbimax, and then multiply the limiting torque Tbi of each variable pump by yi to obtain the limiting torque equivalent Tli = Tbi × yi.

[0206] S6. Outlet pressure monitoring and current calculation of the variable pump.

[0207] Monitor the outlet pressure pbi of each variable pump, calculate the current input to the electro-hydraulic control variable pump according to the formula QNi = Tli / pbi + QNimin, and ensure that the actual output current QN' ≤ QNi.

[0208] S7. Calculation of the actual output parameters of the variable pump.

[0209] Displacement calculation: Calculate the actual output displacement according to the formula qbi = qbimax × (QNi - QNimin) / (QNimax - QNimin).

[0210] Torque calculation: Calculate the actual output torque according to the formula Tbi' = qbi × pbi / 2 / π.

[0211] S8. Output percentage calculation and display.

[0212] Calculation: Calculate the output percentage Xi′ = Tbi′ / Tb×100% and the set percentage Xi = Tbi / Tb×100% respectively.

[0213] Display: Display the relevant percentage data on the display, and at the same time, the controller records the power distribution situation, including the output percentage and the set percentage of each variable pump.

[0214] In this embodiment, a pump group power distribution system is also provided. This device is used to implement the above-mentioned embodiments and preferred implementation manners, and those that have been described will not be repeated. As used hereinafter, the term "module" can be a combination of software and / or hardware that realizes a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware is also possible and contemplated.

[0215] This embodiment provides a pump group power distribution system, as Figure 6 shown, including:

[0216] A first acquisition module 401, configured to acquire the total pump group static torque corresponding to the target engine and the fixed torque corresponding to the fixed pump group in the target engine;

[0217] A subtraction module 402, configured to acquire the distributable torque corresponding to the variable pump group of the target engine based on the total pump group static torque and the fixed torque;

[0218] A second acquisition module 403, configured to acquire the current speed data corresponding to each variable pump in the variable pump group;

[0219] A determination module 404, configured to determine the target torque corresponding to each variable pump based on the distributable torque and the current speed data corresponding to each variable pump.

[0220] In some alternative implementation manners, the current speed data is a current speed sequence composed of the current moment speed and the historical moment speeds corresponding to a preset number of historical moments before the current moment; the determination module 404 is specifically configured to input the total pump group static torque, the distributable torque, and the current speed sequence corresponding to each variable pump into a preset torque distribution model; the preset torque distribution model extracts features from the total pump group static torque, the distributable torque, and the current speed sequence corresponding to each variable pump, and outputs the target torque corresponding to each variable pump.

[0221] In some alternative embodiments, the determining module 404 is specifically configured to obtain the current pressure data and the current flow rate data corresponding to each variable pump; construct a spatial feature matrix based on the current pressure data and the current flow rate data; input the spatial feature matrix into a preset graph neural network, use each variable pump as a node, and the relationship between each variable pump as an edge to construct a graph structure; based on the graph structure, output node embedding vectors; input the static torque of the total pump group, the distributable torque, the current speed sequence corresponding to each variable pump, and the node embedding vectors into a preset torque distribution model; correspondingly, the preset torque distribution model extracts features from the static torque of the total pump group, the distributable torque, the current speed sequence corresponding to each variable pump, and the node embedding vectors, and outputs the target torque corresponding to each variable pump.

[0222] In some alternative embodiments, the determining module 404 is specifically configured to the preset torque distribution model extracts features from each current speed sequence based on convolutional kernels of different sizes, and outputs speed sequence feature vectors; perform a fusion process on each speed sequence feature vector, the static torque of the total pump group, the distributable torque, and the node embedding vectors to generate an initial fusion feature; perform a transformation on the initial fusion feature to generate a transformed feature; fuse the initial fusion feature and the transformed feature to generate a target feature; based on the target feature, output the target torque corresponding to each variable pump.

[0223] In some alternative embodiments, the determining module 404 is specifically configured to perform downsampling operations on the initial fusion feature at different scales to obtain multiple sub-fusion features; perform transformation processing on each sub-fusion feature based on a multi-hidden layer perceptron corresponding to each sub-fusion feature to generate sub-transformed features corresponding to each sub-fusion feature; perform a fusion process on each sub-transformed feature to generate a transformed feature.

[0224] In some alternative embodiments, the determining module 404 is specifically configured to the preset torque distribution model extracts features from the static torque of the total pump group, the distributable torque, and the current speed sequence corresponding to each variable pump, and outputs candidate torques corresponding to each variable pump; based on each candidate torque, determine the variable pump state information corresponding to each variable pump; according to the variable pump state information corresponding to each variable pump, determine the engine state information corresponding to the target engine; adjust each candidate torque according to the engine state, and output the target torque corresponding to each variable pump.

[0225] In some alternative embodiments, the first obtaining module 401 is specifically configured to obtain the current engine speed corresponding to the target engine;

[0226] Determine the rated torque of the target engine at the current engine speed according to the engine external characteristic curve corresponding to the target engine;

[0227] Calculate the static torque of the master cylinder group corresponding to the target engine according to the engine torque absorption ratio corresponding to the target engine.

[0228] The further function descriptions of the above-mentioned various modules and units are the same as those in the corresponding above-mentioned embodiments, and will not be repeated here.

[0229] The pump group power distribution system in this embodiment is presented in the form of functional units. Here, the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and a memory that execute one or more software or fixed programs, and / or other devices that can provide the above functions.

[0230] The embodiment of the present invention also provides an electronic device having the above-mentioned Figure 6 shown pump group power distribution system.

[0231] Please refer to Figure 7 , Figure 7 is a schematic structural diagram of an electronic device provided by an optional embodiment of the present invention. As shown in Figure 7 , the electronic device includes: one or more processors 10, a memory 20, and interfaces for connecting various components, including a high-speed interface and a low-speed interface. Each component communicates with each other using different buses and can be installed on a common motherboard or installed in other ways as needed. The processor can process instructions executed within the electronic device, including instructions stored in the memory or on the memory to display graphical information of the GUI on an external input / output device (such as a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Similarly, multiple electronic devices can be connected, and each device provides some necessary operations (such as a server array, a set of blade servers, or a multi-processor target engine). Figure 7 In

[0232] The electronic device further includes a communication interface 30 for the electronic device to communicate with other devices or communication networks.

[0233] The embodiment of the present invention also provides an engine, including: an engine body and Figure 7 the shown electronic device, and the electronic device is used to execute the pump group power distribution method described in any of the above embodiments.

[0234] The embodiment of the present invention also provides a work machine including the above engine.

[0235] Embodiments of the present invention also provide a computer-readable storage medium. The method according to the embodiments of the present invention can be implemented in hardware, firmware, or be implemented as computer code that can be recorded on a storage medium, or be implemented as computer code that is originally stored in a remote storage medium or a non-transitory machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the method described herein can be stored as such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only memory, a random access memory, a flash memory, a hard disk, or a solid-state drive, etc.; further, the storage medium can also include a combination of the above types of memories. It can be understood that a computer, a processor, a microprocessor controller, or programmable hardware includes a storage component that can store or receive software or computer code, and when the software or computer code is accessed and executed by the computer, the processor, or the hardware, the method shown in the above embodiments is implemented.

[0236] A part of the present invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should be able to understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Herein, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible by the computer.

[0237] Although the embodiments of the present invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the present invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for power distribution of a pump unit, characterized in that The method includes: Obtaining the static torque of the master pump group corresponding to the target engine and the fixed torque corresponding to the fixed pump group in the target engine; Based on the static torque of the master pump group and the fixed torque, obtaining the distributable torque corresponding to the variable pump group of the target engine; Obtaining the current speed data corresponding to each variable pump in the variable pump group; Based on the distributable torque and the current speed data corresponding to each variable pump, determining the target torque corresponding to each variable pump.

2. The method according to claim 1, wherein The current speed data is a current speed sequence composed of the speed at the current moment and the historical moment speeds corresponding to a preset number of historical moments before the current moment; The determining the target torque corresponding to each variable pump based on the distributable torque and the current speed corresponding to each variable pump includes: Inputting the static torque of the master pump group, the distributable torque, and the current speed sequence corresponding to each variable pump into a preset torque distribution model; The preset torque distribution model extracts features from the static torque of the master pump group, the distributable torque, and the current speed sequence corresponding to each variable pump, and outputs the target torque corresponding to each variable pump.

3. The method according to claim 2, wherein The inputting the static torque of the master pump group, the distributable torque, and the current speed sequence corresponding to each variable pump into a preset torque distribution model includes: Obtaining the current pressure data and the current flow data corresponding to each variable pump; Constructing a spatial feature matrix based on the current pressure data and the current flow data; Inputting the spatial feature matrix into a preset graph neural network, and constructing a graph structure with each variable pump as a node and the relationship between each variable pump as an edge; Based on the graph structure, outputting a node embedding vector; Inputting the static torque of the master pump group, the distributable torque, the current speed sequence corresponding to each variable pump, and the node embedding vector into a preset torque distribution model; Correspondingly, the preset torque distribution model extracts features from the static torque of the master pump group, the distributable torque, the current speed sequence corresponding to each variable pump, and the node embedding vector, and outputs the target torque corresponding to each variable pump.

4. The method according to claim 3, characterized in that The preset torque distribution model extracts features from the static torque of the master pump group, the distributable torque, the current speed sequence corresponding to each variable pump, and the node embedding vector, and outputs the target torque corresponding to each variable pump, including: The preset torque distribution model extracts features from each current speed sequence based on convolutional kernels of different sizes, and outputs a speed sequence feature vector; Performing a fusion process on each speed sequence feature vector, the static torque of the master pump group, the distributable torque, and the node embedding vector to generate an initial fusion feature; Performing a transformation on the initial fusion feature to generate a transformed feature; Fusing the initial fusion feature with the transformed feature to generate a target feature; Based on the target feature, outputting the target torque corresponding to each variable pump.

5. The method according to claim 4, characterized in that, The performing a transformation on the initial fusion feature to generate a transformed feature includes; Perform downsampling operations on the initial fusion features at different scales to obtain multiple sub-fusion features; Based on the multi-hyperlayer perceptrons corresponding to each of the sub-fusion features, perform transformation processing on each of the sub-fusion features to generate sub-transformation features corresponding to each of the sub-fusion features; Perform fusion processing on each of the sub-transformation features to generate the transformation features.

6. The method according to claim 2, wherein The preset torque distribution model extracts features from the static torque of the master pump group, the distributable torque, and the current speed sequences corresponding to each of the variable pumps, and outputs the target torque corresponding to each of the variable pumps, including: The preset torque distribution model extracts features from the static torque of the master pump group, the distributable torque, and the current speed sequences corresponding to each of the variable pumps, and outputs candidate torques corresponding to each of the variable pumps; Based on each of the candidate torques, determine the variable pump status information corresponding to each of the variable pumps; According to the variable pump status information corresponding to each of the variable pumps, determine the engine status information corresponding to the target engine; Adjust each of the candidate torques according to the engine status, and output the target torque corresponding to each of the variable pumps.

7. The method according to claim 1, wherein The obtaining of the static torque of the master pump group corresponding to the target engine includes: Obtain the current engine speed corresponding to the target engine; According to the engine external characteristic curve corresponding to the target engine, determine the rated torque of the target engine at the current engine speed; According to the engine torque absorption ratio corresponding to the target engine, calculate the static torque of the master pump group corresponding to the target engine.

8. A pump set power distribution system, characterized in that, The system includes: A first acquisition module for acquiring the static torque of the master pump group corresponding to the target engine and the fixed torque corresponding to the fixed pump group in the target engine; A subtraction module for obtaining the distributable torque corresponding to the variable pump group of the target engine based on the static torque of the master pump group and the fixed torque; A second acquisition module for acquiring the current speed data corresponding to each variable pump in the variable pump group; A determination module for determining the target torque corresponding to each variable pump based on the distributable torque and the current speed data corresponding to each variable pump.

9. An engine, characterized in that, Includes: An engine body and an electronic device, the electronic device includes: a memory and a processor, the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to execute the pump group power distribution method according to any one of claims 1 to 7.

10. An earthmoving machine, characterized in that, The work machine includes the engine according to claim 9.

Images