An intelligent management construction method and system for hydraulic metal structure equipment based on digital twin

Abstract

The present invention provides an intelligent management construction method and system for hydraulic metal structure equipment based on digital twin. The method includes: constructing a physical space including the physical environment of the water conservancy project and the physical entities of the hydraulic metal structures, and simultaneously constructing an information space including the twin environment of the water conservancy project and the digital twins of the hydraulic metal structures. The physical space and the information space interact through twin data including historical data, online data, and prediction data. By constructing digital twins of the hydraulic metal structure equipment and its operating environment, the present invention can achieve state monitoring, operation and maintenance, and safety assessment throughout the entire life cycle of the hydraulic metal structure equipment. The formed intelligent management system for hydraulic metal structure equipment has the characteristics of being well-organized, digitalized, integrated, and visualized, and can, to a certain extent, reduce the problems of inadequate safety monitoring, incomplete information integration, low resource regulation efficiency, and untimely risk warning during the operation and service of the hydraulic metal structure equipment.

Description

Technical Field

[0001] The present invention relates to the field of management of hydraulic metal structure equipment, and in particular to an intelligent management construction method and system for hydraulic metal structure equipment based on digital twin. Technical Background

[0002] Hydraulic metal structure equipment, such as gates, hoists, penstocks, etc., as key equipment for water flow regulation and control in water conservancy and hydropower projects, its safety is extremely important. Compared with civil engineering structures, hydraulic metal structure equipment has a high failure rate and fast changes in failure states, and has high requirements for the real-time nature of monitoring and emergency handling. Whether the operation and maintenance of metal structures can be carried out efficiently will directly affect the stable operation of water conservancy and hydropower projects.

[0003] For the detection, evaluation and management of hydraulic metal structure equipment, relevant specifications have been formed in the industry, such as "SL105-2016 Anti-corrosion Specification for Hydraulic Metal Structures", "SL101-2014 Safety Detection Technical Regulations for Hydraulic Steel Gates and Hoists", "JTS304-2019 Technical Specifications for Inspection and Evaluation of Hydraulic Structures in Waterway Engineering", "JTS153-3-2007 Anti-corrosion Technology Specifications for Steel Structures in Seaport Engineering", etc. However, these specification standards do not form a unified comprehensive system for safety management, and various detection and evaluation methods mainly rely on manual regular inspections, with untimely information interaction, and there is also a lack of comprehensive evaluation of the operation status, corrosion status, etc. of hydraulic metal structures. In terms of safety management, a systematic system has not been formed, and there is a lack of an information-based intelligent management method for hydraulic metal structure equipment.

[0004] Therefore, in order to better achieve the status supervision of the entire life cycle of hydraulic metal structure equipment and comprehensively evaluate the health and safety of hydraulic metal structure equipment, it is urgent to change the traditional management mode, integrate various work operations informatization, build an intelligent management system for hydraulic metal structure equipment, and realize the comprehensive operation and maintenance management of each metal structure equipment. Summary of the Invention

[0005] The first object of the present invention is to provide an intelligent management construction method for hydraulic metal structure equipment based on digital twin, and form an intelligent management system for hydraulic metal structure equipment with functions of data management, dynamic monitoring, global display, status diagnosis, risk assessment, and comprehensive regulation.

[0006] To this end, the above object of the present invention is achieved through the following technical solutions:

[0007] An intelligent management construction method for hydraulic metal structure equipment based on digital twin, characterized in that: the construction method includes the following steps:

[0008] S1. Construct a physical space, which includes the physical environment of the water conservancy project and the physical entities of hydraulic metal structures; the physical entities of hydraulic metal structures include at least one of gates, hoists, steel pipes, and unit volutes.

[0009] S2. Construct an information space, which includes the digital twin environment of the water conservancy project and the digital twin of hydraulic metal structures.

[0010] S3. Generate twin data for the interaction between the physical space and the information space. The twin data includes historical data, online data, and prediction data.

[0011] While adopting the above technical solutions, the present invention can also adopt or combine the following technical solutions:

[0012] As a preferred technical solution of the present invention: In step S2, the construction of the information space includes the construction of an environment model, a geometric model, a physical model, and a rule model; the parameters of the environment model include water level, rainfall, temperature, humidity, pH value, medium flow rate, and solution conductivity; the geometric model includes geometric dimensions, constituent materials, combined structures, and assembly logics.

[0013] As a preferred technical solution of the present invention: The construction of the rule model is based on management systems, norms and standards, standard libraries, expert experience, and intelligent comparison.

[0014] As a preferred technical solution of the present invention: In step S3, the real-time information formed by the physical environment of the water conservancy project and the physical entities of hydraulic metal structures is compared instantaneously with the output information of the information space.

[0015] When the comparison result does not meet the rule model, data analysis is carried out. On the one hand, when the analysis result shows that there are model defects in the information space, the model is corrected; on the other hand, when the analysis result shows that the physical entities of hydraulic metal structures are abnormal, behavior control is carried out.

[0016] As a preferred technical solution of the present invention: The data analysis includes historical fault analysis, safety status assessment, and risk accident prediction.

[0017] As a preferred technical solution of the present invention: The safety status assessment is based on the historical working conditions and historical faults, to excavate and deduce the fault types and causes of the hydraulic metal structure equipment, and combined with real-time data, to carry out evaluation and analysis, and finally form the safety status assessment of the hydraulic metal structure equipment.

[0018] As a preferred technical solution of the present invention: The risk accident prediction is aimed at the digital twin of the hydraulic metal structure in the information space. Relying on the artificial neural network technology, the input parameters are the environmental data in the environmental model and the online monitoring data of the metal structure equipment in the physical model. Based on the working condition history and the fault history, multi-scenario training is carried out for different gate openings, different opening and closing forces, different vibration types, different structural stress states, different structural corrosion degrees, and different component actions to form a fault probability model. And according to the real-time data, twin data is formed and fused with the fault probability model to obtain the probability prediction of the risk accident.

[0019] The above safety status assessment and risk accident prediction rely on the artificial neural network technology, and the input parameters are the environmental data in the environmental model and the online monitoring data of the metal structure equipment in the physical model.

[0020] In the implementation process of the above solution, first, for the selected hydraulic metal structure equipment, real-time data is obtained; and the obtained parameters are used as input parameters to generate digital twin simulation data; and the environmental model, geometric model, physical model, and rule model are constructed by comparing with the physical space. After the model is formed, twin data is used for training, and then data analysis is carried out, including historical fault analysis, safety status assessment, and risk accident prediction; based on the data analysis results, the rule model is compared to perform model correction and behavior control; among them, the behavior control includes fault warning, feedback control, fault operation, shutdown error reporting, and information back transmission.

[0021] Based on the risk accident prediction, an early warning of the fault is formed, the deteriorating unit is identified and located, and through active behavior control, the fault risk is eliminated in time; in particular, for the gate structure, through the "water level - opening - vibration" analysis, a resonance avoidance motion interval is formed to carry out safe operation; for the hoist structure, through the "water level - opening and closing force" analysis, an equivalent hoist capacity prediction is formed to avoid insufficient opening and closing or overload.

[0022] After the analysis by the artificial neural network technology, accurate fault point positioning, condition-based maintenance, and maintenance reminder can be formed on the data twin physical model; and visual accurate scheduling can be realized through comprehensive monitoring and twin scenarios.

[0023] As a preferred technical solution of the present invention: The risk accident prediction is based on the multi-dimensional and multi-parameter criterion, where the multi-dimension is reflected in the "project - structure / equipment - parameter" three-level pyramid model. The top layer is the "project" safety dimension, the middle layer is the "structure / equipment" safety dimension, and the bottom layer is the "parameter" safety dimension. The overall safety degree of the "project" can be expressed as:

[0024] Γ=F{f 1 ,f 2 ,f 3 ,f 4 ,f5}

[0025] where f i (i = 1, 2, 3, 4, 5) represents the "structure / equipment" in the middle layer of the pyramid, i.e., including gates, hoists, valves, steel pipes, and ship lifts. Among them, the overall safety of the top-layer "project" is related to the weight assignment priority of the middle-layer "structure / equipment". The "structure / equipment" with high-frequency dynamics has a higher priority than the "structure / equipment" with low-frequency static state, that is, the gates and hoists have the weight assignment priority. The overall safety of the "project" after weight assignment can be calculated by the following formula:

[0026]

[0027] where p i (i = 1, 2, 3, 4, 5) represents the weight assignment of each "structure / equipment" and satisfies:

[0028]

[0029] As a preferred technical solution of the present invention: The multi-parameters are embodied as including "determined parameters" and "undetermined parameters" in the bottom-layer "parameters". Therefore, the safety of the middle-layer "structure / equipment" can be expressed as:

[0030] p i = P{Q 1 , Q 2}

[0031] where Q 1 represents the "determined parameters", that is, including the immediate determined parameter information obtained based on monitoring, detection, observation, etc., while Q 2 represents the "undetermined parameters", that is, including the evaluation values, reliability, probability values, etc. of non-immediate determined parameter information obtained based on simulation, training, judgment, etc.

[0032] The bottom-layer "parameters" have differences in weight assignment priority, that is, "parameters directly related to safety" > "parameters indirectly related to safety" > "parameters potentially related to safety". The weight assignment is based on the results of structural safety analysis, historical safety data traceability, and artificial neural network analysis.

[0033] Therefore, the safety of the middle-layer "structure / equipment" can be further expressed as

[0034] p i = P{Q i,j} = P{Q 1,1 , Q 1,2 , Q 1,3 , Q 2,1 , Q 2,2 , Q 2,3}

[0035] i = 1, 2; j = 1, 2, 3

[0036] Where Q i,j in which, i represents "determined parameter" or "undetermined parameter", and j represents "safety directly related parameter" or "safety indirectly related parameter" or "safety potentially related parameter".

[0037] Furthermore, the safety degree of the "structure / equipment" in the middle layer after weight assignment can be calculated by the following formula:

[0038]

[0039] Where, q i,j (i = 1, 2; j = 1, 2, 3) represents the weight assignment of each "parameter", and satisfies:

[0040]

[0041] Since the "safety directly related parameter" or "safety indirectly related parameter" or "safety potentially related parameter" is not limited to 1 item, thus r i,j represents the weight assignment of a certain "safety directly related parameter" or "safety indirectly related parameter" or "safety potentially related parameter", that is

[0042]

[0043] Where, S i (i = 1, 2,..., n) represents a certain "safety directly related parameter" or "safety indirectly related parameter" or "safety potentially related parameter", s i (i = 1, 2,..., n) represents the parameter value, t i (i = 1, 2,..., n) represents the weight assignment of this parameter, and satisfies:

[0044]

[0045] As a preferred technical solution of the present invention: in the risk accident prediction based on the multi - dimensional and multi - parameter criterion, the "parameters" with higher priority include the water discharge state, the concrete hollowing state, the deformation of the gate body, the stress of the gate, the deformation of the hinge, the failure of the main rail, and the remaining thickness of the gate erosion.

[0046] As a preferred technical solution of the present invention: the safety assessment based on digital twin follows three processing modes, namely: real - time data judgment, margin analysis and simulation based on data, and safety assessment based on neural network model.

[0047] The above-mentioned real-time data judgment is made by setting standard parameter ranges for environmental parameters, gate status parameters, hoist status parameters and other data, and comparing the real-time data obtained by monitoring with the standard parameters to carry out safety assessment.

[0048] The above-mentioned data-based margin analysis and simulation uses historical data and real-time data to correct theoretical calculation results and optimize finite element calculation models, conduct online calculations under design conditions, and carry out action simulations under predetermined conditions to examine the margins of parameters including motion range, structural stress, deformation, and hoist capacity.

[0049] The above safety assessment based on the neural network model is implemented by modeling in Matlab mathematical software through the following steps.

[0050] Step 1: Determine the basic data type. Set environmental parameters, gate opening, gate flow, opening and closing force, structural stress, structural deformation, etc. as input data, and gate flow, structural stress, structural deformation, etc. as output data. Based on the basic data, perform structural strength analysis and scheduling analysis.

[0051] Step 2: Normalize the data. After the input data is processed by the algorithm, it is limited to the range of 0 to 1 or -1 to 1. The dimensional expression is converted into a dimensionless expression to facilitate the comparison and weighting of indicators of different units or magnitudes.

[0052] Step 3: Algorithm training based on BP (Back Propagation) neural network.

[0053] The second purpose of the present invention is to provide an intelligent management system for hydraulic metal structure equipment based on digital twins to address the deficiencies in the prior art.

[0054] To this end, the above-mentioned purpose of the present invention is achieved through the following technical solutions:

[0055] An intelligent management system for hydraulic metal structure equipment based on digital twins, the system comprising a basic equipment layer, a perception layer, a data resource layer, a system application layer and a user display layer;

[0056] The perception layer includes intelligent sensing, intelligent control and measurement and control integrated devices;

[0057] The data resource layer is derived from manufacturer data, monitoring data, historical data and operation and maintenance data;

[0058] The system application layer includes a digital archive module, a status monitoring module, an operation and maintenance module and a health assessment module.

[0059] While adopting the above technical solutions, the present invention may also adopt or combine the following technical solutions:

[0060] As a preferred technical solution of the present invention: the digital archive module incorporates management systems, norms and standards, adds 3D models of common metal structure equipment, standard libraries, and disassembly and assembly animations of structure equipment, integrates equipment ledgers, offline data, and online monitoring data, and integrates regular inspection and testing reports and status assessment reports, and recommends common fault types, operation suggestions, and fault handling solutions.

[0061] As a preferred technical solution of the present invention: the status monitoring module includes an environmental parameter monitoring unit, a gate monitoring unit, a hoist monitoring unit, a steel pipe monitoring unit, a trash rack detection unit, and a unit volute monitoring unit.

[0062] As a preferred technical solution of the present invention: the operation and maintenance module includes an electronic sand table, video monitoring, inspection items, inspection plans, task allocation, inspection statistics, and inspection report functions.

[0063] As a preferred technical solution of the present invention: the health assessment module includes off-line and on-line data analysis, historical fault analysis, safety status assessment, and wind accident risk prediction functions.

[0064] In addition, the present invention also provides a smart management operation process for hydraulic metal structure equipment relying on the above-mentioned smart management system for hydraulic metal structure equipment, which is characterized by including the following steps:

[0065] S1. Set monitoring sensors and equipment for different metal structure equipment;

[0066] S2. Set the sampling period and frequency of the collector through the industrial control computer, and automatically store the collected data in the digital archive module;

[0067] S3. Through the status monitoring module, the monitoring results can be observed and reviewed online on the safety information management platform;

[0068] S4. The off-line inspection data and results are manually input into the digital archive module through the digital archive module;

[0069] S5. Through the operation and maintenance module, select specific metal structure equipment and retrieve the status and parameter information for the corresponding period;

[0070] S6. Through the health assessment module, select specific metal structure equipment, and after data analysis and evaluation, infer the type and probability of sudden risk accidents, and export the corresponding conclusion report;

[0071] S7. Based on the operation and maintenance module and the health assessment module, formulate an inspection and maintenance plan and arrange personnel division of labor.

[0072] The present invention can at least include the following beneficial effects:

[0073] Based on precise sensing and with the help of data mining, through data twin, a systematic safety management plan has been developed for hydraulic metal structure equipment. On the one hand, for the numerous existing data accumulated over more than a decade, such as historical water levels, opening and closing forces, gate openings, maintenance implementation plans, and fault handling plans, systematic analysis and management are carried out for the purpose of operation and maintenance. On the other hand, for data types that are hidden in location but related to safety, monitoring plans are supplemented and customized, and the safety management analysis method combining existing data and real-time data is further improved. Specifically, (1) a safety informatization management plan for hydraulic metal structure equipment in a three-level pyramid model of "project - structure / equipment - parameter" has been formulated, clearly defining the target "structure / equipment" levels including gates, hoists, valves, steel pipes, and ship lifts, and dividing the immediate "determined parameters" obtained through monitoring, inspection, observation, etc. and non-immediate "uncertain parameters" such as evaluation values, reliability, and probability values obtained through simulation, training, judgment, etc., to improve safety management and evaluation indicators. (2) A digital twin body based on environmental data and the state data of metal structure equipment has been formed, that is, according to the "parameter" type selected by the user, a twin model of the linkage of key environmental hydrological parameters, equipment state parameters, and structural motion postures is constructed, and combined with the historical gate opening and closing conditions, equipment failure history, and real-time data of water retaining scheduling, accurate fault point positioning during the gate opening and closing process, evaluation of the opening and closing force state of the hoist, engineering safety inspection and maintenance reminder are carried out visually. (3) A precise operation and regulation prediction based on digital twin has been formed. For the gate structure, through the analysis of "water level - opening - vibration", the gate movement process is simulated to form a resonance avoidance movement interval to guide safe operation; for the hoist structure, through the analysis of "water level - opening and closing force" trained by neural network, the equivalent hoist capacity prediction is formed to avoid insufficient or overload opening and closing. Description of the Drawings

[0074] Figure 1 It is the digital twin relationship between the hydraulic metal structure equipment provided by the present invention and its surrounding environment.

[0075] Figure 2 It is the construction form of the information space model provided by the present invention.

[0076] Figure 3 It is the system operation mode provided by the present invention.

[0077] Figure 4 It is the framework composition provided by the present invention.

[0078] Figure 5 It is the composition diagram of the system application layer provided by the present invention.

[0079] Figure 6The composition diagram of the health assessment module provided by the present invention.

[0080] Figure 7 The digital twin model based on artificial neural network technology provided by the present invention.

[0081] Figure 8 The intelligent management flow chart of hydraulic metal structure equipment provided by the present invention.

[0082] Figure 9 The schematic diagram of the metal structure equipment management system of a certain power station.

[0083] Figures 10a - 10c They are respectively the schematic diagrams of the state monitoring, health assessment and operation and maintenance functions in the metal structure equipment management system of a certain power station. Specific implementation manners

[0084] The following further detailed description of the present invention is made with reference to the accompanying drawings, so that those skilled in the art can implement it according to the description in the specification.

[0085] It should be noted that the experimental methods described in the following implementation schemes are all conventional methods unless otherwise specified, and the reagents and materials can be obtained from commercial channels unless otherwise specified; in the description of the present invention, the orientation or positional relationship indicated by the terms is based on the orientation or positional relationship shown in the accompanying drawings, which is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be construed as a limitation to the present invention.

[0086] As Figure 1 shown, it is the digital twin relationship between the hydraulic metal structure equipment and the surrounding environment, mainly including constructing a physical space containing the physical environment of the water conservancy project and the physical entity of the hydraulic metal structure, and at the same time constructing an information space containing the twin environment of the water conservancy project and the digital twin of the hydraulic metal structure. Among them, the physical space and the information space interact through twin data including historical data, online data, and prediction data.

[0087] In order to complete the digital twin, it is necessary to construct an information space, and construct the environmental model of the hydraulic metal structure equipment, the geometric model of the hydraulic metal structure equipment, the physical model of the hydraulic metal structure equipment, and the rule model of the hydraulic metal structure equipment management system, as Figure 2As shown below. First, an environmental model of the hydraulic metal structure equipment is constructed, including parameters such as water level, rainfall, temperature, humidity, pH value, medium flow rate, and solution conductivity. Further, a geometric model of the hydraulic metal structure equipment is constructed, which needs to meet the requirements of geometric dimensions, constituent materials, combined structures, and assembly logic. Further, a physical model of the hydraulic metal structure equipment is constructed, mainly including gates, hoisting machines, steel pipes, and unit volutes. The parameters that need to be designed for the construction of the gate physical model include gate opening, gate inclination, gate vibration, hinge shaft torque, stress and strain, coating status, corrosion depth, bolt torque, hinge shaft operating status, and expansion joint operating status, etc. The parameters that need to be designed for the construction of the hoisting machine physical model include hoisting force, hoisting machine vibration, hydraulic pressure, stress and strain, acceleration, bolt torque, wire rope status, coating status, and corrosion status, etc. The parameters that need to be designed for the construction of the steel pipe physical model include steel pipe vibration, coating status, corrosion depth, erosion degree, and flow velocity inside the pipe, etc. The parameters that need to be designed for the construction of the unit volute physical model include unit vibration, stress and strain, and bolt torque, etc. Further, the construction of the rule model of the hydraulic metal structure equipment management system needs to be based on management systems, specification standards, standard libraries, expert experience, and intelligent comparison, etc.

[0088] In order to realize the intelligent management of the hydraulic metal structure equipment, after the physical environment of the water conservancy project and the physical entity of the hydraulic metal structure are formed, real-time information is collected and compared with the output information in the information space immediately. When the comparison result does not meet the rule model, data analysis is carried out. On the one hand, when the analysis result is that there are model defects in the information space, the model is corrected; on the other hand, when the analysis result is that the physical entity of the hydraulic metal structure is abnormal, behavior control is carried out. Among them, data analysis includes historical fault analysis, safety status assessment, and risk accident prediction; behavior control includes fault warning, fault operation, shutdown error reporting, feedback control, and information back transmission.

[0089] Specifically, for the safety status assessment, its steps include excavating and deriving the fault types and causes of the hydraulic metal structure equipment based on the working condition history and fault history, and combining with real-time data for evaluation and analysis, and finally forming the safety status assessment of the hydraulic metal structure equipment; for the risk accident prediction, its steps include carrying out multi-scenario training for different gate openings, different hoisting forces, different vibration types, different structural stress states, different structural corrosion degrees, and different component actions for the digital twin of the hydraulic metal structure in the information space based on the working condition history and fault history, and then forming a fault probability model, and fusing the twin data formed according to the real-time data with the fault probability model to obtain the probability prediction of the risk accident.

[0090] A system operation mode of the above technical solution is as Figure 3As shown in the figure, first, for the selected hydraulic metal structure equipment, real-time data is obtained; and the obtained parameters are used as input parameters to generate digital twin simulation data; and the environmental model, geometric model, physical model, and rule model are constructed by referring to the physical space. After the model is formed, the twin data is used for training, and then data analysis is carried out, including historical fault analysis, safety status assessment, and risk accident prediction; based on the data analysis results, the rule model is compared to perform model correction and behavior control; among them, the behavior control includes fault warning, feedback control, fault operation, shutdown error reporting, and information back transmission. Further, based on the risk accident prediction, an early fault warning is formed, the deteriorated unit is identified and located, and the fault risk is eliminated in time through active behavior control.

[0091] The safety assessment based on digital twin follows three processing modes, namely: instant data judgment, margin analysis and simulation based on data, and safety assessment based on neural network model.

[0092] The above-mentioned instant data judgment is to set the standard parameter range for data such as environmental parameters, gate state parameters, and hoist state parameters, and compare the monitored instant data with the standard parameters to carry out safety assessment.

[0093] The above-mentioned margin analysis and simulation based on data is to use historical data and instant data to correct the theoretical calculation results and optimize the finite element calculation model, perform online calculation under design conditions and carry out action simulation under predetermined conditions to examine the margins of parameters such as movement range, structural stress, deformation, and hoist capacity.

[0094] The above-mentioned safety assessment based on neural network model is modeled and implemented in Matlab mathematical software through the following steps.

[0095] The first step: determination of the basic data type. Set environmental parameters, gate opening, flow rate through the gate, hoisting force, structural stress, structural deformation, etc. as input data, and set flow rate through the gate, structural stress, structural deformation, etc. as output data. Based on the basic data, structural strength analysis and scheduling analysis are carried out.

[0096] The second step: data normalization. After the input data is processed by the algorithm, it is restricted within the range of 0 to 1 or -1 to 1. The dimensional expression is changed into a dimensionless expression, which is convenient for comparing and weighting indicators with different units or magnitudes. The normalization process is as follows:

[0097] X scale =(X - X min ) / (X max - X min )

[0098] where X scaleis the normalized data, X is the input data, X max and X min are the maximum and minimum values in the input data respectively.

[0099] Step 3: Training based on the BP (Back Propagation) neural network algorithm. The first layer of the BP neural network is the input layer (x 1 , x 2 , … x i ), including the immediate data and calculated data of the input; the middle layer is one or more hidden layers, and the number of layers is adjusted according to the training results; the last is the output layer (y 1 , y 2 , … y i ), including data such as the monitored flow rate and structural strength of the output. Each layer is composed of several neurons (equivalent to each hidden layer). The output value of each layer is determined by the input value, activation function, connection weight w i and threshold θ. After determining the algebraic sum of the products of all input quantities x i and weight w i , the final output value y i is equal to the difference between the above algebraic sum and the possible value θ.

[0100] The final output data needs to be processed by an activation function such as sigmoid, and the processing process is as follows:

[0101]

[0102] Input the data into matlab for neural network training, and the code used is as follows:

[0103] net = newff(minmax(Y scale ), X scale , [a, b], {'logsig', 'purelin'}, 'traingdm'), where X scale and Y scale represent the normalized input and output data respectively, a represents the number of neurons in the hidden layer, b represents the number of neurons in the output layer, 'logsig' is the transfer function of the hidden layer of the neural network, and 'purelin' function is the transfer function of the output layer.

[0104] After the neural network training, the input data should be de-normalized to ensure that the scalar is re-converted into a dimensional number in subsequent predictions to avoid deviation in the results. The processing code used is as follows:

[0105] Y re-scale = mapminmax('reverse', Y train,PS), where Y train is the output value of the trained neural network, Y re-scale is the data after denormalization, and PS is the setting parameter.

[0106] Step 4: Security assessment based on the trained BP neural network. The implementation process is as follows:

[0107] pnew=[a1…an], where a1 to an are actual input data;

[0108] pnewn=mapminmax(pnew), wherein this step represents normalizing the data;

[0109] anewn=sim(net,pnewn), where this step means using the trained BP neural network;

[0110] anew=mapminmax('reverse',anewn,ts), where this step indicates that the result is denormalized and outputted, and the errors between numbers can be compared.

[0111] A system framework for intelligent management of hydraulic metal structure equipment based on digital twins Figure 4 As shown in the figure, its framework includes basic equipment layer, perception layer, data resource layer, system application layer and user display layer. It can be used as an integrated, modular and comprehensive security monitoring platform for hydraulic metal structure equipment, mainly used for data management, status diagnosis and risk assessment. The intelligent management system for hydraulic metal structure equipment relies on and serves hydraulic metal structure equipment. Common system objects include gates, hoists, water diversion steel pipes / steel bifurcated pipes, and unit volutes. Sensors are arranged in areas that need to be paid attention to on hydraulic metal structure equipment to collect and store data, and manage the data obtained from online monitoring of the perception layer. Among them, the data resource layer brings together the monitoring results, offline inspection results, data analysis results and status assessment reports of all monitoring sensors; the user display layer can display applications on computer terminals, mobile terminals, digital large screens and other scenarios according to functional requirements.

[0112] In particular, the system application layer has all the functions of intelligent management, such as Figure 5 As shown in the figure, it mainly includes digital archive module, status monitoring module, operation and maintenance module and health assessment module, while the user display layer needs to select one or part of the functions for design and use based on user characteristics and scenario requirements. For example, for operation and maintenance personnel, only the operation and maintenance module needs to be enabled, and for background analysts, only the digital archive module needs to be enabled.

[0113] In this embodiment, the digital archive module in the intelligent management system for hydraulic metal structure equipment functions as follows: Figure 5As shown in the figure, it includes the following: management system, specification standards, 3D models, standard picture libraries, disassembly and assembly animations, equipment ledgers, offline data, online data, regular inspection reports, common faults, operation suggestions, evaluation reports, and fault handling suggestions. Among them, the management system and specification standards are preset according to the requirements of the unit on which the intelligent management system for hydraulic metal structure equipment relies; the 3D models and standard picture libraries can display the basic structural equipment information of gates, hoisting machines, penstocks / steel bifurcation pipes, and unit spiral casings. For key moving parts, the internal structure and movement principle can be understood through the disassembly and assembly animations of the structural equipment. The equipment ledger can select and display the equipment parameters and functions in all systems. The offline data and online monitoring data are respectively entered or stored in the digital archive module after offline inspection and sensor online monitoring; in addition, the regular inspection and monitoring reports can also be stored and retrieved in the fixed format in the module. Based on the specification standards and inspection and monitoring results, the target structural equipment can be selected in the digital archive module for status evaluation or to carry out prediction and measure suggestions.

[0114] In this embodiment, the functions of the operation and maintenance module in the intelligent management system for hydraulic metal structure equipment are as Figure 5 shown in the figure, including the following: electronic sand table, video monitoring, inspection items, inspection plans, task assignment, inspection statistics, and inspection reports. Among them, the electronic sand table serves as a global status display window for the entire hydraulic metal structure equipment. Specifically, it can display, for example, the operating status of the gate, the stress and strain distribution of key parts such as welds and bearings, and the corrosion potential distribution of the metal structure in flooded or humid areas; video monitoring is used for real-time online observation of key monitoring points; the inspection items, inspection plans, and task assignment are parameterized and step-by-step set and implemented in the form of folded items, floating frames, drop-down menus, etc., and preliminary plan suggestions are made based on offline / online inspection and monitoring data, combined with specification standards and expert experience. During the final implementation process, modifications and optimizations can be made according to actual needs in the operation and maintenance module; the inspection statistics and inspection reports can be viewed, compared, and exported item by item. After the inspection is completed, the inspection statistics and inspection reports can be entered through the operation and maintenance module, and at the same time, the inspection statistics and inspection reports will also be stored in the digital archive module.

[0115] In this embodiment, the functions of the health assessment module in the intelligent management system for hydraulic metal structure equipment are as Figure 6 shown in the figure, including: historical fault analysis, offline / online data analysis, safety status assessment, and risk accident prediction.

[0116] Historical fault analysis is to screen the fault types and judge the causes of faults for specific metal structure equipment based on inspection results and off-line / on-line data. Specifically, historical fault analysis is divided into structural deformation, structural instability, structural crack, cable breakage, equipment wear, equipment malfunction, corrosion failure, and structural vibration according to the fault characteristics.

[0117] Off-line / on-line data analysis is to conduct preliminary screening and optimization based on inspection results and sensor on-line monitoring data to form classified data statistical results, which serve as the basis for historical fault analysis, safety status assessment, and accident risk prediction. Specifically, off-line / on-line data analysis is combined with the fault type, and methods such as least squares, time domain analysis, frequency domain analysis, factor analysis, correlation analysis, regression analysis, fuzzy analysis, Fourier transform, and Bayesian analysis are used to organize and refine the data, and provide a basis for subsequent safety status assessment and risk accident prediction.

[0118] Safety status assessment is to select specific metal structure equipment and evaluate the safety status based on the on-line values such as the opening, stress and strain, inclination angle, acceleration, and corrosion rate measured by sensors, combined with the results of off-line regular inspections, based on operating conditions, specifications and standards, analysis models, and expert experience, and give comprehensive health evaluations of different levels.

[0119] Risk accident prediction is to speculate on the types and probabilities of sudden risk accidents by combining reliable risk assessment methods after obtaining the operating status of specific metal structure equipment. Under the current safety status, comprehensive analysis is carried out relying on various prediction models, such as combining grey models, support vector machine analysis, Monte Carlo simulation, and neural network analysis.

[0120] As Figure 7 shown in the figure, twin data analysis data is carried out relying on artificial neural network technology, where the input parameters are the environmental data in the environmental model and the on-line monitoring data of metal structure equipment in the physical model. Based on the operating conditions history and fault history, multi-scenario training of different gate openings, different opening and closing forces, different vibration types, different structural stress states, different structural corrosion degrees, and different component actions is carried out to form a fault probability model, and obtain the remaining life and failure probability of metal structure equipment; and further combining the twin model and data analyzed by artificial neural network technology can form accurate control of the state of metal structure equipment, fault point positioning, condition-based maintenance, and maintenance reminder; and realize visual and accurate scheduling through comprehensive monitoring and twin scenarios.

[0121] In this specific embodiment, for risk accident prediction based on multi-dimensional and multi-parameter criteria, taking the safety evaluation of only the gate in a certain project as an example, "structure / equipment" is uniquely defined as the gate, and its weight for project safety is 1, and it is affected by the weight assignment of the underlying "parameters". In this embodiment, the "parameters" can be divided into the operating conditions of the gate (u1 ), corrosion degree ($u$ 2 ), stress condition ($u$ 3 ), visual inspection of the gate ($u$ 4 ), inspection results ($u$ 5 ), a total of five categories, namely

[0122] $u=(u$ 1 , $u$ 2 , $u$ 3 , $u$ 4 , $u$ 5 )

[0123] Among them, the operation condition of the gate, corrosion degree, and stress condition can be defined as "parameters directly related to safety", the visual inspection of the gate can be defined as "parameters indirectly related to safety", and the inspection results can be defined as "parameters potentially related to safety". According to the priority of weight assignment of parameters, the operation condition of the gate ($u$ 1 ) > corrosion degree ($u$ 2 ) > stress condition ($u$ 3 ) > visual inspection of the gate ($u$ 4 ) > inspection results ($u$ 5 ). Further, the operation condition of the gate can refer to "indeterminate parameters" and be assigned values based on expert evaluation or neural network analysis; the visual inspection of the gate and inspection results can refer to "indeterminate parameters" and be assigned values based on expert evaluation or empirical analysis; while the corrosion degree and stress condition can refer to "determinate parameters" and be assigned values by calculating the remaining strength through structural strength.

[0124] Furthermore, during the process of parameter assignment, some parameters may include lower-level weight distribution factors. For example, the inspection includes the water flow attitude when the gate is discharging water ($u$ 5,1 ), the water leakage condition when the gate is closed ($u$ 5,2 ), the concrete condition of the gate slot ($u$ 5,3 ), the condition of the pier - breast wall - corbel ($u$ 5,4 ), the ventilation hole unobstructed condition ($u$ 5,5 ), etc. A weight vector can be established as follows:

[0125] $u$ 5 =(u$ 5,1 , $u$ 5,2 , $u$ 5,3 , $u$ 5,4 , $u$ 5,5 )

[0126] And further calculate through the weight matrix to obtain the safety degree of the gate:

[0127]

[0128] Based on the above-mentioned intelligent management system for hydraulic metal structure equipment, the following intelligent management and operation process of hydraulic metal structure equipment can be formed, which is characterized by the following steps: Figure 8 As shown in the figure, the intelligent management and operation process of hydraulic metal structure equipment includes the following steps:

[0129] S1. Set monitoring sensors and equipment for different metal structure equipment as objects;

[0130] S2. Set the sampling period and frequency of the collector through the industrial control computer, and automatically store the collected data into the digital archive module;

[0131] S3. Through the status monitoring module, the monitoring results can be observed and reviewed online on the safety information management platform;

[0132] S4. The off-line inspection data and results are manually input into the digital archive module through the digital archive module;

[0133] S5. Through the operation and maintenance module, select specific metal structure equipment and retrieve the status and parameter information of the corresponding period;

[0134] S6. Through the health assessment module, select specific metal structure equipment, and after data analysis and evaluation, infer the types and probabilities of sudden risk accidents, and export the corresponding conclusion report;

[0135] S7. Based on the operation and maintenance module and the health assessment module, formulate an inspection and maintenance plan and arrange personnel division of labor.

[0136] As Figure 9 shown, it is a schematic diagram of the metal structure equipment management system of a certain power station. The monitoring objects are the gates and hoisting machines of the power station. The status monitoring module needs to arrange monitoring components on the gates and hoisting machines. The monitoring components are required to meet the working conditions of long-term underwater operation and are connected to a separate control cabinet in a wired manner. The monitoring module is independent of the hoisting machine control system, and the monitored data is transmitted to the remote monitoring terminal for display through the local area network switch. In this embodiment, the stress state of the equipment during operation is observed through the stress sensors arranged on the structure equipment. As Figure 10a shown, it is the online status monitoring data curve of some stress sensors. The current structural strength of the equipment can be judged by the stress magnitude, and whether there is any abnormality in the equipment can be judged by whether there is a sudden change in the monitoring curve. At the same time, as Figure 10b shown, based on the off-line and online data, the health assessment of the gates and hoisting machines can be carried out to form the assessment results and assessment reports. Based on the monitoring results and assessment results, the management system can also give suggestions for subsequent operation and maintenance, formulate an inspection plan, and perform online operation control with reference to the preset operation instructions. As Figure 10cAs shown, since the evaluation results of some metal structure equipment are "unsafe", the management system retrieves the recent equipment inspection records. At the same time, in the operation and maintenance function module, the subsequent operation plan is formulated.

[0137] The above embodiments are only the preferred technical solutions of the present invention. Those skilled in the art should understand that without departing from the principles and essence of the present invention, the technical solutions or parameters in the embodiments can be modified or replaced, and all should be covered within the protection scope of the present invention.

Claims

1. An intelligent management construction method for hydraulic metal structure equipment based on digital twin, characterized in that: The construction method includes the following steps: S1. Construct a physical space, which includes the physical environment of the water conservancy project and the physical entities of the hydraulic metal structure; the physical entities of the hydraulic metal structure include at least one of gates, hoists, steel pipes, and unit spiral casings; S2. Construct an information space, which includes the twin environment of the water conservancy project and the digital twin of the hydraulic metal structure; In step S2, the construction of the information space includes the construction of an environment model, a geometric model, a physical model, and a rule model; the parameters of the environment model include water level, rainfall, temperature, humidity, pH value, medium flow velocity, and solution conductivity; the geometric model includes geometric dimensions, constituent materials, combined structures, and assembly logic; The construction of the rule model is based on management systems, specifications and standards, standard libraries, expert experience, and intelligent comparison; S3. Generate twin data for the interaction between the physical space and the information space, and the twin data includes historical data, online data, and prediction data; In step S3, the real-time information formed by the physical environment of the water conservancy project and the physical entities of the hydraulic metal structure is immediately compared with the output information of the information space; When the comparison result does not meet the rule model, data analysis is carried out. On the one hand, when the analysis result is that there are model defects in the information space, the model is corrected. On the other hand, when the analysis result is that the physical entity of the hydraulic metal structure is abnormal, behavior control is carried out; The data analysis includes historical fault analysis, safety status assessment, and risk accident prediction; the behavior control includes fault warning, fault operation, shutdown error reporting, feedback control, and information back transmission; The safety assessment based on digital twin follows three processing modes, namely: real-time data judgment, margin analysis and simulation based on data, and safety assessment based on neural network model; among them, the training of the neural network model is based on working condition history, fault history, and real-time data; The risk accident prediction is aimed at the digital twin of the hydraulic metal structure in the information space. Relying on artificial neural network technology, the input parameters are the environmental data in the environment model and the online monitoring data of the metal structure equipment in the physical model. Based on the working condition history and fault history, multi-scenario training of different gate openings, different hoisting forces, different vibration types, different structural stress states, different structural corrosion degrees, and different component actions is carried out to form a fault probability model, and twin data is formed based on real-time data and fused with the fault probability model to obtain the probability prediction of risk accidents; The risk accident prediction is based on a multi-dimensional and multi-parameter criterion. The multi-dimension is reflected in the "project-structure / equipment-parameter" three-level pyramid model, where "structure / equipment" includes gates, hoists, valves, steel pipes, and ship lifts; the multi-parameter is reflected in that "parameters" include "determined parameters" and "undetermined parameters". "Determined parameters" include immediate determined parameter information obtained based on monitoring, detection, and observation, and "undetermined parameters" include non-immediate determined parameter information such as evaluation values, reliability, and probability values obtained based on simulation, training, and judgment; For the risk accident prediction based on the multi - dimensional and multi - parameter criterion, the underlying "parameters" have different priority levels for weight assignment. "Safety directly - related parameters" are greater than "safety indirectly - related parameters" which are greater than "safety potentially - related parameters". The weight assignment is based on the results of structural safety analysis, historical safety data traceability, and artificial neural network analysis; in the middle - layer "structure / equipment", the gates and hoisting machines have priority levels for weight assignment, and the high - frequency dynamic priority is higher than the low - frequency static priority. For the risk accident prediction based on the multi - dimensional and multi - parameter criterion, the "parameters" with higher priorities include the water discharge state, the concrete hollowing state, the deformation of the gate body, the deformation of the support hinge, the failure of the main rail, and the remaining thickness of the gate erosion.

2. An intelligent management system for hydraulic metal structure equipment based on digital twin It is characterized in that The system is based on the construction method of the intelligent management of hydraulic metal structure equipment based on digital twin as described in claim 1, and includes a basic equipment layer, a perception layer, a data resource layer, a system application layer, and a user display layer; The perception layer includes intelligent sensing, intelligent control, and measurement and control integrated devices; The data resource layer comes from manufacturer data, monitoring data, historical data, and operation and maintenance data; The system application layer includes a digital archive module, a status monitoring module, an operation and maintenance module, and a health assessment module.

Images