Method and device for determining dismounting and mounting layout strategy of flowmeter calibration platform position, and medium

Abstract

The application discloses a kind of flowmeter verification station position dismounting layout strategy determination method, device and medium.The method comprises: obtaining station position operation associated data;The station position operation associated data is input into the deep reinforcement learning model pre-trained, the deep reinforcement learning model is based on feature extraction and multi-head attention weight distribution, and the flowmeter verification station position dismounting layout decision sequence that meets the preset process constraint is output;Wherein, preset process constraint includes station position total length constraint, flowmeter type adaptation constraint and flowmeter spacing constraint;According to the dismounting layout decision sequence, generate visual layout scheme.The application improves the efficiency, compliance and optimization level of flowmeter verification station position dismounting layout.

Description

Technical Field

[0001] This invention relates to the field of optimizing the disassembly and assembly layout strategy of flowmeter calibration stations, and particularly to a method, apparatus and medium for determining the disassembly and assembly layout strategy of flowmeter calibration stations. Background Technology

[0002] With the rapid development of industrial testing technology, flowmeter calibration is increasingly widely used in fields such as petrochemicals and natural gas transportation. During flowmeter calibration, it is necessary to rationally arrange straight pipe sections and measuring pipes on a finite-length calibration platform according to the specifications and installation requirements of different types of flowmeters, in order to meet complex process constraints such as the length of upstream and downstream straight pipe sections, rectifier installation location, and flowmeter spacing. For a long time, the optimization of the disassembly and assembly layout strategy for flowmeter calibration platforms has mainly relied on manual experience-based decisions. This method suffers from problems such as low layout efficiency, difficulty in considering global optimality, easy violation of constraints, and inability to dynamically adapt to various flowmeter combinations.

[0003] In recent years, deep reinforcement learning methods have been rarely used in the optimization of flowmeter calibration station layout strategies in China. However, combining deep reinforcement learning algorithms with sequential decision optimization can achieve a globally optimal layout through multi-step decision-making, improving the automation level and optimization efficiency of the layout. Existing flowmeter calibration station layout methods suffer from limitations such as limited optimization effects, difficulty in meeting complex constraints, decision-making limited to local optima, and lack of adaptive learning capabilities, making it difficult to maintain efficient and stable layout performance in dynamically changing calibration tasks. Summary of the Invention

[0004] This invention provides a method, device, and medium for determining the disassembly and assembly layout strategy of flowmeter calibration stations, so as to realize the intuitive display of the layout results and improve the efficiency, compliance, optimization level, and long-term operational stability of the disassembly and assembly layout of flowmeter calibration stations.

[0005] According to one aspect of the present invention, a method for determining the disassembly and assembly layout strategy of a flowmeter calibration station is provided, the method comprising: Obtain the associated data for station operation; The associated data of the test station operation is input into a pre-trained deep reinforcement learning model. The deep reinforcement learning model outputs a decision sequence for the disassembly and assembly layout of the flow meter verification station that satisfies preset process constraints based on feature extraction and multi-head attention weight allocation. The preset process constraints include total station length constraints, flow meter type adaptation constraints, and flow meter spacing constraints. A visual layout scheme is generated based on the disassembly and assembly layout decision sequence.

[0006] According to another aspect of the present invention, an apparatus for determining the disassembly and assembly layout strategy of a flowmeter calibration station is provided, the apparatus comprising: The data acquisition module is used to acquire data related to the operation of the workstations. The decision sequence determination module is used to input the station operation association data into a pre-trained deep reinforcement learning model. The deep reinforcement learning model outputs a decision sequence for the disassembly and assembly layout of the flow meter verification station that satisfies preset process constraints based on feature extraction and multi-head attention weight allocation. The preset process constraints include total station length constraints, flow meter type adaptation constraints, and flow meter spacing constraints. The layout scheme generation module is used to generate a visual layout scheme based on the disassembly and assembly layout decision sequence.

[0007] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising: At least one processor; and memory that is communicatively connected to at least one processor; The memory stores a computer program that can be executed by at least one processor, which is then executed by the at least one processor to enable the at least one processor to execute the method for determining the disassembly and assembly layout strategy of the flowmeter calibration station according to any embodiment of the present invention.

[0008] According to another aspect of the present invention, a computer-readable storage medium is provided, which stores computer instructions for causing a processor to execute a method for determining the disassembly and assembly layout strategy of a flowmeter calibration station according to any embodiment of the present invention.

[0009] The technical solution of this invention solves the problems of fragmented and incomplete data collection in traditional manual layout by acquiring station operation-related data, thus achieving completeness of the data foundation for layout decisions. By inputting data into a pre-trained deep reinforcement learning model and outputting a compliant decision sequence, it solves the problems of low efficiency, easy violation of process constraints, difficulty in achieving global optimization, and lack of adaptive learning ability in manual layout, thus achieving automation, intelligence, global optimization, and long-term stability of layout decisions. By generating a visual layout scheme, it solves the problems of abstract and difficult-to-understand traditional layout schemes and poor readability, thus achieving intuitive display of layout results. Ultimately, it improves the efficiency, compliance, optimization level, and long-term operational stability of flowmeter calibration station disassembly and assembly layout.

[0010] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0011] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 A flowchart illustrating a method for determining the disassembly and assembly layout strategy of a flowmeter calibration station, as provided in an embodiment of the present invention; Figure 2a A flowchart illustrating another method for determining the disassembly and assembly layout strategy of a flowmeter calibration station, as provided in an embodiment of the present invention; Figure 2b This is a schematic diagram of the system functional structure of a method for determining the disassembly and assembly layout strategy of a flowmeter calibration station according to an embodiment of the present invention; Figure 3 A schematic diagram of the structure of a device for determining the disassembly and assembly layout strategy of a flowmeter calibration station provided in an embodiment of the present invention; Figure 4 A schematic diagram of the structure of an electronic device for implementing a method for determining the disassembly and assembly layout strategy of a flowmeter calibration station according to an embodiment of the present invention. Detailed Implementation

[0013] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0014] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0015] Figure 1This is a flowchart illustrating a method for determining the disassembly and assembly layout strategy of a flowmeter calibration station according to an embodiment of the present invention. This embodiment is applicable to the optimization of the disassembly and assembly layout strategy of a flowmeter calibration station. This method can be executed by a device for determining the disassembly and assembly layout strategy of the flowmeter calibration station. This device can be implemented in hardware and / or software and can be configured in an electronic device. Figure 1 As shown, the method specifically includes the following steps: S110, Obtain the associated data of the station operation.

[0016] Among them, the station operation-related data can be understood as the complete set of basic data required to support the decision-making on the disassembly and layout of the flow meter calibration station.

[0017] Specifically, comprehensive data related to the operation and layout of the flow meter calibration station is collected, including key information on core components such as the station, flow meter, pipelines, and expansion joints. This provides a complete and reliable data foundation for subsequent model input, preventing decision-making biases due to missing or incomplete data and ensuring the accuracy and comprehensiveness of subsequent intelligent decisions.

[0018] Optionally, the station operation associated data includes flow meter library data, pipeline inventory data, current station layout status data, and expansion joint parameter data.

[0019] Among them, the flow meter library data can be understood as a dataset storing information such as the model, size, type, parameters, and applicable operating conditions of various flow meters; the pipeline inventory data can be understood as a dataset recording information such as the length, specifications, quantity, material, and applicable range of existing pipelines; the current station layout status data can be understood as real-time status information such as the installation location of existing equipment at the station, the space occupied, the remaining available space, and the historical layout; and the expansion joint parameter data can be understood as key parameter information such as the expansion range, length limit, specifications, and adjustment accuracy of the expansion joint.

[0020] Specifically, flow meter database data provides accurate basis for flow meter selection and ensures selection suitability; pipeline inventory data ensures that pipeline combinations fit existing inventory and avoids resource waste; current station layout status data ensures that the layout fits the real-time operating conditions and historical layout patterns of the station; expansion joint parameter data is used to accurately calculate station length constraints. The four types of data work together and comprehensively cover the dimensions required for layout decision-making, providing detailed, accurate and complete data support for subsequent model input.

[0021] S120. Input the associated data of the station operation into a pre-trained deep reinforcement learning model. The deep reinforcement learning model outputs a decision sequence for the disassembly and assembly layout of the flow meter calibration station that meets the preset process constraints, based on feature extraction and multi-head attention weight allocation.

[0022] The preset process constraints include total station length constraints, flowmeter type compatibility constraints, and flowmeter spacing constraints. The deep reinforcement learning model can be understood as an intelligent model that integrates deep learning feature extraction capabilities with reinforcement learning decision optimization capabilities. After training with historical data, it can automatically output the optimal layout decision. Feature extraction can be understood as the process of filtering and refining key and effective information about station layout from raw data, eliminating redundant information. Multi-head attention weight allocation can be understood as a mechanism that assigns differentiated weights to extracted key features, prioritizing the processing of core process constraint information. Preset process constraints can be understood as the rigid process conditions that ensure the compliance, safety, and accuracy of flowmeter verification. The disassembly and assembly layout decision sequence can be understood as an ordered set of multi-step layout operation instructions, including flowmeter selection, pipeline combination, and equipment installation sequence.

[0023] Specifically, the acquired station operation correlation data is first imported into a pre-trained deep reinforcement learning model. Inside the model, feature extraction is performed to extract deep key features such as station length, flowmeter parameters, pipe specifications, and equipment spacing from the station operation correlation data. Then, through a multi-head attention weight allocation mechanism, differentiated weights are assigned to key features, focusing on three core constraint features: total station length, flowmeter type compatibility, and flowmeter spacing. Based on the weighted features, the model generates probability distributions for flowmeter selection and pipe combination, conducts multi-step sequential decision-making, rigorously verifies whether all preset process constraints are met, screens compliant solutions, and finally outputs a complete and feasible disassembly and assembly layout decision sequence, achieving automated and intelligent global optimal decision-making under complex constraints.

[0024] Optionally, the total length constraint of the station includes the total length of the flowmeter, straight pipe section, measuring pipe and expansion joint within the station area not exceeding a preset maximum length limit; the flowmeter type adaptation constraint includes distinguishing between rectifier-equipped and rectifier-free operating conditions for gas ultrasonic flowmeters, ensuring that the upstream and downstream straight pipe section lengths meet the corresponding process coefficient requirements, and that the upstream and downstream straight pipe section lengths for gas turbine flowmeters, mass flowmeters and rotary flowmeters meet the corresponding independent process coefficient requirements; the flowmeter spacing constraint includes the center distance between adjacent flowmeters not being less than a preset minimum spacing threshold.

[0025] In this context, a straight pipe section can be understood as a standard straight pipe section with no bends, branches, or obstructions upstream and downstream of the flowmeter; a measuring pipe can be understood as a dedicated pipe used for flowmeter calibration that meets calibration standards; a nominal diameter can be understood as the standard nominal diameter of the flowmeter and the pipe; a process coefficient can be understood as a fixed value preset according to the flowmeter type and calibration process requirements, used to calculate the length of the upstream and downstream straight pipe sections; and a minimum spacing threshold can be understood as the minimum center-to-center distance between adjacent flowmeters to ensure normal operation of the flowmeter and that the calibration accuracy is not disturbed. The preset minimum spacing threshold can be pre-set based on experience, and this embodiment does not impose specific restrictions on it.

[0026] Specifically, the total length constraint of the test site requires that the total length of the flowmeter, straight pipe section, measuring pipe, and expansion joint within the test site area strictly not exceed the preset maximum length limit to prevent equipment from exceeding the test site installation range and ensure layout feasibility; the flowmeter type adaptation constraint distinguishes between two operating conditions of gas ultrasonic flowmeters with and without rectifiers, corresponding to different upstream and downstream straight pipe section process coefficients, and also clarifies the independent process coefficient requirements for gas turbine flowmeters, mass flowmeters, and rotary flowmeters to ensure that the calibration accuracy of different types of flowmeters meets the standards; the flowmeter spacing constraint limits the center distance between adjacent flowmeters to not less than the preset minimum spacing threshold to prevent mutual interference between equipment and affect the accuracy of calibration results. These three types of constraints comprehensively cover the core requirements for layout compliance, ensuring that the decision sequence fully complies with process standards and calibration requirements.

[0027] Optionally, the step of inputting the platform operation association data into a pre-trained deep reinforcement learning model includes: cleaning the platform operation association data to remove duplicate and abnormal data; converting character features in the cleaned data into numerical features to obtain standardized data; constructing a system state vector based on the standardized data; and inputting the system state vector into the pre-trained deep reinforcement learning model.

[0028] Data cleaning can be understood as the process of removing duplicate, erroneous, and invalid information from the original data; standardized data can be understood as the process of converting original data of different formats, types, and units into standardized data in a unified numerical format; and system state vector can be understood as a set of numerical values ​​that integrates all standardized data and structurally represents the current operating and layout status of the station.

[0029] Specifically, the process begins by cleaning the operational data of each workstation, removing invalid data such as duplicate entries, incorrect formats, abnormal values, and missing key information to prevent dirty data from interfering with model decision-making and reducing decision accuracy. Next, character features such as text, symbols, and categories in the cleaned data are uniformly converted into numerical features to obtain standardized data, which meets the numerical computation requirements of deep reinforcement learning models. Finally, the standardized data is integrated to construct a system state vector, transforming scattered, multi-class, and heterogeneous data into structured data that the model can directly recognize and process. This ensures that the data input format is compliant, the information is complete, and the dimensions are consistent, effectively improving the accuracy, stability, and efficiency of model feature extraction and decision generation.

[0030] Optionally, the deep reinforcement learning model includes a pre-built Transformer encoder; the step of outputting a flowmeter calibration station disassembly and assembly layout decision sequence that satisfies preset process constraints based on feature extraction and multi-head attention weight allocation includes: performing deep feature extraction on the input data based on the Transformer encoder, and weighting the extracted deep features based on a multi-head self-attention mechanism; generating a flowmeter selection strategy probability distribution and a pipeline combination selection strategy probability distribution based on the weighted features; determining an initial decision sequence based on the flowmeter selection strategy probability distribution and the pipeline combination selection strategy probability distribution; verifying the initial decision sequence against preset process constraints, and determining the initial decision sequence as a flowmeter calibration station disassembly and assembly layout decision sequence and outputting it if the initial decision sequence satisfies all preset process constraints.

[0031] The Transformer Encoder can be understood as the core network structure in a deep reinforcement learning model used for deep feature extraction; the flow meter selection strategy probability distribution can be understood as the probability set of each type of flow meter being selected for this layout; the pipe combination selection strategy probability distribution can be understood as the probability set of each type of pipe combination being selected for this layout; the initial decision sequence can be understood as the set of multi-step layout operation instructions initially generated by the model and not yet verified by process constraints; and the process constraint verification can be understood as the verification process of checking whether the initial decision sequence meets the preset process conditions such as the total length of the station, flow meter type adaptation, and flow meter spacing.

[0032] Specifically, firstly, the input data is subjected to deep feature extraction using a Transformer encoder to uncover hidden information such as platform constraints and equipment adaptation. Then, a multi-head self-attention mechanism is used to assign weights to the deep features, highlighting the impact of key constraint features. Based on the weighted features, a probability distribution for flow meter selection and pipeline combination is generated to determine the initial decision sequence. The initial decision sequence is then verified against preset process constraints, and only when it meets all the constraints is it determined as the final disassembly and assembly layout decision sequence and output.

[0033] Optionally, before inputting the station operation correlation data into the pre-trained deep reinforcement learning model, the method further includes: collecting historical experience data during the disassembly and assembly layout process of the flow meter calibration station, wherein the historical experience data includes state data, action data, reward data, and next state data; dividing the historical experience data into multi-layer experience pools according to decision complexity or reward value; wherein each layer of experience pool independently stores corresponding experience data; extracting experience data from the multi-layer experience pools according to a preset sampling probability to form training batches; iteratively updating the model parameters based on the training batches using a proximal policy optimization algorithm; and completing the training of the deep reinforcement learning model when the model loss value converges, the cumulative reward value stabilizes, or the preset maximum number of iterations is reached.

[0034] Historical experience data can be understood as operational data, status data, action data, and result data accumulated during the disassembly and assembly of flowmeter calibration stations in the past; the experience pool can be understood as a dedicated dataset that centrally stores historical experience data for model training; the hierarchical experience pool can be understood as an optimized data structure that stores experience data hierarchically according to decision complexity or reward value to improve the utilization rate of high-value experience; the proximal policy optimization algorithm can be understood as a mainstream optimization algorithm used to iteratively update reinforcement learning model parameters and balance training stability and optimization efficiency; the loss value can be understood as the error value between the model prediction result and the actual result, used to measure the accuracy of model prediction; the cumulative reward value can be understood as the sum of rewards obtained during the model decision-making process, used to quantify the quality of the decision scheme; the number of iterations can be understood as the number of loops for updating model parameters. The preset maximum number of iterations can be pre-set based on experience, and this embodiment does not impose specific restrictions on it.

[0035] Specifically, model training and optimization are performed before data is input into the model. First, historical experience data from the disassembly and assembly of flowmeter calibration stations is collected, including state data at the time of decision-making, action data, reward data, and subsequent state data, forming experience tuples stored in the experience pool. Then, the experience pool is divided into multiple layers based on the complexity of the decision sequence or the reward value, with each layer independently storing corresponding experience data. High-value and high-reward experiences are prioritized to improve training efficiency and effectiveness. During training, experience data is extracted from each layer of the experience pool according to a preset sampling probability to form training batches. Based on these training batches, a proximal strategy optimization algorithm is used to iteratively update the model parameters, gradually reducing the model loss value, increasing the cumulative reward value, and optimizing the rationality of decisions. Finally, training is terminated when the model loss value converges, the cumulative reward value stabilizes, or the preset maximum number of iterations is reached, completing the optimization of the deep reinforcement learning model and ensuring that the model has the ability to output optimal, compliant, and stable decision sequences.

[0036] S130. Generate a visual layout scheme based on the disassembly and assembly layout decision sequence.

[0037] The visual layout scheme can be understood as transforming the abstract, digital, and textual layout decision sequence into intuitive chart-based displays.

[0038] Specifically, the abstract disassembly and assembly layout decision sequence output by the deep reinforcement learning model is transformed into a visually viewable and understandable chart solution. The visualization content covers dimensions such as equipment installation sequence, pipeline changes, constraint compliance, and layout efficiency. Operators can quickly grasp the overall layout design, equipment installation sequence, and constraint satisfaction without having to interpret complex data, thus improving the readability, understanding efficiency, and on-site execution convenience of the solution.

[0039] Optionally, the visualization layout scheme includes a time-series Gantt chart, a pipeline variation heatmap, a constraint satisfaction radar chart, and a global efficiency index panel; generating the visualization layout scheme based on the disassembly and assembly layout decision sequence includes: calculating the constraint satisfaction data of each preset process constraint, and generating the time-series Gantt chart, the pipeline variation heatmap, the constraint satisfaction radar chart, and the global efficiency index panel based on the disassembly and assembly layout decision sequence and the constraint satisfaction data.

[0040] Among them, constraint satisfaction data can be understood as numerical data that quantifies the degree to which each process constraint meets the standard and reflects the level of compliance of the layout scheme.

[0041] Specifically, firstly, for the disassembly and assembly layout decision sequence output by the deep reinforcement learning model, the compliance status of three types of preset process constraints—total station length, flow meter type compatibility, and flow meter spacing—is calculated one by one. The corresponding constraint satisfaction data is calculated to accurately quantify the compliance degree of each constraint and intuitively reflect whether the layout scheme meets the standards and the level of compliance. Then, combining the disassembly and assembly layout decision sequence and constraint satisfaction data, a time-series Gantt chart, a pipeline change heatmap, a constraint satisfaction radar chart, and a global efficiency indicator panel are generated simultaneously. The time-series Gantt chart shows the equipment installation sequence, the pipeline change heatmap presents the pipeline adjustment situation, the constraint satisfaction radar chart intuitively reflects the compliance of each constraint, and the global efficiency indicator panel quantifies the overall efficiency of the layout. This achieves a multi-dimensional and comprehensive visualization of the layout scheme, making it easy for operators to quickly grasp the details, compliance level, and efficiency of the scheme.

[0042] The technical solution of this invention solves the problems of fragmented and incomplete data collection in traditional manual layout by acquiring station operation-related data, thus achieving completeness of the data foundation for layout decisions. By inputting data into a pre-trained deep reinforcement learning model and outputting a compliant decision sequence, it solves the problems of low efficiency, easy violation of process constraints, difficulty in achieving global optimization, and lack of adaptive learning ability in manual layout, thus achieving automation, intelligence, global optimization, and long-term stability of layout decisions. By generating a visual layout scheme, it solves the problems of abstract and difficult-to-understand traditional layout schemes and poor readability, thus achieving intuitive display of layout results. Ultimately, it improves the efficiency, compliance, optimization level, and long-term operational stability of flowmeter calibration station disassembly and assembly layout.

[0043] Figure 2a This is a flowchart illustrating another method for determining the disassembly and assembly layout strategy of a flowmeter calibration station, provided by an embodiment of the present invention. Based on the above embodiments, this embodiment further optimizes the method for generating a visual layout scheme as described above. Specific implementation details can be found in the technical solution of this embodiment. Technical terms that are the same as or corresponding to those in the above embodiments will not be repeated here. Figure 2a As shown, the method specifically includes the following steps: S140. Receive adjustment instructions input by the user, modify the disassembly and assembly layout decision sequence according to the adjustment instructions, and output the updated visual layout scheme.

[0044] Among them, adjustment instructions can be understood as modification, optimization and adjustment instructions proposed by operators based on actual on-site working conditions, special needs or human experience regarding the visual layout scheme.

[0045] Specifically, after generating the visual layout scheme, the system opens a human-machine interface to receive adjustment instructions input by operators in real time. The instructions may include changing the flow meter model, adjusting the pipe specifications, modifying the equipment installation sequence, and fine-tuning the constraints. The system accurately analyzes the core content and modification requirements of the adjustment instructions, modifies the original disassembly and assembly layout decision sequence accordingly, and updates relevant parameters, constraint calculation results, and layout logic. Based on the modified decision sequence, the system recalculates the constraint satisfaction data and generates an updated visual layout scheme, balancing the global optimality of intelligent decision-making with the on-site adaptability of manual adjustments. This effectively adapts to special and unexpected needs in actual working conditions, further improving the practicality, executability, and on-site adaptability of the layout scheme.

[0046] The technical solution of this invention, by receiving user adjustment instructions, modifying the decision sequence, and outputting an updated visual layout scheme, solves the problems of traditional intelligent layout schemes being fixed and unable to adapt to special on-site working conditions and human experience optimization needs. It realizes the organic combination of intelligent global optimal decision-making and flexible human intervention, effectively improving the on-site adaptability, practicality, and executability of the flowmeter calibration station disassembly and assembly layout scheme, and further ensuring that the layout scheme can be implemented efficiently and compliantly.

[0047] Figure 2b This is a schematic diagram of the system functional structure of a method for determining the disassembly and assembly layout strategy of a flowmeter calibration station according to an embodiment of the present invention. Figure 2b As shown, the basic hardware components of the flow meter calibration station in this embodiment are as follows: Main mounting frame: The load-bearing structure of the inspection platform, which limits the total length, width and mounting plane of the installation, and is the foundation for the installation of all equipment; The flow meter installation unit, which includes calibration equipment for ultrasonic gas flow meters, turbine gas flow meters, mass flow meters, and rotary flow meters, is the core component of the layout. Piping assembly: includes straight pipe sections and measuring pipes, used to meet the upstream and downstream length requirements of the flow meter, and is the hardware carrier of pipeline inventory characteristics; Expansion joint: used to compensate for installation length errors and adjust the overall assembly length; its minimum / maximum / working length are the core parameters. Connection and positioning components: including flanges, elbows, pipe joints, and rectifiers, used for pipe connection and flow field regulation, and are the hardware basis for constraints; Detection and control interface: used for station status acquisition and signal transmission, providing data source for the status sensing module.

[0048] The status awareness module optimizes the flow meter calibration station layout globally by taking the station location as the optimization target. Specifically, it performs the following steps: 1.1) Data Acquisition: Obtain flow meter inventory information, pipeline inventory information, current station layout status, and expansion joint parameters; 1.2) Data cleaning: This includes removing duplicate and abnormal information; 1.3) Data type conversion: Converting character-type features in the data into numeric-type features; 1.4) Constructing the State Vector: Constructing the data into a system state vector. : ; in, Indicates the total length constraint. Indicates the range of the expansion joint. Represents the feature set of the flow meter. Represents the set of characteristics of pipeline inventory. This represents the historical state vector of the current stage layout.

[0049] The deep reinforcement learning decision module receives the system state vector processed by the state-aware module. The following steps are performed to generate a multi-step optimal decision strategy: 2.1) State encoding: encoding the state vector The input is fed into the Transformer encoder for feature extraction to obtain a deep feature representation; 2.2) Attention Mechanism Calculation: A multi-head self-attention mechanism is used to process features. The calculation formula is as follows: ; ; ; in, Represents the query matrix. Represents the key matrix. Let K be the value matrix, and K be the transpose matrix. For the input feature matrix, This represents the concatenation function. Let M represent the dimension of the key vector, and M represent the attention mask matrix. This represents the weight matrix for multi-head output fusion. , , Let represent the projection weight matrices of the query, key, and value for the i-th attention head, respectively, and h represent the number of attention heads; 2.3) Multi-step and two-step decision generation: Generating the future Two-step decision sequence for installing a flow meter: ; in, This represents the probability distribution of the flow meter selection strategy at step t. Let represent the probability distribution of the pipeline combination selection strategy at step t; 2.4) Constraint Verification: Verify whether the generated decision sequence satisfies all constraints, including: 2.4.1) Total length constraint: ; in, Indicates the length of the flow meter to be installed. This represents the sum of the total lengths of all straight pipe sections to be installed and measuring pipes. Indicates the actual usable length of the expansion joint. This indicates the maximum allowable total length of the testing station area; 2.4.2) Constraints related to flow meter type: For gas ultrasonic flow meters with rectifiers, the following requirements must be met: , And upstream of the flow meter The location must be a pipe section connection point.

[0050] For gas ultrasonic flow meters without a rectifier, the following must be met: , .

[0051] For gas turbine flow meters, mass flow meters, and rotary flow meters, the following requirements must be met: , ; in, This indicates the total length of the upstream straight pipe section and the measuring pipe of the flow meter. This indicates the total length of the downstream straight pipe section and the measuring pipe of the flow meter, where D represents the nominal diameter of the current flow meter. This represents the upstream length coefficient when the gas ultrasonic flow meter has a rectifier. This represents the downstream length coefficient when the gas ultrasonic flow meter has a rectifier. This represents the upstream length coefficient of a gas ultrasonic flow meter without a rectifier. This represents the downstream length coefficient of a gas ultrasonic flow meter without a rectifier. This represents the upstream length coefficient for gas turbine flow meters, mass flow meters, and rotary flow meters. This indicates the downstream length coefficient for gas turbine flow meters, mass flow meters, and rotary flow meters. This indicates the rectifier installation location coefficient.

[0052] 2.4.3) Flowmeter spacing constraints: ; in, This represents the center-to-center distance between flowmeter i and flowmeter j. This indicates the nominal diameter of flow meter i. This indicates the nominal diameter of flow meter j. Indicates the flow meter spacing coefficient; 2.5) Decision Optimization: The near-end policy optimization algorithm is used to optimize the decision strategy. The objective function is: ; in, Indicates model parameters, Represents the expected value of the trajectory sample. Indicates the discount factor. Indicates the probability ratio. This represents the generalized advantage estimation. Indicates timing difference error. Denotes the parameters for generalized dominance estimation. Indicates the clipping threshold. Representing state Value function estimation; 3) Strategy display module 3: Receive decision sequences generated by the deep reinforcement learning decision module Perform the following steps for visualization and interactive adjustments: 3.1) Decision Sequence Visualization: Displays decision sequences through various visualization components, including time-series Gantt charts. Pipeline variation heat map Constraint satisfaction radar chart and global efficiency metrics panel 3.2) Constraint Satisfaction Calculation: Calculate the constraint satisfaction vector. The calculation formula is: ; in, This represents the amount of violation of the i-th constraint. Indicates the maximum permissible violation amount; 3.3) Interactive adjustment: Provides a user interface that allows users to adjust the generated decision sequence, with the adjustment amount... It acts on the original decision sequence; The model training module provides continuous optimization capabilities for the deep reinforcement learning decision-making module, and performs the following training steps: 4.1) Experience collection: After each decision is executed by the system, the current state is collected. Execution of actions Receive rewards and the next state Forming empirical tuples 4.2) Hierarchical experience replay: The experience pool is divided into layers according to the complexity of the decision sequence or the reward value. Layers, each layer has an experience pool. Independent storage of experience; during training, according to sampling probability Training batches are formed by extracting experience from each layer, among which The calculation formula is: ; in, This represents the probability of sampling from the i-th layer of the experience pool. This indicates the priority of the i-th level of experience. The calculation formula is: , Let represent the number of experiences in the i-th layer experience pool, and e represent a single experience sample. This represents the temporal difference error of the empirical value e at step t. Indicates the decision-making step size. Indicates the discount factor. Indicates the total number of floors. Use small positive numbers to prevent division by zero. 4.3) Reward function calculation: In each training iteration, a multi-step reward function is calculated for the sampled empirical sequence to evaluate the long-term value of the action. The calculation formula is as follows: ; Where R represents the cumulative reward across multiple steps. This represents the reward for successfully completing step t. This indicates the penalty for violating the constraint at step t. This represents the efficiency reward at step t. This represents the smoothness reward at step t. This represents the number of pipes that change in step t. This represents the total number of pipes used in step t. , The weighting coefficient for efficiency rewards and , , The weighting coefficient for smoothness reward. This indicates the degree of variation in the flow meter selection. Indicates the degree of change in pipeline combination selection; 4.4) Course learning: In the early stages of training, start with smaller decision steps. Start training, and increase the number of training steps. As the threshold increases, the decision-making step size is gradually increased. The specific step size adjustment formula is as follows: ; in, This indicates the decision step size used in the current training. Indicates the maximum decision step size supported by the system. Indicates the current training step count. This indicates that the step size increases over time. 4.5) Dynamic weight adjustment: During training, the weights of each component in the reward function are dynamically adjusted according to the training progress. The specific weight adjustment formula is as follows: ; in, This indicates the weight of the i-th reward item in the training steps. The value of time, This represents the initial value of the weight. Indicates the magnitude of the weight change. This indicates the preset maximum number of training steps. 4.6) Global Optimization: By maximizing the expected value of the cumulative reward over multiple steps, the decision-making strategy is learned from the local optimum to the global optimum. The global optimization objective is: ; ; in, Let represent the composite reward function at step t. This represents the reward for successfully completing step t. This indicates the penalty for violating the constraint at step t. This represents the efficiency reward at step t. This represents the smoothness reward at step t; 4.7) Parameter Update: Using gradient descent, the model parameters of the deep reinforcement learning decision module are updated based on sampled empirical data and the calculated reward and advantage functions. Specifically, the gradient update rule of the near-end policy optimization algorithm is used to ensure the stability of policy updates. After each parameter update, the new policy is used for subsequent decision generation, forming a closed-loop iterative process of "decision-experience collection-training-parameter update" to continuously improve the model's decision-making ability.

[0053] The technical solution of this invention utilizes a deep reinforcement learning optimization system that integrates state perception, multi-step sequence decision-making, interactive display, and adaptive learning capabilities. This system includes a state perception module, a deep reinforcement learning decision-making module, a strategy display module, and a model training module. The system uses the state perception module to collect flow meter parameters, pipeline inventory, and station layout status in real time. The deep reinforcement learning decision-making module generates the optimal decision sequence for the future installation of multiple flow meters. The strategy display module visualizes this sequence and provides interactive adjustment functionality. The model training module continuously optimizes the decision model parameters based on real-time experience data, forming a complete perception-decision-display-optimization closed loop. This significantly improves the global optimization effect and automation level of the flow meter calibration station installation and removal layout strategy.

[0054] Figure 3 This is a schematic diagram of a device for determining the disassembly and assembly layout strategy of a flowmeter calibration station, provided in an embodiment of the present invention. Figure 3 As shown, the device includes: a data acquisition module 310, a decision sequence determination module 320, and a layout scheme generation module 330.

[0055] The data acquisition module 310 is used to acquire station operation-related data; the decision sequence determination module 320 is used to input the station operation-related data into a pre-trained deep reinforcement learning model, which outputs a flowmeter calibration station disassembly and assembly layout decision sequence that satisfies preset process constraints based on feature extraction and multi-head attention weight allocation; wherein the preset process constraints include station total length constraints, flowmeter type adaptation constraints, and flowmeter spacing constraints; and the layout scheme generation module 330 is used to generate a visual layout scheme based on the disassembly and assembly layout decision sequence.

[0056] The technical solution of this invention solves the problems of fragmented and incomplete data collection in traditional manual layout by acquiring station operation-related data, thus achieving completeness of the data foundation for layout decisions. By inputting data into a pre-trained deep reinforcement learning model and outputting a compliant decision sequence, it solves the problems of low efficiency, easy violation of process constraints, difficulty in achieving global optimization, and lack of adaptive learning ability in manual layout, thus achieving automation, intelligence, global optimization, and long-term stability of layout decisions. By generating a visual layout scheme, it solves the problems of abstract and difficult-to-understand traditional layout schemes and poor readability, thus achieving intuitive display of layout results. Ultimately, it improves the efficiency, compliance, optimization level, and long-term operational stability of flowmeter calibration station disassembly and assembly layout.

[0057] Optionally, the station operation associated data includes flow meter library data, pipeline inventory data, current station layout status data, and expansion joint parameter data.

[0058] Optionally, the decision sequence determination module 320 includes: The data cleaning unit is used to clean the operation-related data of the station, removing duplicate and abnormal data; The standardization processing unit is used to convert character features in the cleaned data into numerical features to obtain standardized data. The vector construction unit is used to construct a system state vector based on the standardized data and input the system state vector into a pre-trained deep reinforcement learning model.

[0059] Optionally, the deep reinforcement learning model includes a pre-built Transformer encoder; correspondingly, the decision sequence determination module 320 includes: The feature extraction and weight allocation unit is used to perform deep feature extraction on the input data based on the Transformer encoder, and to allocate weights to the extracted deep features based on the multi-head self-attention mechanism. The probability distribution generation unit is used to generate the flow meter selection strategy probability distribution and the pipeline combination selection strategy probability distribution based on the features after weight allocation. The sequence determination unit is used to determine an initial decision sequence based on the probability distribution of the flow meter selection strategy and the probability distribution of the pipeline combination selection strategy; The verification output unit is used to verify the initial decision sequence against preset process constraints. If the initial decision sequence satisfies all preset process constraints, the unit determines the initial decision sequence as the flow meter calibration station disassembly and assembly layout decision sequence and outputs it.

[0060] Optionally, the total length constraint of the station includes the fact that the sum of the total lengths of the flowmeter, straight pipe section, measuring pipe and expansion joint within the station area does not exceed a preset maximum length limit; The flow meter type adaptation constraints include distinguishing between rectifier-equipped and rectifier-free operating conditions for gas ultrasonic flow meters, ensuring that the upstream and downstream straight pipe lengths meet the corresponding process coefficient requirements, and ensuring that the upstream and downstream straight pipe lengths of gas turbine flow meters, mass flow meters, and rotary flow meters meet the corresponding independent process coefficient requirements. The flow meter spacing constraint includes that the center distance between adjacent flow meters is not less than a preset minimum spacing threshold.

[0061] Optionally, the device further includes: The historical data collection module is used to collect historical experience data during the disassembly and assembly layout process of the flow meter calibration station before inputting the station operation correlation data into the pre-trained deep reinforcement learning model. The historical experience data includes state data, action data, reward data, and next state data. The experience pool partitioning module is used to divide the historical experience data into multiple experience pools according to the decision complexity or reward value; wherein, each experience pool independently stores the corresponding experience data. The model iterative training module is used to extract empirical data from a multi-layered empirical pool according to a preset sampling probability to form training batches, and to iteratively update model parameters based on the training batches using a proximal strategy optimization algorithm. The model determination module is used to complete the training of the deep reinforcement learning model when the model loss value converges, the cumulative reward value stabilizes, or the preset maximum number of iterations is reached.

[0062] Optionally, the visualization layout scheme includes a time-series Gantt chart, a pipeline variation heatmap, a constraint satisfaction radar chart, and a global efficiency index panel; correspondingly, the layout scheme generation module is specifically used for: Calculate the constraint satisfaction data for each preset process constraint, and generate the time-series Gantt chart, the pipeline variation heat map, the constraint satisfaction radar chart, and the global efficiency index panel based on the disassembly and assembly layout decision sequence and the constraint satisfaction data.

[0063] Optionally, the device further includes: The scheme adjustment module is used to receive adjustment instructions input by the user after generating a visual layout scheme based on the disassembly and assembly layout decision sequence, modify the disassembly and assembly layout decision sequence according to the adjustment instructions, and output an updated visual layout scheme.

[0064] The device for determining the disassembly and assembly layout strategy of flowmeter calibration station provided in this embodiment of the invention can execute the method for determining the disassembly and assembly layout strategy of flowmeter calibration station provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the method.

[0065] Figure 4 This is a schematic diagram of an electronic device used to implement the method for determining the layout strategy of the flowmeter calibration station according to embodiments of the present invention. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workbenches, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0066] like Figure 4 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0067] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0068] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as determining the disassembly and assembly layout strategy for the flow meter calibration station.

[0069] In some embodiments, the determination of the method flowmeter calibration station setup and disassembly strategy can be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program can be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of determining the method flowmeter calibration station setup and disassembly strategy described above can be performed. Alternatively, in other embodiments, processor 11 can be configured to perform the determination of the method flowmeter calibration station setup and disassembly strategy by any other suitable means (e.g., by means of firmware).

[0070] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), complex programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations may include: implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0071] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0072] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0073] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0074] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0075] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0076] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0077] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for determining the layout strategy of flowmeter calibration station assembly and disassembly, characterized in that, include: Obtain the associated data for station operation; The associated data of the test station operation is input into a pre-trained deep reinforcement learning model. The deep reinforcement learning model outputs a decision sequence for the disassembly and assembly layout of the flow meter verification station that satisfies preset process constraints based on feature extraction and multi-head attention weight allocation. The preset process constraints include total station length constraints, flow meter type adaptation constraints, and flow meter spacing constraints. A visual layout scheme is generated based on the disassembly and assembly layout decision sequence.

2. The method of claim 1, wherein, The associated data for station operation includes flow meter database data, pipeline inventory data, current station layout status data, and expansion joint parameter data.

3. The method according to claim 1, characterized in that, The step of inputting the platform operation association data into a pre-trained deep reinforcement learning model includes: The operation-related data of the aforementioned stations is cleaned to remove duplicate and abnormal data; The character features in the cleaned data are converted into numerical features to obtain standardized data; A system state vector is constructed based on the standardized data, and the system state vector is input into a pre-trained deep reinforcement learning model.

4. The method according to claim 1, characterized in that, The deep reinforcement learning model includes a pre-built Transformer encoder; the decision sequence for the disassembly and assembly layout of the flowmeter calibration station, based on feature extraction and multi-head attention weight allocation and satisfying preset process constraints, includes: The Transformer encoder is used to extract deep features from the input data, and a multi-head self-attention mechanism is used to assign weights to the extracted deep features. Generate the flow meter selection strategy probability distribution and the pipeline combination selection strategy probability distribution based on the features after weight allocation; The initial decision sequence is determined based on the probability distribution of the flow meter selection strategy and the probability distribution of the pipeline combination selection strategy. The initial decision sequence is verified by preset process constraints. If the initial decision sequence satisfies all preset process constraints, the initial decision sequence is determined to be the flow meter calibration station disassembly and assembly layout decision sequence and output.

5. The method according to claim 1, characterized in that, The total length constraint of the station includes the fact that the sum of the total lengths of the flowmeter, straight pipe section, measuring pipe and expansion joint within the station area does not exceed the preset maximum length limit; The flow meter type adaptation constraints include distinguishing between rectifier-equipped and rectifier-free operating conditions for gas ultrasonic flow meters, ensuring that the upstream and downstream straight pipe lengths meet the corresponding process coefficient requirements, and ensuring that the upstream and downstream straight pipe lengths of gas turbine flow meters, mass flow meters, and rotary flow meters meet the corresponding independent process coefficient requirements. The flow meter spacing constraint includes that the center distance between adjacent flow meters is not less than a preset minimum spacing threshold.

6. The method according to claim 1, characterized in that, Before inputting the station operation association data into the pre-trained deep reinforcement learning model, the method further includes: Collect historical experience data during the disassembly and assembly layout of the flow meter calibration station. The historical experience data includes status data, action data, reward data, and next status data. The historical experience data is divided into multiple experience pools according to the decision complexity or reward value; each experience pool independently stores the corresponding experience data. Experience data is extracted from a multi-layered experience pool according to a preset sampling probability to form a training batch, and the model parameters are iteratively updated based on the training batch using a proximal strategy optimization algorithm. The training of the deep reinforcement learning model is completed when the model loss value converges, the cumulative reward value stabilizes, or the preset maximum number of iterations is reached.

7. The method according to claim 1, characterized in that, The visualization layout scheme includes a time-series Gantt chart, a pipeline variation heatmap, a constraint satisfaction radar chart, and a global efficiency index panel; the generation of the visualization layout scheme based on the disassembly and assembly layout decision sequence includes: Calculate the constraint satisfaction data for each preset process constraint, and generate the time-series Gantt chart, the pipeline variation heat map, the constraint satisfaction radar chart, and the global efficiency index panel based on the disassembly and assembly layout decision sequence and the constraint satisfaction data.

8. The method according to claim 1, characterized in that, After generating the visual layout scheme based on the disassembly and assembly layout decision sequence, the method further includes: Receive adjustment instructions from the user, modify the disassembly and assembly layout decision sequence according to the adjustment instructions, and output the updated visual layout scheme.

9. A device for determining the layout strategy of a flowmeter calibration station, characterized in that, include: The data acquisition module is used to acquire data related to the operation of the workstations. The decision sequence determination module is used to input the station operation association data into a pre-trained deep reinforcement learning model. The deep reinforcement learning model outputs a decision sequence for the disassembly and assembly layout of the flow meter verification station that satisfies preset process constraints based on feature extraction and multi-head attention weight allocation. The preset process constraints include total station length constraints, flow meter type adaptation constraints, and flow meter spacing constraints. The layout scheme generation module is used to generate a visual layout scheme based on the disassembly and assembly layout decision sequence.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the method for determining the disassembly and assembly layout strategy of the flowmeter calibration station as described in any one of claims 1-8.

Images