Lubricating hydraulic equipment operation and maintenance method and system based on online hydraulic model

Abstract

The invention discloses a lubricating hydraulic equipment operation and maintenance method and system based on an online hydraulic model, and relates to the technical field of intelligent operation and maintaining.The lubricating hydraulic equipment operation and maintenance method based on the online hydraulic model comprises the following steps that a structure diagram and operation principle description of lubricating hydraulic equipment are obtained; constructing an online hydraulic model, obtaining observation information at the current moment, and comparing the observation information with prediction data of the online hydraulic model to generate observation features and dependency features; pre-decision-making is carried out based on the observation features and the dependency features, and a potential operation and maintenance scheme is determined; generating a hierarchical communication network based on a minimum spanning tree algorithm; and generating a plurality of joint decision-making schemes based on the potential operation and maintenance schemes, optimizing the generated joint decision-making schemes by using the hierarchical communication network, evaluating all the joint decision-making schemes by using a state action value function, and generating a final operation and maintenance scheme according to an evaluation result. It is ensured that the operation and maintenance scheme can be efficiently transmitted and implemented during execution, and scientificity and reliability of decision making are ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent operation and maintenance, in particular to a lubricated hydraulic equipment operation and maintenance method and system based on an online hydraulic model. BACKGROUND

[0002] Lubricated hydraulic equipment is a kind of equipment used to control fluid pressure, flow and direction. They transmit power through hydraulic oil and use lubricating oil to reduce the wear of mechanical parts, ensuring the efficient operation of the system. These devices are widely used in industrial manufacturing, engineering machinery, transportation and other fields.

[0003] The operation and maintenance of lubricated hydraulic equipment refers to the process of operating, maintaining and managing the lubricated hydraulic system and its various components to ensure the normal operation of the equipment, prolong its service life and improve its work efficiency. For example, regular inspection is the basis for ensuring the normal operation of the lubricated hydraulic system; taking certain maintenance measures before the equipment fails to prevent failure; when the equipment fails, quickly diagnose and repair to reduce downtime, etc.

[0004] For example, Chinese patent 202210679014.5 discloses a hydraulic forming intelligent equipment manufacturing process fault diagnosis and operation and maintenance method and system, which constructs an expert system and a knowledge graph of intelligent equipment, and realizes fast and accurate matching of operation and maintenance schemes. However, the above method still has the following shortcomings: the method does not construct an online hydraulic model, the understanding of the equipment is not comprehensive enough, and multiple operation and maintenance schemes are not generated for comparison and optimization, which may lead to the final selected operation and maintenance scheme not being comprehensive and optimal.

[0005] At present, there is no effective solution to the problems in the related art. SUMMARY

[0006] In view of the problems in the related art, the present application proposes a lubricated hydraulic equipment operation and maintenance method and system based on an online hydraulic model to overcome the above technical problems existing in the prior art.

[0007] To this end, the specific technical solutions adopted by the present application are as follows: According to one aspect of the present application, a lubricated hydraulic equipment operation and maintenance method based on an online hydraulic model is provided, which comprises the following steps: S1, obtaining the structure data and operation principle data of the lubricated hydraulic equipment, identifying the key components and their relationships, and obtaining the structure diagram and operation principle description.

[0008] S2, based on the structure diagram and operation principle description, constructing an online hydraulic model, obtaining observation information at the current time, and comparing it with the prediction data of the online hydraulic model to generate observation features and dependency features.

[0009] S3, making a pre-decision based on the observation features and the dependency features, and determining a potential operation and maintenance scheme.

[0010] S4, generating a hierarchical communication network based on a minimum spanning tree algorithm, and realizing selective directional communication.

[0011] S5, generating a plurality of joint decision schemes based on the potential operation and maintenance scheme, optimizing the generated joint decision schemes by using the hierarchical communication network, evaluating all the joint decision schemes by using a state-action value function, and generating a final operation and maintenance scheme according to the evaluation results.

[0012] Generating a hierarchical communication network based on a minimum spanning tree algorithm, and realizing selective directional communication includes the following steps: S41, taking each key component in the lubricating hydraulic equipment as a node in the network, and establishing edges between the nodes according to the dependency relationship and functional connection between the key components, to obtain an initial graph; S42, setting the weight of each edge to obtain a device hierarchical network graph with weights; S43, determining the number of clustering partitions, and clustering the nodes by using a clustering algorithm to obtain a clustering result; S44, taking the nodes in the clustering result as nodes of the communication network, and finding a minimum spanning tree in the clustering network by using a minimum spanning tree algorithm to obtain a hierarchical communication network based on a minimum spanning tree; S45, defining directional communication rules, and configuring the network to realize selective directional communication.

[0013] Further, an online hydraulic model is constructed, and observation information at the current time is obtained and compared with predicted data of the online hydraulic model to generate observation features and dependency features, which includes the following steps: S21, based on a structure diagram and an operation principle description, obtaining the internal fluid flow of the lubricating hydraulic equipment, and constructing a fluid dynamics model; S22, constructing a thermodynamic model according to the thermodynamic characteristics of the lubricating hydraulic equipment; S23, integrating the fluid dynamics model and the thermodynamic model to form an online hydraulic model, and inputting initial conditions and parameters at the current time into the online hydraulic model to generate predicted data of the operation state at the current time; S24, obtaining observation information at the current time, calculating the difference between the observation information and the predicted data, and identifying abnormal values that exceed the expected range; S25, calculating the residual error of the observation data and the predicted data, extracting observation features and dependency features based on difference analysis and residual error calculation.

[0014] Further, the pre-decision is made based on the observation features and the dependency features, and the potential operation and maintenance scheme is determined including the following steps: S31, extracting key features from the observation features and the dependency features; S32, classifying the fault mode by using a machine learning algorithm based on the extracted key features; S33, analyzing potential causes for the identified fault mode; S34, matching an appropriate potential operation and maintenance scheme from an existing operation and maintenance scheme library according to the fault mode and the potential causes.

[0015] Further, the fault mode is classified by using a machine learning algorithm based on the extracted key features including the following steps: S321, collecting observation features and dependency features under normal operation and various fault modes in history, and corresponding key features; S322, labeling the key features in S321 to indicate the operation state of the lubricated hydraulic equipment at that time; S323, dividing the labeled data set into a training set and a test set; S324, training a support vector machine model using the training set and the test set, and evaluating and adjusting the support vector machine model; S325, inputting new observation features and dependency features, and using the trained and evaluated support vector machine model to predict the current state of the lubricated hydraulic equipment.

[0016] Further, the number of clustering partitions is determined, and the nodes are clustered by using a clustering algorithm to obtain a clustering result including the following steps: S431, determining the number of clustering partitions by using the elbow rule; S432, randomly selecting k nodes as initial clustering centers; S433, calculating the distance from each node to each clustering center using the Euclidean distance; S434, according to the calculated distance, each node is classified into the nearest cluster; S435, calculating the new center of each cluster; S436, repeating the steps of S433-S435 until a preset iteration number is reached or the clustering center no longer changes significantly.

[0017] Further, the nodes in the clustering result are used as nodes of the communication network, and a minimum spanning tree algorithm is used to find a minimum spanning tree in the clustering network to obtain a hierarchical communication network based on the minimum spanning tree including the following steps: S441, regarding all nodes in each clustering partition as independent nodes in the network; S442, construct an undirected graph, wherein the edges of the undirected graph represent the connections between the individual nodes, and assign a weight to each edge to obtain a weighted undirected graph; S443, based on the weighted undirected graph and performing a minimum spanning tree algorithm, selecting the minimum weight edges to gradually build a tree until all nodes are included; S444, according to the structure of the minimum spanning tree, determine the hierarchical relationship between the nodes, and identify the root node in the minimum spanning tree; S445, determine the path from the root node to all other nodes, form a hierarchical structure.

[0018] Further, based on the potential operation and maintenance scheme, a plurality of joint decision schemes are generated, and the generated joint decision schemes are optimized using the hierarchical communication network, and all joint decision schemes are evaluated using the state-action value function, according to the evaluation result, the final operation and maintenance scheme is generated, including the following steps: S51, combine different potential operation and maintenance schemes to generate a plurality of joint decision schemes; S52, optimize the generated joint decision scheme using the hierarchical relationship network; S53, evaluate each joint decision scheme using the state-action value function, and calculate the state-action value; S54, sort all joint decision schemes according to the state-action value, and select the optimal scheme.

[0019] Further, the state-action value function is used to evaluate each joint decision scheme, and the state-action value is calculated, including the following steps: S531, define the state of the operation and maintenance scheme and the operation and maintenance action; S532, initialize the state-action value table for each potential operation and maintenance scheme, and design a reward function; S533, execute the operation and maintenance action in each joint decision scheme, and obtain new state information; S534, update the state-action value of each joint decision scheme based on the new state information, the reward obtained and the prediction of future returns.

[0020] Further, the formula of the reward function is: ; The update formula of the state-action value is: ; In the formula, R imd ( t ) represents the immediate reward at time point t ; P t represents the time pointt actual measurement value of the performance index, P ref reference value of the performance index, C represents the cost of executing the operation and maintenance scheme, R represents the risk index at the time point t R long represents the long-term return; w 1 represents the weight coefficient of the performance index, w 2 represents the weight coefficient of the operation and maintenance cost, w 3 represents the weight coefficient of the risk index; Q old S A represents the Q value in the state A S before the operation and maintenance action Q new S A represents the updated Q value; gamma represents the discount factor, α represents the learning rate; Q S next A next represents the expected Q value of the operation and maintenance action S next A next in the new state max Anext Q S next A next represents the maximum value in the Q values of all operation and maintenance actions in the new state S next

[0021] According to another aspect of the present application, there is also provided an online hydraulic model-based lubricated hydraulic equipment operation and maintenance system, which comprises an equipment data acquisition module, a feature acquisition module, a potential scheme generation module, a network generation module and a final scheme generation module; wherein the equipment data acquisition module, the feature acquisition module, the potential scheme generation module, the network generation module and the final scheme generation module are sequentially connected.

[0022] ​​​​​​​​​​​​An equipment data acquisition module is configured to acquire structural data and operation principle data of the lubricated hydraulic equipment, identify key components and mutual relations, and obtain a structural diagram and an operation principle description.

[0023] A feature acquisition module is configured to construct an online hydraulic model based on the structural diagram and the operation principle description, acquire observation information at a current time, compare the observation information with predicted data of the online hydraulic model, and generate observation features and dependent features.

[0024] A potential scheme generation module is configured to make a pre-decision based on the observation features and the dependent features, and determine a potential operation and maintenance scheme.

[0025] A network generation module is configured to generate a hierarchical communication network based on a minimum spanning tree algorithm, and realize selective directional communication.

[0026] A final scheme generation module is configured to generate a plurality of joint decision schemes based on the potential operation and maintenance scheme, optimize the generated joint decision schemes by using the hierarchical communication network, evaluate all the joint decision schemes by using a state-action value function, and generate a final operation and maintenance scheme according to an evaluation result.

[0027] The lubricated hydraulic equipment operation and maintenance method and system based on the online hydraulic model have the following advantages: (1) The lubricated hydraulic equipment operation and maintenance method and system based on the online hydraulic model provide basic data for subsequent steps by acquiring a structural diagram and an operation principle description of the lubricated hydraulic equipment, and can accurately simulate the operation status of the equipment, predict possible performance changes and faults, make a pre-decision by using observation features and dependent features, ensure that the pre-decision is based on sufficient information, reduce blindness and subjectivity, and help to better plan maintenance work by making a potential operation and maintenance scheme in advance.

[0028] (2) The lubricated hydraulic equipment operation and maintenance method and system based on the online hydraulic model generate a hierarchical communication network by using a minimum spanning tree algorithm, ensure that the communication path is the shortest and most efficient, reduce the delay and resource consumption of information transmission, help to realize hierarchical processing and distribution of information, reduce communication conflicts and information redundancy, ensure that the joint decision scheme generated by optimizing the hierarchical communication network can be efficiently transmitted and implemented when executed, reduce execution errors and delays, quantitatively evaluate the decision by using a state-action value function, ensure the scientificity and reliability of the decision, reduce the deviation caused by subjective judgment, and finally generate an operation and maintenance scheme that has high feasibility and execution efficiency, helps to realize the best maintenance effect and equipment performance. BRIEF DESCRIPTION OF DRAWINGS

[0029] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed in the embodiments. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained from these drawings without creative labor.

[0030] Figure 1 is a flow chart of a lubricating hydraulic equipment operation and maintenance method based on an online hydraulic model according to an embodiment of the present application; Figure 2 is a principle block diagram of a lubricating hydraulic equipment operation and maintenance system based on an online hydraulic model according to an embodiment of the present application.

[0031] In the drawings: 1, equipment data acquisition module; 2, feature acquisition module; 3, potential scheme generation module; 4, network generation module; 5, final scheme generation module. DETAILED DESCRIPTION

[0032] To further illustrate the embodiments, the present application provides drawings, which are part of the disclosure of the present application, mainly used to illustrate the embodiments, and can explain the operating principle of the embodiments in conjunction with the related description of the specification. With reference to these contents, those skilled in the art should understand other possible embodiments and advantages of the present application. The components in the drawings are not drawn to scale, and similar component symbols are usually used to represent similar components.

[0033] According to an embodiment of the present application, a lubricating hydraulic equipment operation and maintenance method and system based on an online hydraulic model are provided.

[0034] The present application will be further described in conjunction with the drawings and specific embodiments. As shown in Figure 1 According to an embodiment of the present application, a lubricating hydraulic equipment operation and maintenance method based on an online hydraulic model is provided, which includes the following steps: S1, obtaining structure data and operation principle data of the lubricating hydraulic equipment, identifying key components and mutual relationship, and obtaining structure diagram and operation principle description. Specifically: Collecting equipment documents and equipment composition of the lubricating hydraulic equipment, and obtaining equipment operation principle data; Identifying key components and mutual relationship, such as key pumps, main valves, key sensors, etc.; hydraulic oil flow path, pressure change path, temperature change path, etc.; Displaying the physical layout of each component and the connection relationship of the key components. Explain the function and interaction of each part, describe the working process and operation steps under different working conditions.

[0035] S2, based on the structure diagram and the operating principle description, an online hydraulic model is constructed, including a fluid dynamics model and a thermodynamics model, and observation information (such as pressure, temperature, flow, etc.) at the current time is obtained and compared with the prediction data of the online hydraulic model to generate observation features and dependent features.

[0036] In further embodiments, constructing the online hydraulic model and obtaining the observation information at the current time and comparing it with the prediction data of the online hydraulic model to generate observation features and dependent features includes the following steps: S21, based on the structure diagram and the operating principle description, the internal fluid flow of the lubricated hydraulic equipment is obtained, a fluid dynamics model is constructed, including the fluid flow equations and interactions of components such as pipes, valves, and pumps.

[0037] S22, according to the thermodynamic characteristics of the lubricated hydraulic equipment, a thermodynamics model is constructed, including the thermodynamic equations of heat sources, heat conduction, and heat dissipation processes.

[0038] S23, the fluid dynamics model and the thermodynamics model are integrated to form an online hydraulic model, and the initial conditions and parameters at the current time are input to the online hydraulic model to generate prediction data for the current operating state.

[0039] S24, the observation information at the current time is obtained, the difference between the observation information and the prediction data is calculated, and abnormal values outside the expected range are identified.

[0040] S25, the residual error of the observation data and the prediction data is calculated, and based on the difference analysis and residual error calculation, observation features and dependent features are extracted.

[0041] For example, the observation parameters obtained in the lubricated hydraulic equipment include pressure, temperature, and flow.

[0042] Observation information at the current time: Observed pressure, 150 bar. Observed temperature, 80℃. Observed flow, 20L / min.

[0043] Prediction data of the online hydraulic model: Predicted pressure, 140 bar. Predicted temperature, 78℃. Predicted flow, 22L / min.

[0044] Difference analysis and residual error calculation: Pressure difference, 10 bar. Temperature difference, ℃. Flow difference, -2L / min.

[0045] Through comparative analysis, it is found that: The residual error of pressure is large, indicating that the actual pressure is higher than the predicted value. The residual error of flow is negative, indicating that the actual flow is lower than the predicted value.

[0046] Observation features, high pressure and low flow are the key observation features, which suggest that there is a problem of blockage or pump efficiency reduction in the system.

[0047] Dependency features, through further analysis, it is found that the reduction of flow mainly depends on the increase of pressure, which is caused by the increase of internal resistance of the system.

[0048] S3, based on the observation features and dependency features, pre-decision is made to determine the potential operation and maintenance scheme.

[0049] In further embodiments, the pre-decision based on the observation features and dependency features to determine the potential operation and maintenance scheme includes the following steps: S31, extract key features from observation features and dependency features, such as constant values of pressure, temperature, flow, etc. such as mean, maximum, minimum), change trend (such as slope, fluctuation range).

[0050] S32, based on the extracted key features, use machine learning algorithm to classify possible failure modes, define possible failure modes according to past experience and equipment working principle, such as leakage, blockage, wear and tear, etc.

[0051] S33, for the identified failure mode, combine expert knowledge and historical data to analyze potential causes, such as equipment leakage, pipeline blockage, sensor failure, improper parameter setting, etc.

[0052] S34, according to the failure mode and potential cause, use matching algorithm (such as similarity calculation, rule matching, etc.) to match the appropriate potential operation and maintenance scheme from the existing operation and maintenance scheme library, including replacing parts, adjusting parameters, adding lubricating oil, etc.

[0053] In further embodiments, based on the extracted key features, using machine learning algorithm to classify failure modes includes the following steps: S321, collect observation features and dependency features under normal operation and various failure modes in history, as well as corresponding key features.

[0054] S322, label the key features in S321, indicating the operating state of the lubricating hydraulic equipment at that time (normal or specific failure mode).

[0055] S323, divide the labeled data set into training set and test set by randomization.

[0056] S324, select support vector machine as the fault classification model, determine the kernel function used, train the support vector machine model using the training set and test set, and evaluate and adjust the support vector machine model.

[0057] S325, input the new observation features and dependent features, and use the support vector machine model trained and evaluated to predict the state of the lubricated hydraulic equipment, including fault mode or normal state.

[0058] S4, generate a hierarchical communication network based on the minimum spanning tree algorithm to realize selective directed communication, and share observation information and intention information through the hierarchical communication network.

[0059] In further embodiments, generating a hierarchical communication network based on the minimum spanning tree algorithm to realize selective directed communication includes the following steps: S41, each key component in the lubricated hydraulic equipment is taken as a node in the network, and edges between nodes are established according to the dependency relationship and functional connection between key components to obtain an initial graph.

[0060] S42, weights of each edge are set based on flow, pressure change or other key parameters between key components to obtain a device hierarchical network graph with weights.

[0061] S43, the number of cluster partitions is determined, and a clustering algorithm is used to cluster nodes to obtain a clustering result; a suitable clustering algorithm such as K-means clustering algorithm is selected to cluster nodes. If the number of nodes is small, clustering can not be performed.

[0062] S44, nodes in the clustering result are taken as nodes of the communication network, and a minimum spanning tree algorithm is used to find a minimum spanning tree in the clustering network to obtain a hierarchical communication network based on the minimum spanning tree; S45, a directed communication rule is defined, and the network is configured to realize selective directed communication.

[0063] Examples: The nodes include a pump, a main valve, an auxiliary valve, a cooler, a pressure sensor 1, a pressure sensor 2, a temperature sensor, and a flow sensor.

[0064] The edges include pump→main valve, weight 1; pump→auxiliary valve, weight 1; pump→cooler, weight 1; main valve→pressure sensor 1, weight 0.5; auxiliary valve→pressure sensor 2, weight 0.5; cooler→temperature sensor, weight 0.5; pump→flow sensor, weight 1.

[0065] Construct a minimum spanning tree: Pump.

[0066] ├── Main valve.

[0067] │ └── Pressure sensor 1.

[0068] ├── Auxiliary valve.

[0069] │ └── pressure sensor 2.

[0070] ├── cooler.

[0071] │ └── temperature sensor.

[0072] └── flow sensor.

[0073] From the above minimum spanning tree, it is ensured that each node in the tree is connected to the pump, forming an effective hierarchical communication network.

[0074] A directed communication needs to be established between nodes.

[0075] Pump → main valve. Pump → auxiliary valve. Pump → cooler. Main valve → pressure sensor 1. Auxiliary valve → pressure sensor 2. Cooler → temperature sensor. Pump → flow sensor.

[0076] Define communication rules.

[0077] Pump → main valve: transmit pump speed and flow setting, once per second.

[0078] Pump → auxiliary valve: transmit pump speed and flow setting, once per second.

[0079] Pump → cooler: transmit pump speed and cooling requirement, once per second.

[0080] Main valve → pressure sensor 1: transmit pressure data, once per second.

[0081] Auxiliary valve → pressure sensor 2: transmit pressure data, once per second.

[0082] Cooler → temperature sensor: transmit temperature data, once per second.

[0083] Pump → flow sensor: transmit flow data, once per second.

[0084] In further embodiments, the number of cluster partitions is determined, and the nodes are clustered using a clustering algorithm to obtain a clustering result, including the following steps: S431, determine the number of cluster partitions using the elbow rule.

[0085] S432, randomly select k nodes as initial cluster centers, which will serve as reference points for the initial clusters.

[0086] S433, calculate the distance from each node to each cluster center using the Euclidean distance, with shorter distances indicating closer proximity to the center.

[0087] S434, according to the calculated distance, each node is assigned to the nearest cluster.

[0088] S435, calculate the new center of each cluster, i.e. calculate the mean of all nodes in the cluster as the new cluster center.

[0089] S436, repeat the steps of S433-S435 until a preset number of iterations is reached or the cluster centers no longer change significantly.

[0090] In further embodiments, the nodes in the clustering result are taken as nodes of the communication network, and a minimum spanning tree algorithm is used to find the minimum spanning tree in the clustering network, and a hierarchical communication network based on the minimum spanning tree is obtained including the following steps: S441, take all nodes in each cluster partition as independent nodes in the network.

[0091] S442, construct an undirected graph, where the edges of the undirected graph represent the connections between the individual independent nodes, and each edge is assigned a weight, obtaining a weighted undirected graph; the weight between the nodes after clustering is based on the connection strength between the original node sets they represent, which can be the sum or average of the weights of the edges between the original nodes.

[0092] S443, based on the weighted undirected graph and performing a minimum spanning tree algorithm, selecting edges with the smallest weights to gradually build a tree until all nodes are included.

[0093] It should be noted that the minimum spanning tree algorithm includes: Prim algorithm: starting from an arbitrary node, gradually expanding to all nodes, and adding the node connected to the existing tree and the edge with the smallest weight each time.

[0094] Kruskal algorithm: sort all edges by weight from small to large, select the smallest weight edge that does not form a loop, and gradually build a tree.

[0095] S444, determine the hierarchical relationship between nodes according to the structure of the minimum spanning tree, and identify the root node in the minimum spanning tree, which is usually the node connected by the edge with the smallest weight.

[0096] S445, determine the path from the root node to all other nodes to form a hierarchical structure.

[0097] S5, generate a number of joint decision schemes based on the potential operation and maintenance scheme, optimize the generated joint decision scheme using the hierarchical communication network, and evaluate all joint decision schemes using the state-action value function, and generate the final operation and maintenance scheme according to the evaluation result.

[0098] In further embodiments, based on the potential operation and maintenance scheme, a number of joint decision schemes are generated, and the generated joint decision scheme is optimized using the hierarchical communication network, and all joint decision schemes are evaluated using the state-action value function, and the final operation and maintenance scheme is generated according to the evaluation result, including the following steps: S51, combine different potential maintenance schemes according to their affected device components and operation types to generate several joint decision schemes.

[0099] S52, use hierarchical relationship network optimization to generate joint decision schemes, ensuring efficient information transmission and collaborative operation between different device components.

[0100] S53, evaluate each joint decision scheme using state-action value function (Q function) to calculate state-action value (Q value).

[0101] S54, sort all joint decision schemes according to state-action value, and select the optimal scheme.

[0102] For example, potential maintenance scheme A: adjust pump speed, increase flow. Potential maintenance scheme B: adjust main valve, optimize main system pressure. Potential maintenance scheme C: adjust auxiliary valve, optimize auxiliary system pressure. Potential maintenance scheme D: clean cooler, reduce system temperature.

[0103] Joint decision scheme 1: scheme A + scheme B + scheme D.

[0104] Joint decision scheme 2: scheme A + scheme C + scheme D.

[0105] Joint decision scheme 1: Pump → Main valve → Cooler → Pressure sensor 1, temperature sensor.

[0106] Collaborative operation: after increasing pump speed, immediately adjust pressure through main valve and simultaneously clean cooler to ensure temperature drop.

[0107] Joint decision scheme 2: Pump → Auxiliary valve → Cooler → Pressure sensor 2, temperature sensor.

[0108] Collaborative operation: after increasing pump speed, immediately adjust pressure through auxiliary valve and simultaneously clean cooler to ensure temperature drop.

[0109] In further embodiments, evaluating each joint decision scheme using state-action value function to calculate state-action value includes the following steps: S531, define the state of the maintenance scheme, including device running state, resource usage, historical maintenance records, etc., define maintenance actions such as replacing components, adjusting configurations, restarting services, etc.

[0110] S532, initialize the state-action value table for each potential maintenance scheme, record the Q value of each state-action pair, and design a reward function to evaluate the immediate effect and long-term impact of each action. Rewards can be based on system performance improvement, cost savings, risk reduction, etc.

[0111] S533, performing the operation and maintenance action in each joint decision scheme, and obtaining new state information, i.e., obtaining the actual impact of the operation and maintenance action on the lubricating hydraulic equipment.

[0112] S534, updating the state-action value of each joint decision scheme based on the new state information, the reward obtained, and the prediction of future return.

[0113] Among them, the acquisition of new state: after performing the action, the environment will feedback a new state, which is the starting point in the next update process.

[0114] Obtaining rewards: obtaining an immediate reward according to the executed action. This reward is usually calculated according to the performance index of the new state, reflecting the direct result of the action.

[0115] Prediction of future return: the discount factor and the maximum expected Q value of the new state are used together to predict the future return. The discount factor determines the importance of future rewards relative to current rewards, and the maximum expected Q value represents the maximum return expected from all possible next actions in the new state.

[0116] In further embodiments, the formula of the reward function is: ; The update formula of the state-action value is: ; In the formula, R imd ( t ) represents the immediate reward at time point t ; P t represents the actual measurement value of the performance index at time point t , P ref represents the reference value or ideal value of the performance index, C represents the cost of executing the operation and maintenance scheme, R represents the risk index at time point t , which considers the failure probability, potential downtime or safety risk, etc. R long represents the long-term return, which considers the expected future reward; w 1 represents the weight coefficient of the performance index, w 2 represents the weight coefficient of the operation and maintenance cost, w 3 represents the weight coefficient of the risk index; Q old ( S , A ) represents the operation and maintenance actionA Previous state S Q value under Q new ( S , A ) represents the updated Q value; gamma γ represents the discount factor, used to calculate the current value of future rewards, α α represents the learning rate; Q ( S next , A next ) represents the expected Q value of taking the operation and maintenance action S next under the new state A next ; max Anext Q ( S next , A next ) represents the maximum value among all Q values of operation and maintenance actions under the new state S next .

[0117] For example, the following parameters are set.

[0118] Performance indicator: the current performance indicator value of the device (e.g., traffic), ranging from 0 to 100.

[0119] Reference value of performance indicator: the performance indicator value in the ideal state, set to 90.

[0120] Operation and maintenance cost: the cost of executing each operation and maintenance action, for example, the cost of adjusting the pump speed is 10, and the cost of cleaning the cooler is 20.

[0121] Risk indicator: based on failure probability, potential downtime, or safety risk, etc., set to a value between 0 and 1, for example, 0.05.

[0122] Long-term return: the expected future reward, which can be set to a fixed value or estimated based on certain prediction models, for example, 100.

[0123] Weight coefficient w 1, w 2, w 3: set to 1, 2, and 1 respectively.

[0124] Q old ( S , A ) is set to 50. The discount factor is set to 0.9. The learning rate is set to 0.1. The maximum expected Q value under the new state is 80.

[0125] At a certain time point t , the actual performance index of the equipment is equal to 85, the operation and maintenance action in joint decision scheme 1 is executed, including adjusting the pump speed and cleaning the cooler, and the total cost is 30. The updated Q value is 62.65.

[0126] According to the high and low of the Q value, all joint decision schemes are sorted, and the optimal scheme is selected, that is, the scheme with the highest Q value is selected.

[0127] As Figure 2 shown, according to another embodiment of the application, a lubricated hydraulic equipment operation and maintenance system based on an online hydraulic model is also provided, which comprises an equipment data acquisition module 1, a feature acquisition module 2, a potential scheme generation module 3, a network generation module 4 and a final scheme generation module 5; wherein the equipment data acquisition module 1, the feature acquisition module 2, the potential scheme generation module 3, the network generation module 4 and the final scheme generation module 5 are connected in sequence.

[0128] The equipment data acquisition module 1 is used to acquire the structure data and operation principle data of the lubricated hydraulic equipment, identify the key components and mutual relationship, and obtain the structure diagram and operation principle description.

[0129] The feature acquisition module 2 is used to construct an online hydraulic model based on the structure diagram and operation principle description, acquire observation information at the current time, and compare it with the predicted data of the online hydraulic model to generate observation features and dependent features.

[0130] The potential scheme generation module 3 is used to make pre-decision based on the observation features and dependent features to determine potential operation and maintenance schemes.

[0131] The network generation module 4 is used to generate a hierarchical communication network based on the minimum spanning tree algorithm to realize selective directional communication.

[0132] The final scheme generation module 5 is used to generate a plurality of joint decision schemes based on the potential operation and maintenance schemes, optimize the generated joint decision schemes using the hierarchical communication network, evaluate all joint decision schemes using the state action value function, and generate the final operation and maintenance scheme according to the evaluation result.

[0133] In summary, the lubricating hydraulic equipment operation and maintenance method and system based on an online hydraulic model provided by the application provide basic data for subsequent steps by acquiring the structure diagram and operation principle description of the lubricating hydraulic equipment, and can accurately simulate the operation condition of the equipment, predict possible performance changes and faults; the observation features and dependent features are used for pre-decision making, ensuring that the pre-decision making is based on sufficient data, reducing blindness and subjectivity. The potential operation and maintenance scheme is formulated in advance, which is helpful for better planning of maintenance work. The application generates a hierarchical communication network through a minimum spanning tree algorithm, ensures that the communication path is the shortest and most efficient, reduces the delay and resource consumption of information transmission, is helpful for realizing hierarchical processing and distribution of information, reduces communication conflicts and information redundancy, and ensures that the joint decision scheme generated through optimization of the hierarchical communication network can be efficiently transmitted and implemented when executed, reduces execution errors and delays. Finally, the state-action value function is used for quantitative evaluation, ensuring the scientificity and reliability of the decision making, and reducing the deviation caused by subjective judgment. The finally generated operation and maintenance scheme is subjected to multiple evaluations and optimizations, has high feasibility and execution efficiency, and is helpful for realizing the best maintenance effect and equipment performance.

[0134] The above merely describes the preferred embodiments of the application and is not intended to limit the application, and any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the application shall be included in the protection scope of the application.

Claims

1. A lubrication hydraulic equipment operation and maintenance method based on an online hydraulic model, characterized in that: The lubrication hydraulic equipment operation and maintenance method based on the online hydraulic model includes the following steps: S1. Obtain the structural data and operating principle data of the lubrication hydraulic equipment, identify the key components and their relationships, and obtain a structural diagram and operating principle description; S2. Based on the structural diagram and the description of the operating principle, an online hydraulic model is constructed, and the observation information at the current moment is obtained and compared with the predicted data of the online hydraulic model to generate observation features and dependency features; S3. Make preliminary decisions based on observation features and dependency features to determine potential operation and maintenance solutions; S4, based on the minimum spanning tree algorithm, generates a hierarchical communication network to achieve selective directed communication; S5. Generate several joint decision-making plans based on potential operation and maintenance plans, optimize the generated joint decision-making plans using the hierarchical communication network, evaluate all joint decision-making plans using the state-action value function, and generate a final operation and maintenance plan based on the evaluation results; The method of generating a hierarchical communication network based on a minimum spanning tree algorithm and realizing selective directed communication includes the following steps: S41, treating each key component in the lubrication hydraulic equipment as a node in the network, and establishing edges between the nodes based on the dependency relationships and functional connections between the key components to obtain an initial graph; S42. Set the weight of each edge to obtain a weighted device hierarchical network diagram; S43, determining the number of cluster partitions, and clustering the nodes using a clustering algorithm to obtain a clustering result; S44, using the nodes in the clustering result as nodes of the communication network, and using the minimum spanning tree algorithm to find a minimum spanning tree in the clustering network to obtain a hierarchical communication network based on the minimum spanning tree; S45. Define directed communication rules and configure the network to implement selective directed communication.

2. The lubrication hydraulic equipment operation and maintenance method based on the online hydraulic model according to claim 1 is characterized in that: The online hydraulic model is constructed, and observation information at the current moment is obtained and compared with the predicted data of the online hydraulic model to generate observation features and dependency features, including the following steps: S21. Based on the structural diagram and operating principle description, obtain the internal fluid flow of the lubricating hydraulic equipment and construct a fluid dynamics model; S22. Construct a thermodynamic model based on the thermodynamic characteristics of lubricating hydraulic equipment; S23. Integrate the fluid dynamics model and the thermodynamics model to form an online hydraulic model, and input the initial conditions and parameters at the current moment into the online hydraulic model to generate prediction data for the current operating state; S24. Obtain the observation information at the current moment, calculate the difference between the observation information and the predicted data, and identify abnormal values ​​that exceed the expected range; S25. Obtain the residuals between the observed data and the predicted data, and extract the observed features and dependent features based on difference analysis and residual calculation.

3. The lubrication hydraulic equipment operation and maintenance method based on the online hydraulic model according to claim 2 is characterized in that: The pre-decision-making based on the observation characteristics and the dependent characteristics to determine the potential operation and maintenance solution includes the following steps: S31. Extract key features from observation features and dependent features; S32. Based on the extracted key features, the machine learning algorithm is used to classify the fault modes; S33. Analyze potential causes of identified failure modes; S34. According to the failure mode and potential cause, match the appropriate potential operation and maintenance solution from the existing operation and maintenance solution library.

4. The lubrication hydraulic equipment operation and maintenance method based on the online hydraulic model according to claim 3 is characterized in that: The method of classifying the fault mode using a machine learning algorithm based on the extracted key features includes the following steps: S321. Collect historical observation characteristics and dependency characteristics under normal operation and various failure modes, as well as corresponding key characteristics; S322. Mark the key features in S321 to indicate the operating status of the lubricating hydraulic equipment at that time; S323, dividing the labeled data set into a training set and a test set; S324, training a support vector machine model using the training set and the test set, and evaluating and adjusting the support vector machine model; S325. For the new observation features and dependent features input, the trained and evaluated support vector machine model is used to predict the current state of the lubrication hydraulic equipment.

5. The lubrication hydraulic equipment operation and maintenance method based on the online hydraulic model according to claim 4 is characterized in that: Determining the number of cluster partitions and clustering the nodes using a clustering algorithm to obtain a clustering result includes the following steps: S431. Determine the number of cluster partitions using the elbow rule; S432, randomly select k nodes as initial cluster centers; S433, using Euclidean distance to calculate the distance between each node and each cluster center; S434, classifying each node into the nearest cluster according to the calculated distance; S435, calculating the new center of each cluster; S436. Repeat steps S433-S435 until the preset number of iterations is reached or the cluster center no longer changes significantly.

6. The lubrication hydraulic equipment operation and maintenance method based on the online hydraulic model according to claim 5 is characterized in that: The method of using the nodes in the clustering results as nodes of the communication network and using the minimum spanning tree algorithm to find the minimum spanning tree in the clustering network to obtain a hierarchical communication network based on the minimum spanning tree includes the following steps: S441, treating all nodes of each cluster partition as independent nodes in the network; S442. Construct an undirected graph, wherein the edges of the undirected graph represent connections between independent nodes, and a weight is assigned to each edge to obtain a weighted undirected graph; S443, based on the weighted undirected graph and executing the minimum spanning tree algorithm, selecting the edge with the minimum weight and gradually building the tree until all nodes are included; S444. Determine the hierarchical relationship between nodes based on the structure of the minimum spanning tree, and identify the root node in the minimum spanning tree; S445. Determine the path from the root node to all other nodes to form a hierarchical structure.

7. The method for operating and maintaining lubricating hydraulic equipment based on an online hydraulic model according to claim 6, characterized in that: The method of generating a plurality of joint decision-making schemes based on potential operation and maintenance schemes, optimizing the generated joint decision-making schemes using a hierarchical communication network, and evaluating all joint decision-making schemes using a state-action value function, and generating a final operation and maintenance scheme based on the evaluation results includes the following steps: S51. Combining different potential operation and maintenance solutions to generate several joint decision-making solutions; S52. Joint decision-making scheme generated by optimizing hierarchical relationship network; S53, using the state-action value function to evaluate each joint decision-making scheme and calculate the state-action value; S54. Sort all joint decision-making plans according to the state-action value and select the best plan.

8. The lubrication hydraulic equipment operation and maintenance method based on the online hydraulic model according to claim 7 is characterized in that: The state-action value function is used to evaluate each joint decision-making scheme, and the calculation of the state-action value includes the following steps: S531. Define the status and operation and maintenance actions of the operation and maintenance plan; S532. Initialize a state-action value table for each potential operation and maintenance solution and design a reward function; S533. Execute the operation and maintenance actions in each joint decision-making solution and obtain new status information; S534: Based on the new state information, the acquisition of rewards, and the prediction of future returns, the state action value of each joint decision-making solution is updated.

9. The lubrication hydraulic equipment operation and maintenance method based on an online hydraulic model according to claim 8, characterized in that: The formula of the reward function is: ； The update formula of the state action value is: ； Where, R imd ( t ) indicates the time point t Instant rewards; P t Indicates a time point t The actual measured value of the performance indicator, P ref Indicates the reference value of the performance indicator, C Represents the cost of executing the operation and maintenance plan, and R represents the time point t risk indicators, R long Indicates long-term returns; w 1 represents the weight coefficient of the performance index, w 2 represents the weight coefficient of operation and maintenance cost, w 3 represents the weight coefficient of the risk indicator; Q old ( S , A ) indicates that operation and maintenance actions are being taken A Previous status S The Q value under Q new ( S , A ) represents the updated Q value; γ represents the discount factor, α represents the learning rate; Q ( S next , A next ) indicates that in the new state S next Operation and maintenance actions taken A next The expected Q value of max Anext Q ( S next , A next ) indicates that in the new state S next The maximum Q value of all operation and maintenance actions.

10. A lubricating hydraulic equipment operation and maintenance system based on an online hydraulic model, used to implement the lubricating hydraulic equipment operation and maintenance method based on an online hydraulic model according to any one of claims 1 to 9, characterized in that: The lubrication hydraulic equipment operation and maintenance system based on the online hydraulic model includes an equipment data acquisition module, a feature acquisition module, a potential solution generation module, a network generation module and a final solution generation module; wherein the device data acquisition module, the feature acquisition module, the potential solution generation module, the network generation module, and the final solution generation module are sequentially connected; The equipment data acquisition module is used to obtain the structural data and operating principle data of the lubrication hydraulic equipment, identify key components and their interrelationships, and obtain a structural diagram and operating principle description; The feature acquisition module is used to build an online hydraulic model based on the structural diagram and the operating principle description, and obtain the observation information at the current moment, and compare it with the predicted data of the online hydraulic model to generate observation features and dependent features; The potential solution generation module is used to make a preliminary decision based on the observation characteristics and the dependent characteristics to determine the potential operation and maintenance solution; The network generation module is used to generate a hierarchical communication network based on a minimum spanning tree algorithm to achieve selective directed communication; The final solution generation module is used to generate several joint decision solutions based on potential operation and maintenance solutions, and optimize the generated joint decision solutions using a hierarchical communication network, and use a state-action value function to evaluate all joint decision solutions, and generate a final operation and maintenance solution based on the evaluation results.

Images