Air compression station constant pressure gas supply intelligent regulation and control system based on digital twinning

Abstract

The application relates to the technical field of air supply regulation of air compression stations, in particular to an air compression station constant-pressure air supply intelligent regulation system fusing digital twinning, which comprises the following: a topology weight time sequence coupling modeling module, which is used for generating a weight time sequence coupling topology graph; a disturbance path reverse deduction module, which generates a disturbance propagation path chain based on the correlation between flow mutation energy and time; a segmented regulation trigger sequence construction module, which divides time periods according to propagation delay and generates a trigger sequence according to pressure deviation; and a regulation instruction dynamic configuration module, which executes instruction merging and optimization configuration to generate a result set. The application extracts pressure response characteristics through synchronous sampling and sliding comparison, identifies a disturbance source by using path inversion, realizes time sequence alignment of regulation actions and disturbances in combination with propagation delay, effectively eliminates space-time coupling interference, and significantly improves the control precision, stability and operation efficiency of the air supply system.

Description

Technical Field

[0001] This invention relates to the field of air compressor station gas supply control technology, and in particular to an intelligent control system for constant pressure air supply in air compressor stations that integrates digital twins. Background Technology

[0002] As air compressor station gas supply control technology develops towards intelligence and precision, digital twin technology, due to its advantages in virtual-real mapping, synchronous sensing, and dynamic control, is gradually being introduced into the field of constant pressure gas supply in air compressor stations. However, current air compressor station control systems still have significant shortcomings in areas such as spatiotemporal coupling analysis of pressure fluctuations, disturbance path tracing, and timing alignment of adjustment actions, making it difficult to achieve high-precision modeling, real-time optimization, and stable control of the entire gas supply process in complex pipeline network environments.

[0003] In existing air compressor station supply control schemes, pressure sensors and flow meters are typically used to collect pipeline network data, and multiple air compressors are sequentially started and stopped or their loads switched according to a set pressure range, achieving basic pressure feedback regulation. However, most of these schemes establish control logic based on instantaneous pressure data, lacking a deep characterization of the time lag characteristics during pipeline pressure transmission. They also fail to construct a dynamic topology weight model that evolves synchronously with the physical pipeline network, making it impossible to achieve accurate mapping based on pressure response lag characteristics. This leads to the system being prone to response lag or pressure overshoot when facing drastic fluctuations in gas load, limiting the flexibility of regulation.

[0004] On the other hand, some intelligent control systems for air compressor stations attempt to drive control operations in a virtual environment by establishing a correspondence between equipment operating parameters and virtual models, using real-time data. Although such solutions introduce a virtual-real mapping mechanism, they are still limited to static parameter mapping and global synchronous adjustment. They do not consider the energy attenuation characteristics of disturbance signals during pipeline propagation, lack the ability to reverse-engineer multi-source disturbance paths and adaptively configure segmented adjustment trigger sequences, and are unable to cope with the nonlinear and time-varying challenges brought about by multi-node interactions in complex pipeline networks.

[0005] The aforementioned problems indicate that existing air compressor station control systems still have significant shortcomings in areas such as pressure hysteresis characterization, disturbance propagation path identification, control command timing alignment, and dynamic feedback accuracy. Therefore, this invention proposes an intelligent control system for constant pressure air supply in air compressor stations that integrates digital twins. The aim is to construct a high-fidelity, real-time, and invertible digital twin model, achieving closed-loop optimization across the entire chain from pressure characteristic inversion and disturbance path deduction to segmented command execution, thereby comprehensively improving the stability, control accuracy, and system operating efficiency of air compressor station air supply. Summary of the Invention

[0006] The purpose of this invention is to overcome the shortcomings of existing technologies and propose an intelligent control system for constant pressure air supply in air compressor stations that integrates digital twins.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: an intelligent control system for constant pressure air supply in an air compressor station integrating digital twins, the system comprising:

[0008] The topology weight temporal coupling construction module synchronously samples the pressure data of each node in the compressed air pipeline network, calculates the sum of pressure differences under different time offsets through time sliding comparison, selects the offset corresponding to the minimum difference as the pressure response lag time, generates a propagation consistency index in combination with pipeline length, and performs enhancement or weakening processing on the weights of the connecting edges based on the index to generate a weight temporal coupling topology graph.

[0009] The disturbance path reverse deduction module, based on the weighted temporal coupled topology graph, performs abrupt change detection on the node flow sequence and calculates the abrupt change energy value corresponding to the square of the flow difference. According to the energy decay law and the time interval increase relationship, it performs logical discrimination on the energy ratio and abrupt change time difference of adjacent nodes to generate a disturbance propagation path chain with physical propagation characteristics.

[0010] The segmented adjustment trigger sequence construction module extracts the arrival time of disturbances at each node in the disturbance propagation path chain, calculates the propagation delay sequence between nodes and divides the time period according to the magnitude of the delay, performs pressure deviation absolute value calculation on the adjustment target in each time period, filters out nodes with excessive deviations and determines increase or decrease instructions according to the deviation direction, and generates a segmented adjustment trigger sequence.

[0011] The dynamic configuration module for control commands compares the segmented adjustment trigger sequence with the preset command interval threshold, calls the current opening and speed parameters of the execution device, assigns the command sequences that meet the interval constraints to the continuous execution sequence, merges the remaining commands and assigns them to the discrete execution sequence, and generates a dynamic configuration result set of control commands by combining the device execution command library.

[0012] As a further embodiment of the present invention, the weighted temporal coupling topology graph includes pressure response lag time, propagation consistency index, connection edge enhancement coefficient, and priority path marker; the disturbance propagation path chain includes mutation energy value, energy attenuation ratio, disturbance propagation direction identifier, and path truncation marker; the segmented adjustment trigger sequence includes time period division index, node pressure deviation, equipment adjustment amplitude, and segmented execution interval; and the dynamic configuration result set of the control command includes continuous adjustment command, merged adjustment command, equipment opening correction value, and speed compensation parameter.

[0013] As a further aspect of the present invention, the topology weight temporal coupling construction module includes:

[0014] The pressure data acquisition submodule performs equal-interval sampling of the instantaneous pressure of each node in the compressed air pipeline network through pressure sensors, clocks all pressure sequences according to a unified time scale, performs cross-correlation sliding comparison on the pressure sequences of any two nodes, calculates the cumulative sum of pressure differences under different discrete shift step sizes, and generates a node pressure response characteristic table.

[0015] The lag feature extraction submodule calls the pressure difference cumulative sum sequence in the pressure response feature table of the node, retrieves the time offset corresponding to the minimum cumulative sum, marks it as the pressure response lag time, obtains the actual physical length of the pipeline between the nodes, calculates the ratio of the two, and obtains the corresponding propagation consistency index of each node.

[0016] The weight matrix generation submodule performs thresholding processing based on the deviation of the propagation consistency index from the set reference value. It enhances the weight of connection edges whose deviation is close to the reference value, weakens the weight of connection edges whose deviation exceeds a preset proportion, and maps the weights of node pairs in ascending order of lag time to generate a weighted temporal coupled topology graph.

[0017] As a further aspect of the present invention, the process of performing cross-correlation sliding comparison on the pressure sequences of any two nodes and calculating the cumulative sum of pressure differences under different discrete shift step sizes specifically involves: selecting a fixed-length sliding time window for all pressure sequences under a unified time scale; shifting successively within the sliding time window according to a preset discrete shift step size; performing difference calculation on the pressure values ​​at corresponding times and accumulating the results after absolute value processing to obtain the cumulative sum sequence of pressure differences corresponding to the shift amount; the process of retrieving the time offset corresponding to the minimum value of the cumulative sum specifically involves: screening candidate values ​​that satisfy the local minimum condition in the cumulative sum sequence of pressure differences and performing continuity judgment on the candidate values. The time offset that satisfies the minimum stability within a preset continuous shift interval is selected as the pressure response hysteresis time. The process of performing thresholding based on the deviation of the propagation consistency index from the set reference value is as follows: a deviation range between the propagation consistency index and the set reference value is set. When the propagation consistency index is within the first preset deviation range, the corresponding connection edge is marked as the enhancement type and assigned a first weight coefficient. When the propagation consistency index is outside the second preset deviation range, the corresponding connection edge is marked as the weakening type and assigned a second weight coefficient. The first weight coefficient is greater than the second weight coefficient, and both the first weight coefficient and the second weight coefficient are fixed calibration parameters.

[0018] As a further aspect of the present invention, the operation of the disturbance path reverse deduction module includes:

[0019] The flow mutation detection submodule acquires the values ​​of adjacent sampling points in the flow sequence of the pipeline node, calculates the flow difference, compares the difference with the set mutation threshold, adds a mutation timestamp to the sampling points that exceed the threshold, and performs a square operation on the flow difference to obtain the mutation energy value, generating a node mutation energy feature vector.

[0020] The energy sequence arrangement submodule arranges the mutation energy values ​​of each node in time sequence based on the mutation timestamp in the node mutation energy feature vector, calculates the energy decay ratio between adjacent nodes, marks the node with the highest energy value as the potential disturbance source point, and performs correlation mapping according to the physical topology path with energy decreasing step by step to generate disturbance evolution candidate paths.

[0021] The path chain construction submodule extracts the actual time interval between mutations of adjacent nodes based on the node sequence in the perturbation evolution candidate path, determines whether it satisfies the unidirectional increasing characteristic on the time axis, performs a retention operation on the path that satisfies the time increasing and energy decreasing, and performs path cutting and recombining of fragments on the nodes with time reversal to generate the perturbation propagation path chain.

[0022] As a further aspect of the present invention, the segmented adjustment trigger sequence construction module includes:

[0023] The delayed sequence analysis submodule obtains the disturbance arrival time label of each node in the disturbance propagation path chain, performs subtraction operation between adjacent nodes to obtain the propagation delay duration, re-sorts the nodes according to the delay duration values ​​from largest to smallest, and divides the entire path into several independent time response segments according to the delay distribution gradient, generating a path segment index dataset.

[0024] The adjustment target screening submodule extracts the current pressure value of the node based on the set of nodes in each time period in the path segment index dataset and calculates the deviation with the set constant pressure target value. It performs absolute value calculation on the deviation, filters the nodes whose absolute deviation value exceeds the dead zone threshold, and determines the gain or attenuation direction of the adjustment actuator according to the sign of the deviation, thus obtaining a segmented adjustment target list.

[0025] The sequence initialization generation submodule calls the segmented adjustment target list, extracts the execution device ID corresponding to the target node, maps the pressure deviation value to the corresponding adjustment amplitude instruction, completes the instruction logic encapsulation according to the order of the time period, and associates the propagation delay parameter of the corresponding segment to generate the segmented adjustment trigger sequence.

[0026] As a further aspect of the present invention, the dynamic configuration module for control commands includes:

[0027] The interval constraint discrimination submodule obtains the instruction execution interval time of adjacent time periods in the segmented adjustment trigger sequence, compares it with the preset minimum adjustment response threshold, constructs a Boolean logic discrimination array that satisfies the minimum response interval, and establishes an instruction interval constraint discrimination result set.

[0028] The instruction merging and classification submodule performs a time-domain aggregation operation on adjacent instructions with an interval less than a threshold based on the Boolean identifier in the instruction interval constraint discrimination result set. This merges multiple pulse-type adjustment actions into a continuous ramp-type adjustment instruction, and keeps the instructions with the interval meeting the standard in a discrete trigger state, thereby obtaining an instruction execution mode classification set.

[0029] The configuration result generation submodule calls the execution mode identifier in the instruction execution mode classification set, matches the opening control protocol and speed frequency table in the device execution instruction library, assigns an execution priority label to each type of instruction, adds the index code of the corresponding control cycle, and generates a dynamic configuration result set of control instructions.

[0030] As a further aspect of the present invention, the process of obtaining the instruction execution interval time of adjacent time periods in the segmented adjustment trigger sequence and comparing it with the preset minimum adjustment response threshold specifically involves: extracting the start timestamps of adjacent time periods in the segmented adjustment trigger sequence according to the time order and performing difference calculations to obtain the corresponding instruction execution interval time; comparing the instruction execution interval time with the preset minimum adjustment response threshold item by item to generate corresponding Boolean values; the process of performing time-domain aggregation operation on adjacent instructions with intervals less than the threshold specifically involves: performing amplitude superposition and execution duration reconstruction on adjacent instructions with Boolean values ​​of no according to the time order; accumulating the adjustment amplitudes of multiple instructions; and expanding the execution time according to the minimum response interval to form a continuously changing adjustment instruction sequence; the process of assigning an execution priority label to each type of instruction and adding the index code of the corresponding control cycle specifically involves: dividing the priority interval according to the category of the execution mode identifier in the instruction execution mode classification set; generating the corresponding control cycle index code according to the time period order in the segmented adjustment trigger sequence; and combining the priority label and the control cycle index code according to a fixed field format and writing them into the dynamic configuration result set of the control instruction.

[0031] As a further aspect of the present invention, the system further includes:

[0032] The closed-loop optimization execution module calls the adjustment instructions in the dynamic configuration result set of the control instructions, matches the priority path identifier in the weight time-series coupling topology diagram, extracts the dynamic feedback correction parameters of the associated execution devices, and combines them to generate a smart control instruction package for the entire gas supply process.

[0033] The intelligent control instruction package for the entire gas supply process includes a segmented execution time axis, a list of equipment adjustment quantities, a path coupling weight table, and a response delay correction value.

[0034] As a further aspect of this invention, the closed-loop optimization execution module includes:

[0035] The instruction timing parsing submodule obtains the time index in the dynamic configuration result set of the control instructions, performs instruction execution timing alignment processing for different execution modes, constructs a relative execution time axis based on the trigger time of the disturbance source point, and generates a control instruction timing sequence.

[0036] The parameter correlation submodule, based on the device number in the timing sequence of the control command, calls the edge weights and path markers in the weight timing coupling topology graph, calculates the adjustment amplitude correction coefficient based on weighted average, performs a product operation on the original adjustment amount to complete the parameter correction, and obtains the device adjustment correction parameter set;

[0037] The execution package submodule calls the correction adjustment amount and the control command timing sequence from the device adjustment correction parameter set, performs structured encapsulation according to a unified field format, adds a check bit and a control cycle end identifier, and establishes a smart control command package for the entire gas supply process.

[0038] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0039] In this invention, by synchronously sampling pressure data and performing sliding comparison, the pressure response lag time and propagation consistency index are explicitly extracted, and a dynamic topological weight model reflecting the physical propagation characteristics is constructed. The energy sequence generated by combining the square of the energy difference of flow mutation is used to perform disturbance path inversion, accurately identify the disturbance source and propagation chain, and divide the adjustment time period based on the propagation delay time. The absolute value of pressure deviation is calculated to screen the adjustment target, realizing the temporal alignment of adjustment actions and disturbance propagation process. By constraining the execution of action merging and optimization through command interval constraints, the timeliness of pressure compensation, the coordination of control commands, and the stability of system response are ensured under complex pipeline network conditions. The spatiotemporal coupling interference in pressure fluctuations is effectively eliminated, and the control accuracy and operating efficiency of constant pressure gas supply in air compressor stations are improved. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the system flow of the present invention;

[0041] Figure 2 This is a flowchart illustrating the acquisition process of the topology weight temporal coupling construction module of the present invention.

[0042] Figure 3 This is a flowchart illustrating the acquisition process of the disturbance path reverse deduction module of the present invention;

[0043] Figure 4 This is a flowchart illustrating the acquisition process of the segmented adjustment trigger sequence construction module of the present invention.

[0044] Figure 5 This is a flowchart illustrating the acquisition process of the dynamic configuration module for control commands in this invention.

[0045] Figure 6 This is a flowchart illustrating the acquisition process of the closed-loop optimization execution module of this invention. Detailed Implementation

[0046] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0047] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0048] In the embodiments of this invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, they convey the same meaning.

[0049] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0050] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0051] Please see Figure 1 This invention provides a technical solution: an intelligent control system for constant pressure air supply in an air compressor station that integrates digital twins, the system comprising:

[0052] The topology weight temporal coupling construction module synchronously samples the pressure data of each node in the compressed air pipeline network, calculates the sum of pressure differences under different time offsets through time sliding comparison, selects the offset corresponding to the minimum difference as the pressure response lag time, generates a propagation consistency index in combination with pipeline length, and performs enhancement or weakening processing on the weights of the connecting edges based on the index to generate a weight temporal coupling topology graph.

[0053] The disturbance path reverse inference module, based on the weighted temporal coupled topology graph, performs abrupt change detection on the node flow sequence and calculates the abrupt change energy value corresponding to the square of the flow difference. According to the energy decay law and the time interval increase relationship, it performs logical discrimination on the energy ratio and abrupt change time difference of adjacent nodes, and generates a disturbance propagation path chain with physical propagation characteristics.

[0054] The segmented adjustment trigger sequence construction module extracts the arrival time of disturbances at each node in the disturbance propagation path chain, calculates the propagation delay sequence between nodes and divides the time period according to the magnitude of the delay, performs pressure deviation absolute value calculation on the adjustment target in each time period, filters out nodes with excessive deviation and determines the increase or decrease instructions according to the deviation direction, and generates the segmented adjustment trigger sequence.

[0055] The dynamic configuration module for control commands compares the segmented adjustment trigger sequence with the preset command interval threshold, calls the current opening and speed parameters of the execution equipment, assigns the command sequence that meets the interval constraint to the continuous execution sequence, merges the remaining execution sequences and assigns them to the discrete execution sequence, and generates a dynamic configuration result set of control commands by combining the equipment execution command library.

[0056] The closed-loop optimization execution module calls the adjustment instructions in the dynamic configuration result set of the control instructions, matches the priority path identifiers in the weighted time-series coupling topology diagram, extracts the dynamic feedback correction parameters of the associated execution devices, and combines them to generate a smart control instruction package for the entire gas supply process.

[0057] The weighted temporal coupling topology diagram includes pressure response lag time, propagation consistency index, connection edge enhancement coefficient, and priority path marker. The disturbance propagation path chain includes mutation energy value, energy attenuation ratio, disturbance propagation direction identifier, and path truncation marker. The segmented adjustment trigger sequence includes time period division index, node pressure deviation, equipment adjustment range, and segmented execution interval. The dynamic configuration result set of control instructions includes continuous adjustment instructions, merged adjustment instructions, equipment opening correction value, and speed compensation parameters. The intelligent control instruction package for the entire gas supply process includes segmented execution time axis, equipment adjustment quantity list, path coupling weight table, and response delay correction value.

[0058] Please see Figure 2 The topology weight temporal coupling construction module includes:

[0059] The pressure data acquisition submodule performs equal-interval sampling of the instantaneous pressure of each node in the compressed air pipeline network through pressure sensors, clocks all pressure sequences according to a unified time scale, performs cross-correlation sliding comparison on the pressure sequences of any two nodes, calculates the cumulative sum of pressure differences under different discrete shift step sizes, and generates a node pressure response characteristic table.

[0060] In complex industrial fluid pipeline environments, air compressor station systems are often accompanied by strong electromagnetic interference, mechanical vibration noise, and nonlinear airflow pulsations. Therefore, a highly robust edge sensing network was constructed. Through multi-protocol gateways such as Industrial Ethernet, Profinet, and Modbus-TCP, high-bandwidth real-time communication links were established with various pressure transmitters deployed in the field. During multi-channel synchronous acquisition, high-precision diffused silicon pressure sensors were deployed at core nodes of the pipeline network (such as air tank outlets, main pipeline bifurcation points, and terminal core equipment inlets), capturing instantaneous pressure values ​​(e.g., 0.75 MPa) at a high-frequency sampling rate of 100 Hz. To eliminate high-frequency white noise and sensor drift errors caused by airflow turbulence, the system embedded an Extended Kalman Filter (EKF) algorithm on the edge computing side to dynamically smooth and clean the original sequence. Simultaneously, to eliminate phase differences caused by data transmission delays, the system's underlying layer forcibly introduced the IEEE 1588 (PTP) precision clock synchronization protocol or an NTP server to achieve microsecond-level clock alignment across all network nodes. For the cleaned and aligned pressure sequence... and Within a 5-second sliding time window, the system sets the discrete shift step size to 10ms, performs high-frequency cross-correlation sliding comparison, and calculates the cumulative sum of pressure differences. This generates a high-dimensional table of node-pair stress response features, laying an absolutely accurate data foundation for subsequent spatiotemporal feature extraction.

[0061] The lag feature extraction submodule calls the cumulative sum sequence of pressure difference values ​​in the pressure response feature table of the nodes, retrieves the time offset corresponding to the minimum value of the cumulative sum, marks it as the pressure response lag time, obtains the actual physical length of the pipeline between the nodes, calculates the ratio of the two, and obtains the corresponding propagation consistency index for each node.

[0062] After obtaining the node-to-pressure response characteristic table, due to the discreteness and massive volume of industrial data, the system did not adopt the computationally expensive global exhaustive search method. Instead, it introduced an optimization algorithm based on gradient descent and quadratic parabolic interpolation to automatically retrieve the accumulated sum with sub-millisecond computational precision. The global minimum point of the sequence, for example, the algorithm precisely locates the optimal offset corresponding to the minimum sum between nodes A and B. The value of seconds is precisely labeled as the actual physical response lag time of the aerodynamic pressure wave within this pipe segment. Simultaneously, the system deeply accesses the 3D static topology database of the digital twin model (such as a BIM architecture system) via API interfaces to extract the physical length of the straight lines between nodes. (e.g., 50 meters) More importantly, the system will convert local resistance components along the pipeline, such as elbows, reducers, and pneumatic valves, into equivalent physical lengths based on empirical fluid mechanics formulas and include them in the total length. Then, using the formula... The propagation speed consistency index is calculated. Considering that the sound speed of compressed air is significantly affected by the medium temperature, the system will also acquire the pipe temperature parameters in real time for compensation and calibration. If the index... Within the standard aerodynamic propagation velocity range under the current temperature and pressure conditions, this confirms that the physical connection has extremely high transmission fidelity, providing a solid physical basis for subsequent weight allocation.

[0063] The weight matrix generation submodule performs thresholding processing based on the deviation of the propagation consistency index from the set reference value. It enhances the weight of the connection edges whose deviation is close to the reference value, weakens the weight of the connection edges whose deviation exceeds the preset proportion, and maps the weights of the nodes in ascending order of lag time to generate a weight time-series coupled topology graph.

[0064] Propagation consistency index obtained from prior calculation The system performs a deep comparison and thresholding process with the dynamic reference threshold under standard operating conditions (combined with a local sonic empirical model based on real-time temperature and pressure compensation). For high-quality physical connections with a deviation of less than 5%, it indicates that the inner wall of the pipe segment is smooth and the transmission is unobstructed. The system assigns a weight enhancement coefficient of 1.2 to strengthen its control priority in the main transmission path. For abnormal connection edges with a deviation of more than 20%, the system determines that the pipe segment is highly likely to have hidden micro-leakage, condensate accumulation blockage, or complex unmodeled throttling effects, and assigns a weakening coefficient of 0.8 to actively reduce its command dependence in global control. After calculating the final coupling weight, the system calculates the coupling weight based on the lag time. The topology is sorted in ascending order and indexed using an efficient hash table structure. Edge weights are written into the dynamic topology matrix in real time to generate a weight-time coupled topology graph. This dynamically generated topology graph not only records the topological relationships, but also makes the implicit degradation characteristics of the pipeline network, such as physical wear and environmental interference, explicit and quantified, so as to ensure that the virtual twin model can adaptively match the real working conditions of the physical pipeline network at a high frequency.

[0065] Table 1: Weighted Temporal Coupling Topology Image Segment

[0066]

[0067] As shown in Table 1, this table records in detail the spatiotemporal coupling properties of pressure wave propagation in the pipeline network, ensuring that the topology in the virtual space can dynamically respond to the transmission characteristics of the physical pipeline network.

[0068] Please see Figure 3 The disturbance path reverse deduction module includes:

[0069] The flow mutation detection submodule acquires the values ​​of adjacent sampling points in the flow sequence of the pipeline node, calculates the flow difference, compares the difference with the set mutation threshold, adds a mutation timestamp to the sampling points that exceed the threshold, and performs a square operation on the flow difference to obtain the mutation energy value, generating a node mutation energy feature vector.

[0070] To accurately extract real gas consumption disturbance signals from massive amounts of noisy industrial operating data, high-sensitivity ultrasonic or thermal mass flow meters were deployed on key branch pipelines of the pipeline network. The system extracts the instantaneous flow values ​​of adjacent sampling points with an extremely short sampling period and performs first-order forward differential calculations. Unlike traditional control systems that use fixed and rigid alarm thresholds, this module introduces a dynamic confidence interval threshold mechanism based on adaptive learning from historical operating conditions. For example, when a threshold is detected... The flow rate at any time is 50m 3 / min suddenly increases to 80m with an unusually steep slope. 3 At / min, calculate The system compares this difference with a dynamic threshold. Once the threshold is exceeded, it is determined to be a genuine abrupt change, and a high-precision microsecond-level timestamp is added to the sampling point. To further filter out pseudo-signals of the "water hammer effect" caused by pipe resonance and minor internal leakage of valves, the system performs a nonlinear square operation on the flow difference. To obtain the mutation energy value (e.g.) This energy amplification mechanism can significantly widen the magnitude gap between real gas impact and occasional noise. Ultimately, the mutation energy of each node, along with the timestamp, is encapsulated into a high-dimensional node mutation energy feature vector, providing high signal-to-noise ratio data for subsequent three-dimensional localization of targeted disturbance sources.

[0071] The energy sequence arrangement submodule arranges the mutation energy values ​​of each node in time sequence based on the mutation timestamp in the node mutation energy feature vector, calculates the energy decay ratio between adjacent nodes, marks the node with the highest energy value as the potential disturbance source, and performs correlation mapping according to the physical topology path with progressively decreasing energy to generate disturbance evolution candidate paths.

[0072] The system's built-in deduction logic deeply draws upon the energy dissipation equations in fluid mechanics. It employs a generalized depth-first search algorithm from graph theory to perform rigorous reverse tracing of mutation energy. First, the system retrieves high-dimensional feature vectors and then arranges the mutation energy values ​​of each node temporally on the global time axis, locking... The value reaches a global peak (e.g.) The node with the earliest trigger timestamp is highlighted in the twin space as a "potential disturbance source" (usually representing the sudden opening of a heavy-duty gas-consuming device or the action of a large valve). Subsequently, the algorithm calculates the energy attenuation ratio between its downstream nodes and the source node along the preset physical connection direction of the pipeline network. Since compressed air inevitably suffers from frictional loss and local resistance loss during propagation in pipelines, the actual physical propagation path must strictly follow the objective law of decreasing energy gradient. Based on this constraint, the system dynamically removes pseudo-fluctuation branches caused by multi-source coupling interference and selects the correlation mapping that best matches the physical pipeline network direction as the candidate path for disturbance evolution. This inference mechanism fundamentally overturns the passive mode of blindly pressurizing the entire network in traditional constant pressure gas supply and achieves accurate characterization of the fluctuation evolution path.

[0073] The path chain construction submodule extracts the actual time interval between mutations of adjacent nodes based on the node sequence in the perturbation evolution candidate path, determines whether it satisfies the unidirectional increasing characteristic on the time axis, performs a retention operation on the path that satisfies the time increasing and energy decreasing, and performs path cutting and recombining of fragments on the nodes with time reversal to generate the perturbation propagation path chain.

[0074] After generating the initial candidate paths, the system must introduce strict causal and temporal constraints for secondary logical cleaning and reconstruction. Specifically, the algorithm extracts any adjacent nodes from the candidate path sequence. and High-frequency mutation timestamps and If the calculated difference satisfies If the propagation link conforms to the unidirectional time-incrementing characteristic of aerodynamic pressure waves in physical space from near to far, and simultaneously satisfies the preceding energy attenuation logic, the system determines that the propagation link has physical legitimacy and retains it. However, if during the comparison... This "time reversal" phenomenon, which violates the basic physical causality law, will cause the system to directly determine that the downstream node has been affected by another unidentified independent disturbance source (i.e., the multi-point concurrent conflict impact commonly seen in industrial sites). At this time, the system will automatically perform a forced path cutting operation at the node, decoupling and separating the originally chaotic fluctuation pattern. Finally, the system uses multi-threaded parallel processing technology to reconnect and combine all the path segments that have undergone rigorous legality verification, generating a disturbance propagation path chain with physical causality law, thus eliminating the misjudgment of control under multi-source concurrent impact from the underlying logic.

[0075] Please see Figure 4 The segmented adjustment trigger sequence construction module includes:

[0076] The delayed sequence analysis submodule obtains the arrival time label of the disturbance at each node in the disturbance propagation path chain, performs subtraction between adjacent nodes to obtain the propagation delay duration, re-sorts the nodes according to the delay duration values ​​from largest to smallest, and divides the entire path into several independent time response segments based on the delay distribution gradient, generating a path segment index dataset.

[0077] Faced with complex, long-distance, multi-branch factory pipeline networks, gas fluctuations affect all nodes in the network with extremely significant time step differences. To achieve refined control, the system first extracts the arrival time labels of all nodes affected by the impact from the legitimate disturbance propagation path chain. Then, it performs subtraction operations on topologically adjacent nodes along the path to obtain the absolute propagation delay sequence between the two points. For example, a set of consecutive delay sequences of 150ms, 160ms, 450ms, and 140ms is calculated. The algorithm incorporates a first-order differential gradient clustering model to dynamically scan the sequence. Once a delay is detected... When abrupt breaks in the sequence, such as a 450ms interval that significantly deviates from the smooth gradient (this phenomenon is usually caused by sudden changes in pipe diameter, a surge in local resistance, or buffering by a large gas storage tank), the system automatically uses this break as a boundary to divide the entire continuous propagation path into several independent time response segments (such as the "near-end transient response segment" and the "far-end hysteresis buffer segment"). Through this time-domain dimensionality reduction and segmentation processing, the system discretizes the originally complex and continuous pressure decay process into easily controllable independent time intervals and generates an accurate path segmentation index dataset, which greatly reduces the concurrent computational load of the controller.

[0078] The adjustment target filtering submodule extracts the current pressure value of the node based on the set of nodes in each time period in the path segment index dataset, calculates the deviation with the set constant pressure target value, performs absolute value calculation on the deviation, filters nodes whose absolute deviation value exceeds the dead zone threshold, and determines the gain or attenuation direction of the adjustment actuator according to the sign of the deviation, thus obtaining a segmented adjustment target list.

[0079] For each set of independent time-period nodes, the system extracts the current transient pressure value of the node from the digital twin pool in real time through a communication protocol, and calculates the deviation with the preset process constant pressure benchmark target value (e.g., 0.70 MPa). To avoid high-frequency oscillations in the controller due to minor fluctuations, the system performs absolute value calculations on the deviation. Subsequently, a dead zone threshold (e.g., 0.01 MPa) adaptively linked to the current unit load rate was introduced. Taking a certain node as an example, its current real-time pressure suddenly drops to 0.66 MPa, and the calculated absolute value of the deviation is 0.04. Since this value far exceeds the dead zone threshold of 0.01, the system immediately marks it as a high-priority target to be adjusted. Next, the system determines the polarity of the deviation: if... The system determines that the current pipeline is in an "under-pressure" state and assigns gain adjustment commands to the associated actuators (such as instructing the variable frequency air compressor to increase the frequency significantly); if the polarity is positive, it assigns attenuation commands (such as reducing the opening of the downstream pressure regulating valve or reducing the frequency of the main unit). Through this comprehensive screening mechanism, the segmented adjustment target list shown in Table 2 is obtained.

[0080] Table 2: Multi-target bias and time delay analysis table

[0081]

[0082] The sequence initialization generation submodule calls the segmented adjustment target list, extracts the execution device ID corresponding to the target node, maps the pressure deviation value to the corresponding adjustment amplitude instruction, completes the instruction logic encapsulation according to the order of the time period, and associates the propagation delay parameter of the corresponding segment to generate the segmented adjustment trigger sequence.

[0083] The abstract logical control intent is transformed into a draft of action instructions that can be parsed by the underlying hardware. After the system calls the pre-generated segmented adjustment target list, it accurately matches the specific execution equipment ID (such as a centrifugal air compressor, frequency converter, or proportional control valve with a specific number) responsible for covering the target pipe segment through a digital twin model. Then, the system calls the controller's built-in nonlinear three-dimensional mapping function (PID mapping surface) to smoothly convert, for example, a pressure deviation value of 0.045MPa into the corresponding physical adjustment amplitude instruction (e.g., to increase the air compressor motor speed by 50r / min or increase the given frequency by 2Hz). After the conversion is completed, the system strictly follows the chronological order of the time period (e.g., from SEG_01 to SEG_02) to complete the encapsulation of the instruction logic structure and deeply associates and binds each adjustment amplitude with the node's unique hysteresis delay parameter (e.g., 155ms). The initial segmented adjustment trigger sequence generated in this process is essentially a precise pre-arrangement of action nodes in the spatiotemporal dimension, ensuring that the pressurized airflow or throttling action released by the equipment can arrive at the target node exactly after the corresponding delay period, perfectly offsetting the pressure pulsation.

[0084] Please see Figure 5 The dynamic configuration module for control commands includes:

[0085] The interval constraint discrimination submodule obtains the instruction execution interval time of adjacent time periods in the segmented adjustment trigger sequence, compares it with the preset minimum adjustment response threshold, constructs a Boolean logic discrimination array that satisfies the minimum response interval, and establishes an instruction interval constraint discrimination result set.

[0086] In large air compressor station clusters, high-power frequency converters or high-voltage motors are extremely sensitive to frequent pulse-type starts and stops and drastic load switching, which can easily cause severe harmonic pollution of the power grid, mechanical bearing wear, and even thermal overload protection tripping. Therefore, this module performs a rigorous time-domain safety interval scan and comparison on command actions in adjacent time periods in the trigger sequence. The system first accurately obtains the theoretical execution interval time of adjacent actions (e.g., 45ms) and compares it with the preset minimum adjustment response threshold in the equipment library (e.g., 200ms dead time for unit-level electrical safety protection). If the actual interval time is too tight (i.e., 45ms < 200ms), the system automatically marks the command pair with a "False" Boolean conflict flag, determining that it does not meet the minimum response interval constraint and there is a risk of equipment damage; conversely, if the interval time is very generous (e.g., 350ms), the system automatically marks the command pair with a "False" Boolean conflict flag, determining that it does not meet the minimum response interval constraint and there is a risk of equipment damage. If the interval is 200ms, a "True" flag is set to allow the signal. The system performs multiple rounds of sliding scans on the global timing sequence and finally constructs a complete Boolean logic discrimination array. This instruction interval constraint discrimination result set is like the system's underlying security firewall, providing an insurmountable electrical red line for subsequent surge protection aggregation calculations and mechanical life protection.

[0087] The instruction merging and classification submodule performs time-domain aggregation operation on adjacent instructions with intervals less than the threshold based on the Boolean identifier in the instruction interval constraint judgment result set, merges multiple pulse-type adjustment actions into continuous ramp-type adjustment instructions, and keeps instructions with the interval meeting the standard in a discrete trigger state, thereby obtaining the instruction execution mode classification set.

[0088] After reading and parsing the Boolean discrimination array, the intelligent splitting and reconstruction mechanism is immediately activated. For high-frequency dense adjacent instruction pairs marked as "False" in the array, the system blocks their independent discrete issuance at the bus level and instead performs a smooth aggregation operation at the time domain level. For example, if the previous instruction of the same device requires "instantaneous increase of 2Hz", and the next instruction only 45ms later requires "continue to increase by 1Hz", the system will use a vector superposition algorithm to forcibly merge them into a ramp-type dynamic integral instruction that "smoothly climbs 3Hz within a continuous period of 300ms", thus transforming the destructive mechanical pulse into a gentle continuous loading action. In contrast, for independent instructions marked as "True", the system determines that they are in the absolute safe period of electrical operation and thus maintains their original discrete triggering characteristics to ensure that the device has the maximum agile response when facing isolated disturbances. This flexible instruction merging and rigid classification mechanism not only significantly reduces the fatigue wear of large mechanical equipment and effectively releases the communication bandwidth occupation of the underlying bus, but also generates a highly flexible instruction execution mode classification set.

[0089] The configuration result generation submodule calls the execution mode identifier in the instruction execution mode classification set, matches the opening control protocol and speed frequency table in the device execution instruction library, assigns an execution priority label to each type of instruction, adds the index code of the corresponding control cycle, and generates a dynamic configuration result set of control instructions.

[0090] For instruction sets with clear classification and reconstruction, the system further performs deep mounting and addressing of the underlying industrial communication protocol. After calling the instruction execution mode classification set, for continuous instructions classified as ramp mode, the system automatically matches specific opening control protocols and inverter speed frequency tables from the equipment execution instruction library (such as calling S-shaped acceleration curve parameters to smooth and suppress current surges). For discrete pulse instructions, it directly maps them to specific Modbus-RTU or TCP register addresses. More importantly, the system combines the node importance evaluated in the weighted coupling topology graph to assign "high priority" hardware interrupt tags to the heavy load regulation of the main pipeline network, giving priority to preempting the PLC execution cycle; and assigns "low priority" polling tags to the slight fluctuations of the terminal branches. Finally, the system adds a unique index code for the global control cycle to the data header to generate the final dynamic configuration result set of control instructions. This enables the heterogeneous air compressor cluster to process the differential pressure compensation task of the cycle in a high-concurrency manner according to the rules of clear hierarchy and unified protocol.

[0091] Please see Figure 6 The closed-loop optimization execution module includes:

[0092] The instruction timing parsing submodule obtains the time index from the dynamic configuration result set of control instructions, performs instruction execution timing alignment processing for different execution modes, constructs a relative execution time axis based on the trigger time of the disturbance source point, and generates a control instruction timing sequence.

[0093] In multi-device collaborative control scenarios, the response delay of each machine varies greatly due to differences in physical location and communication routing, which can easily lead to timing disorder and mutual cancellation in cluster control. After extracting the time index of each instruction from the dynamic configuration result set, the system forces high-precision timing alignment and rearrangement processing in the digital twin space. The algorithm uses the earliest moment when the pipeline network detects a sudden change in traffic as the system's absolute zero point. Taking into account the physical propagation delay of each gas supply path, the processing time of the bus gateway, and the dead zone of the mechanical transmission, the precise intervention time of each air compressor was deduced and rearranged. For example, the No. 1 high-power unit, responsible for compensating for fluctuations in the near-end backbone network, was set to... The entire unit was instantly engaged, releasing a large flow rate to fill the pressure gap; while the No. 2 auxiliary unit, responsible for stabilizing the pressure at the remote end, was strategically delayed until... Once the pressurization intervention is initiated, by constructing this relative execution time axis, the originally loose and isolated control commands are tightly woven into a closely interlocking time sequence, generating a highly coordinated control command time sequence, ensuring that the compensating airflow at each node perfectly offsets the pressure attenuation trough in the spatiotemporal dimension.

[0094] The parameter correlation submodule, based on the device number in the timing sequence of the control command, calls the edge weights and path markers in the weight timing coupling topology graph, calculates the adjustment amplitude correction coefficient based on weighted average, performs a product operation on the original adjustment amount to complete the parameter correction, and obtains the device adjustment correction parameter set;

[0095] Before the final instruction is issued, the system performs feedforward dynamic parameter correction. Based on the specific device ID number carried in the timing sequence of the control instruction, the system calls back the dynamic edge weights and path health markers in the weighted timing coupling topology graph in real time. For example, if the current coupling weight of the gas supply branch under the responsibility of a certain unit is 0.92, it indicates that the local resistance of the path is small and the transmission response is extremely fast. The system then uses an adaptive control algorithm to calculate the adjustment amplitude correction coefficient based on weighted average. Faced with the theoretically issued speed-up command of 100 r / min, the system performs a product operation to obtain the actual equipment adjustment and correction parameters. r / min, this weighted correction mechanism that deeply integrates topological features, cleverly compensates for the physical degradation state identified by the digital twin into the control closed loop, giving each control command the ability to adaptively correct deviations.

[0096] The execution package encapsulation submodule calls the correction adjustment amount and control command timing sequence in the equipment adjustment correction parameter set, encapsulates it in a structured manner according to a unified field format, adds a check bit and a control cycle end identifier, and establishes an intelligent control command package for the entire gas supply process;

[0097] As the final output port of the entire intelligent hub facing the underlying physical devices, it undertakes the core task of compiling and converting high-order algorithm intents into industrial-grade machine messages. The system simultaneously calls the finely corrected adjustment amount from the previous stage (such as 108.4 r / min) and the precisely aligned timing sequence of control instructions (such as execution timestamps). The system strictly adheres to the unified field format of international standard industrial control network communication for structured message splicing and packaging. To prevent data packet bit distortion caused by strong electromagnetic interference generated during the start-up of frequency converters in industrial sites, the system forcibly adds error-proof redundancy check bits (such as high-strength CRC16 check codes) to the message data frames and embeds a clear control cycle end marker (EOF) at the end to ensure the integrity and uniqueness of the PLC reading instruction cycle. After rigorous packaging and compilation verification, the system finally establishes a highly cohesive "intelligent control instruction package for the entire gas supply process". This instruction package is sent to the PLC execution end of each air compressor station through a high-bandwidth industrial gateway in milliseconds, forming a complete closed loop of anti-interference constant pressure gas supply from intelligent sensing, digital inference, timing configuration to physical servo.

[0098] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A smart control system for constant pressure air supply in an air compressor station integrating digital twins, characterized in that, The system includes: The topology weight temporal coupling construction module synchronously samples the pressure data of each node in the compressed air pipeline network, calculates the sum of pressure differences under different time offsets through time sliding comparison, selects the offset corresponding to the minimum difference as the pressure response lag time, generates a propagation consistency index in combination with pipeline length, and performs enhancement or weakening processing on the weights of the connecting edges based on the index to generate a weight temporal coupling topology graph. The disturbance path reverse deduction module, based on the weighted temporal coupled topology graph, performs abrupt change detection on the node flow sequence and calculates the abrupt change energy value corresponding to the square of the flow difference. According to the energy decay law and the time interval increase relationship, it performs logical discrimination on the energy ratio and abrupt change time difference of adjacent nodes to generate a disturbance propagation path chain with physical propagation characteristics. The segmented adjustment trigger sequence construction module extracts the arrival time of disturbances at each node in the disturbance propagation path chain, calculates the propagation delay sequence between nodes and divides the time period according to the magnitude of the delay, performs pressure deviation absolute value calculation on the adjustment target in each time period, filters out nodes with excessive deviations and determines increase or decrease instructions according to the deviation direction, and generates a segmented adjustment trigger sequence. The dynamic configuration module for control commands compares the segmented adjustment trigger sequence with the preset command interval threshold, calls the current opening and speed parameters of the execution device, assigns the command sequences that meet the interval constraints to the continuous execution sequence, merges the remaining commands and assigns them to the discrete execution sequence, and generates a dynamic configuration result set of control commands by combining the device execution command library.

2. The intelligent control system for constant pressure air supply in an air compressor station integrating digital twins as described in claim 1, characterized in that: The weighted temporal coupling topology includes pressure response lag time, propagation consistency index, connection edge enhancement coefficient, and priority path marker. The disturbance propagation path chain includes mutation energy value, energy attenuation ratio, disturbance propagation direction identifier, and path truncation marker. The segmented adjustment trigger sequence includes time period division index, node pressure deviation, equipment adjustment amplitude, and segmented execution interval. The dynamic configuration result set of control instructions includes continuous adjustment instructions, merged adjustment instructions, equipment opening correction value, and speed compensation parameters.

3. The intelligent control system for constant pressure air supply in an air compressor station integrating digital twins as described in claim 1, characterized in that, The topology weight temporal coupling construction module includes: The pressure data acquisition submodule performs equal-interval sampling of the instantaneous pressure of each node in the compressed air pipeline network through pressure sensors, clocks all pressure sequences according to a unified time scale, performs cross-correlation sliding comparison on the pressure sequences of any two nodes, calculates the cumulative sum of pressure differences under different discrete shift step sizes, and generates a node pressure response characteristic table. The lag feature extraction submodule calls the pressure difference cumulative sum sequence in the pressure response feature table of the node, retrieves the time offset corresponding to the minimum cumulative sum, marks it as the pressure response lag time, obtains the actual physical length of the pipeline between the nodes, calculates the ratio of the two, and obtains the corresponding propagation consistency index of each node. The weight matrix generation submodule performs thresholding processing based on the deviation of the propagation consistency index from the set reference value. It enhances the weight of connection edges whose deviation is close to the reference value, weakens the weight of connection edges whose deviation exceeds a preset proportion, and maps the weights of node pairs in ascending order of lag time to generate a weighted temporal coupled topology graph.

4. The intelligent control system for constant pressure air supply in an air compressor station integrating digital twins as described in claim 3, characterized in that: The process of performing cross-correlation sliding comparison on the pressure sequences of any two nodes and calculating the cumulative sum of pressure differences under different discrete shift step sizes is as follows: Select a fixed-length sliding time window for all pressure sequences under a unified time scale, shift successively within the sliding time window according to the preset discrete shift step size, perform difference calculation on the pressure values ​​at the corresponding time, process the absolute value, and then accumulate to obtain the cumulative sum sequence of pressure differences for the corresponding shift amount; The process of retrieving the time offset corresponding to the minimum value of the cumulative sum is as follows: in the pressure difference cumulative sum sequence, candidate values ​​that meet the local minimum condition are screened, and the continuity of the candidate values ​​is judged. The time offset that meets the minimum stability within the preset continuous shift interval is selected as the pressure response lag time. The process of performing thresholding based on the deviation of the propagation consistency index from the set reference value is as follows: a deviation range between the propagation consistency index and the set reference value is set. When the propagation consistency index is within the first preset deviation range, the corresponding connection edge is marked as an enhancement type and assigned a first weight coefficient. When the propagation consistency index is outside the second preset deviation range, the corresponding connection edge is marked as a weakening type and assigned a second weight coefficient. The first weight coefficient is greater than the second weight coefficient, and both the first weight coefficient and the second weight coefficient are fixed calibration parameters.

5. The intelligent control system for constant pressure air supply in an air compressor station integrating digital twins as described in claim 1, characterized in that, The disturbance path reverse deduction module includes: The flow mutation detection submodule acquires the values ​​of adjacent sampling points in the flow sequence of the pipeline node, calculates the flow difference, compares the difference with the set mutation threshold, adds a mutation timestamp to the sampling points that exceed the threshold, and performs a square operation on the flow difference to obtain the mutation energy value, generating a node mutation energy feature vector. The energy sequence arrangement submodule arranges the mutation energy values ​​of each node in time sequence based on the mutation timestamp in the node mutation energy feature vector, calculates the energy decay ratio between adjacent nodes, marks the node with the highest energy value as the potential disturbance source point, and performs correlation mapping according to the physical topology path with energy decreasing step by step to generate disturbance evolution candidate paths. The path chain construction submodule extracts the actual time interval between mutations of adjacent nodes based on the node sequence in the perturbation evolution candidate path, determines whether it satisfies the unidirectional increasing characteristic on the time axis, performs a retention operation on the path that satisfies the time increasing and energy decreasing, and performs path cutting and recombining of fragments on the nodes with time reversal to generate the perturbation propagation path chain.

6. The intelligent control system for constant pressure air supply in an air compressor station integrating digital twins as described in claim 1, characterized in that, The segmented adjustment trigger sequence construction module includes: The delayed sequence analysis submodule obtains the disturbance arrival time label of each node in the disturbance propagation path chain, performs subtraction operation between adjacent nodes to obtain the propagation delay duration, re-sorts the nodes according to the delay duration values ​​from largest to smallest, and divides the entire path into several independent time response segments according to the delay distribution gradient, generating a path segment index dataset. The adjustment target screening submodule extracts the current pressure value of the node based on the set of nodes in each time period in the path segment index dataset and calculates the deviation with the set constant pressure target value. It performs absolute value calculation on the deviation, filters the nodes whose absolute deviation value exceeds the dead zone threshold, and determines the gain or attenuation direction of the adjustment actuator according to the sign of the deviation, thus obtaining a segmented adjustment target list. The sequence initialization generation submodule calls the segmented adjustment target list, extracts the execution device ID corresponding to the target node, maps the pressure deviation value to the corresponding adjustment amplitude instruction, completes the instruction logic encapsulation according to the order of the time period, and associates the propagation delay parameter of the corresponding segment to generate the segmented adjustment trigger sequence.

7. The intelligent control system for constant pressure air supply in an air compressor station integrating digital twins as described in claim 1, characterized in that, The dynamic configuration module for control commands includes: The interval constraint discrimination submodule obtains the instruction execution interval time of adjacent time periods in the segmented adjustment trigger sequence, compares it with the preset minimum adjustment response threshold, constructs a Boolean logic discrimination array that satisfies the minimum response interval, and establishes an instruction interval constraint discrimination result set. The instruction merging and classification submodule performs a time-domain aggregation operation on adjacent instructions with an interval less than a threshold based on the Boolean identifier in the instruction interval constraint discrimination result set. This merges multiple pulse-type adjustment actions into a continuous ramp-type adjustment instruction, and keeps the instructions with the interval meeting the standard in a discrete trigger state, thereby obtaining an instruction execution mode classification set. The configuration result generation submodule calls the execution mode identifier in the instruction execution mode classification set, matches the opening control protocol and speed frequency table in the device execution instruction library, assigns an execution priority label to each type of instruction, adds the index code of the corresponding control cycle, and generates a dynamic configuration result set of control instructions.

8. The intelligent control system for constant pressure air supply in an air compressor station integrating digital twins as described in claim 7, characterized in that: The process of obtaining the instruction execution interval time of adjacent time periods in the segmented adjustment trigger sequence and comparing it with the preset minimum adjustment response threshold is as follows: extract the start timestamp of adjacent time periods in the segmented adjustment trigger sequence according to the time order and perform difference operation to obtain the corresponding instruction execution interval time, and compare the instruction execution interval time with the preset minimum adjustment response threshold item by item to generate the corresponding Boolean value; The process of performing time-domain aggregation operation on adjacent instructions with an interval less than the threshold is as follows: adjacent instructions with a Boolean value of no are processed by summing the amplitude and reconstructing the execution time in chronological order, the adjustment amplitudes of multiple instructions are accumulated, and the execution time is extended according to the minimum response interval to form a continuously changing sequence of adjustment instructions. The process of assigning an execution priority label to each type of instruction and adding an index code for the corresponding control cycle is as follows: priority intervals are divided according to the category of the execution mode identifier in the instruction execution mode classification set, and corresponding control cycle index codes are generated according to the time period order in the segmented adjustment trigger sequence. The priority label and the control cycle index code are combined and written into the control instruction dynamic configuration result set according to a fixed field format.

9. The intelligent control system for constant pressure air supply in an air compressor station integrating digital twins as described in claim 1, characterized in that, The system also includes: The closed-loop optimization execution module calls the adjustment instructions in the dynamic configuration result set of the control instructions, matches the priority path identifier in the weight time-series coupling topology diagram, extracts the dynamic feedback correction parameters of the associated execution devices, and combines them to generate a smart control instruction package for the entire gas supply process. The intelligent control instruction package for the entire gas supply process includes a segmented execution time axis, a list of equipment adjustment quantities, a path coupling weight table, and a response delay correction value.

10. The intelligent control system for constant pressure air supply in an air compressor station integrating digital twins as described in claim 9, characterized in that, The closed-loop optimization execution module includes: The instruction timing parsing submodule obtains the time index in the dynamic configuration result set of the control instructions, performs instruction execution timing alignment processing for different execution modes, constructs a relative execution time axis based on the trigger time of the disturbance source point, and generates a control instruction timing sequence. The parameter correlation submodule, based on the device number in the timing sequence of the control command, calls the edge weights and path markers in the weight timing coupling topology graph, calculates the adjustment amplitude correction coefficient based on weighted average, performs a product operation on the original adjustment amount to complete the parameter correction, and obtains the device adjustment correction parameter set; The execution package submodule calls the correction adjustment amount and the control command timing sequence from the device adjustment correction parameter set, performs structured encapsulation according to a unified field format, adds a check bit and a control cycle end identifier, and establishes a smart control command package for the entire gas supply process.

Images