Self-adaptive cooperative control system for industrial fluid pipe network valve nodes

Abstract

The invention relates to the field of industrial automation process control, and discloses a self-adaptive cooperative control system for industrial fluid pipe network valve nodes, which comprises control units distributed at the pipe network nodes, and the control units monitor target adjustment increment of controlled units and broadcast forecast signals. According to the method, the characteristic that the transmission speed of an electric signal is higher than the conduction speed of physical disturbance is utilized, the reverse feed-forward compensation amount is calculated and superposed according to a preset neighborhood influence weight table before the physical disturbance reaches an associated node, a digital reflex arc faster than physical response is constructed, the qualitative change of a control mode from lag correction to advanced immunity is realized, and the stability of the control mode is improved. Interlocking oscillation caused by the coupling effect of the multi-node network is suppressed, the load of the communication bus is ensured to be in a self-adaptive adjustment state while low computing power overhead is maintained, and the control real-time performance of large-scale networking is improved.

Description

Technical Field

[0001] This invention relates to an adaptive and coordinated control system for valve nodes in an industrial fluid pipeline network, belonging to the field of industrial automation process control technology. Background Technology

[0002] In current heating, petrochemical, and urban water supply fluid transmission networks, a mesh or ring topology consisting of multiple distributed regulating valves is commonly used to regulate the flow and pressure distribution across the entire network. The mainstream control architecture is based on distributed control systems or fieldbus control systems, with each valve node configured with an independent control unit. Single-loop proportional-integral-derivative (PID) algorithms are applied. The control unit calculates and adjusts the local valve opening based on real-time pressure or flow feedback signals collected from local sensors to maintain the stability of the controlled variable. This type of single-node independent feedback control mode has low hardware costs and is easy to deploy, making it suitable for steady-state or weakly coupled systems. Hardware integration solutions are becoming increasingly sophisticated, but if the control logic does not break through local limitations... Sampling and hysteresis feedback modes cannot eliminate the chain reaction of pipeline fluctuations. For example, the utility model patent with authorization announcement number CN208566861U discloses a hydraulic balance regulating device for a heating pipeline system. By configuring a multi-parameter measuring instrument and a fast shut-off device in an integrated box, it realizes centralized acquisition of node flow, temperature and pressure data and water loss alarm. The scheme is a post-hoc monitoring, and the control decision depends on the local parameter offset feedback. Under the condition of strong coupling of fluid medium, the time consumed by signal sampling, logic operation and actuator response is much greater than the time of pressure wave propagation at the speed of sound in the medium. The controller of the disturbed node cannot identify the source of the disturbance and regards it as a conventional external load interference for reverse regulation.

[0003] In actual industrial pipeline network operation, fluid media have incompressible or low-compressibility characteristics. Any valve adjustment action generates pressure waves that propagate at the speed of sound in the closed pipeline, causing transient and strong coupling effects on the fluid state of adjacent or hydraulically related nodes. Under single-loop independent control logic, nodes lack information interaction, and the control unit can only respond with delayed signals after capturing physical parameter deviations from local sensors. When neighboring nodes take action, the controller of the affected node cannot identify the source of the disturbance and treats the physical disturbance as a regular external load interference for reverse adjustment. The physical transmission speed of fluid pressure waves is much faster than the response speed of the sensor sampling to the actuator action feedback loop. The passive defense mechanism is prone to alternating misadjustments and adversarial oscillations between multiple nodes, exacerbating the entire network's problems. Pressure fluctuations can trigger water hammer accidents. Industry attempts have attempted to introduce centralized model predictive control based on global optimization or distributed adaptive decoupling algorithms based on high-dimensional matrix operations. Centralized solutions rely on high-bandwidth, low-latency industrial communication networks and high-performance central servers, which pose a risk of single-point failure and are difficult to adapt to the real-time requirements of large-scale distributed field operations. Distributed solutions based on complex algorithms require terminal node controllers to have high-frequency floating-point operations or matrix inversion capabilities, which contradicts the fact that industrial fields generally use general-purpose microcontrollers with limited computing resources to reduce costs and power consumption. The real-time handshake negotiation communication mechanism between nodes results in protocol stack processing and handshake delays, causing control signals to lag behind the arrival of physical pressure waves, making it impossible to complete pre-compensation before the disturbance takes effect.

[0004] Therefore, how to utilize the limited resources of existing industrial controllers to construct a low-computing-power adaptive decoupling control system that overcomes physical lag limitations and achieves low-computing-power decoupling between multiple nodes has become the technical problem to be solved by this invention. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of the present invention is as follows: An adaptive cooperative control system for valve nodes in an industrial fluid pipeline network, comprising control units distributed at each node of the pipeline network, wherein the control units store a neighborhood influence weight table, the neighborhood influence weight table recording the identifiers of associated nodes within the logical neighborhood and their corresponding coupling coefficients, and the control units operate in the following manner: Monitor the target adjustment increment of the local controlled unit, and when the absolute value of the target adjustment increment exceeds the broadcast trigger threshold, generate a warning signal containing the action amplitude value and action direction identifier and broadcast it to the associated node; Receive the warning signal from the associated node and index the corresponding coupling coefficient in the neighborhood influence weight table according to the identifier; The feedforward compensation value is obtained by multiplying the motion amplitude value by the coupling coefficient, and then the feedforward compensation value is superimposed on the local control command to drive the local controlled unit to perform state evolution. Among them, the warning signal is issued before or simultaneously with the local control unit outputting the command to drive the physical action, so as to establish the time advance of digital signal transmission before the physical medium transmits the disturbance. The superposition operation of feedforward compensation values ​​is limited to the time between the moment the warning signal is received and the moment when the controlled physical parameter fluctuations caused by the physical adjustment action of the associated node are transmitted to the local node; The control unit generates a compensation component that is opposite to the direction of the upcoming parameter fluctuation based on the coupling coefficient, so as to achieve lateral decoupling before the physical disturbance takes effect.

[0006] Preferably, the neighborhood influence weight table is constructed as a multi-dimensional gain scheduling matrix; the control unit performs gain scheduling according to the following rules: obtain the current basic state value of the controlled unit and convert it into a discretized execution interval identifier; encapsulate the discretized execution interval identifier in a warning signal for broadcast; parse the discretized execution interval identifier in the received warning signal as the first index dimension, and use the local current discretized execution interval identifier as the second index dimension; retrieve the specific operating condition coupling coefficient that matches the combination of the first index dimension and the second index dimension in the multi-dimensional gain scheduling matrix, and use it as the basis for calculating the feedforward compensation value.

[0007] Preferably, the control unit executes parameter self-calibration logic: continuously monitors the broadcast data stream in the communication network, and when only a single associated node's warning signal is received within a preset silent judgment time window and the local controlled unit is in a steady state, it collects the actual response characteristic data of the local controlled physical parameters; calculates the residual ratio between the actual response characteristic data and the theoretical response data derived based on the current coupling coefficient; establishes a confidence counter, and when the residual ratio shows a unidirectional deviation characteristic within multiple consecutive valid verification periods, it corrects the corresponding coupling coefficient in the multidimensional gain scheduling matrix according to the preset micro-step update rule.

[0008] Preferably, the broadcast trigger threshold is limited to the dead zone range used to eliminate noise fluctuations during steady-state operation; the data structure of the warning signal consists of source node identifier, discretized action level and action direction identifier, and it is strictly prohibited to include the absolute physical state value of the node, so as to reduce the load occupancy of the communication bus and ensure the real-time transmission of control commands.

[0009] Preferably, the coupling coefficient is a dimensionless constant pre-calibrated based on the pipeline hydraulic model. Its value represents the ratio between the parameter change caused by the physical fluctuations of the associated node when the unit adjustment action is transmitted to the local node and the parameter change caused by the unit adjustment action of the local node. The calculation process of the feedforward compensation value consists only of table lookup index operation and linear algebra multiplication operation.

[0010] Preferably, the control unit also executes security degradation logic: establishes a neighbor node heartbeat monitoring mechanism, and determines that the communication link with the associated node is failed when no heartbeat signal is received from a specific associated node within a preset period; temporarily sets the coupling coefficient corresponding to the failed associated node in the neighborhood influence weight table to zero, and generates local control commands only based on the controlled physical parameters fed back by the local sensing unit.

[0011] Preferably, the discretized execution interval identifier divides the entire stroke of the controlled unit into three logical intervals: low load region, linear control region, and saturation region; the multidimensional gain scheduling matrix stores nine independent coupling coefficients corresponding to the source node and local node under the above different logical interval combinations, so as to compensate for the nonlinear flow gain characteristics of the actuator under different adjustment intervals.

[0012] Preferred residual ratio Calculated according to the following rules: ,in, The residual ratio, and These are the start and end times of the valid verification period, respectively. The actual change in the controlled physical parameters collected by the local sensing unit. This refers to the theoretically expected parameter changes calculated based on the forecast signal and the current coupling coefficient.

[0013] Preferably, the control unit includes a microcontroller, and the neighborhood influence weight table is stored in the internal memory of the microcontroller; the forecast signal is broadcast through the industrial network; and the superposition operation of the feedforward compensation value is performed at the output register level of the microcontroller.

[0014] Preferably, the control unit executes a broadcast priority management strategy and marks the generated forecast signal as a priority processing packet when the absolute value of the local target adjustment increment exceeds a preset mutation level threshold; the cooperative control system also includes a network switching device configured to prioritize forwarding priority processing packets to ensure that the forecast signal arrives within the time lead.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. In the valve nodes of industrial fluid pipeline networks, taking advantage of the physical characteristic that the transmission speed of electrical signals on the communication bus is much higher than that of fluid pressure waves, a parallel physical pipeline network digital feedforward logic is constructed. Before or simultaneously with the actuator driven by the source node, a warning signal containing the amplitude information of the action is issued. This allows the associated nodes to pre-calculate and superimpose the reverse feedforward compensation amount based on the neighborhood influence weight table before the physical disturbance reaches the system through the fluid medium. At this time, the sequential logic establishes a time window that is faster than the physical response within the controlled system, converting the lag deviation correction process into a pre-set offset for the physical coupling effect that is about to occur, thus suppressing the chain oscillation caused by the hydraulic coupling effect in the multi-node network.

[0016] 2. Abandoning the reliance on real-time high-dimensional matrix operations or online global model solutions, this paper adopts a logical mapping based on a statically configured neighborhood influence weight table. The control unit directly queries the locally stored coupling coefficients to perform linear algebra operations, thus achieving approximate decoupling of the hydraulic influence of complex pipe networks. The complex fluid dynamic coupling relationship is pre-fixed into a lightweight parameter table, enabling high-precision collaborative control of multiple nodes on industrial embedded controllers with limited computing resources. This avoids the high computational load and convergence risk of online identification algorithms, ensuring the real-time performance and stability of the control system in industrial environments.

[0017] 3. Set up an event triggering mechanism based on the target adjustment value change threshold at the signal transmitting end. Only when the local action amplitude exceeds the preset dead zone will a broadcast warning signal be generated. This will eliminate minor noise fluctuations during steady-state operation, ensure that the communication network only carries state change information with substantial control significance, and reduce the occupancy rate of industrial fieldbus channels at the physical level through the on-demand broadcast communication strategy. This will avoid network congestion caused by high-frequency periodic polling, ensure priority transmission and immediate response of critical control commands in case of sudden operating conditions, and maintain the determinism of control timing when large-scale node networking is used. Attached Figure Description

[0018] Figure 1 This is a flowchart illustrating the distributed node collaborative control logic and timing of this invention. Figure 2 This is a feature diagram of the multidimensional gain scheduling of the nonlinear operating condition coupling coefficients in this invention. Figure 3 This is a schematic diagram of the system hierarchy architecture and physical deployment of the pipeline network of the present invention. Detailed Implementation

[0019] The present invention will be further described in detail below with reference to specific embodiments. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the scope of protection of the present invention.

[0020] This invention provides an adaptive cooperative control system for valve nodes in an industrial fluid pipeline network. The system comprises multiple control units distributed across various physical nodes in the pipeline network. These control units are connected via an industrial fieldbus or industrial Ethernet. Utilizing the physical characteristic that the transmission speed of electrical signals is higher than that of fluid pressure waves, the system performs lateral feedforward compensation before physical disturbances reach the associated nodes. Each control unit includes a microprocessor, memory, communication module, and input / output interfaces, used to execute local control algorithms and inter-node cooperative logic. The internal memory of each control unit contains a pre-set neighborhood influence weight table, which defines the identifiers of associated nodes within the logical neighborhood and their corresponding coupling coefficients. ,in, The coupling coefficient is a dimensionless constant used to characterize the proportional gain of the reverse adjustment amount required by the local node to maintain the stability of the controlled variable when the pressure wave caused by a unit amplitude of action of the associated node (source node) is transmitted to the local node (disturbed node). The following calibration procedure is used to determine the value: The disturbed node is brought to an open-loop steady state, and the amplitude generated by the driving source node is equal to the full valve stroke. to The step action is detected; the response data of the pressure sensor at the disturbed node is collected, and the pressure response peak value is extracted; the ratio of the pressure response peak value to the action amplitude of the source node is calculated, and the ratio is written into the memory as the basic coupling coefficient.

[0021] During system operation, the control unit executes an event monitoring and broadcasting process based on dead-time logic. The control unit runs a local PID control algorithm at a set period, which is set to [period value missing]. to Calculate the target adjustment increment at the current moment, and set the broadcast trigger threshold. The threshold value is taken as the full stroke of the valve. to When the absolute value of the calculated target adjustment increment exceeds the broadcast trigger threshold At that time, the control unit generates an action prediction signal, which includes a source node identifier, an action direction identifier, and a discretized action level identifier. The action direction identifier indicates forward activation or reverse deactivation, and the discretized action level identifier maps continuous action amplitudes to preset discrete levels. The control unit sends this action prediction signal to the communication network within the same clock cycle before outputting physical drive commands to the local actuator, or at least one communication cycle in advance. After receiving the action prediction signal from the associated node, the receiving control unit performs lateral signal parsing and feedforward compensation logic. The control unit parses the source node identifier in the action prediction signal and uses it as an index to retrieve the corresponding coupling coefficient in the local neighborhood influence weight table. The control unit is based on the formula Calculate feedforward compensation value ,in, The symbol value corresponding to the direction of the action. For the normalized amplitude value corresponding to the discretized action level identifier, the control unit will calculate the feedforward compensation value. This is directly superimposed onto the current control command of the local valve to generate a composite drive command. ,Right now ,in The basic control quantity output by the local PID control algorithm is superimposed through this operation, so that the local valve performs reverse opening adjustment before the pressure wave generated by the source node is physically transmitted to the local valve.

[0022] To address the nonlinear flow characteristics of the control valve, the control unit executes two-dimensional gain scheduling logic based on discrete operating ranges, allowing the system to manage the valve's entire stroke. to The execution is divided into three discretized regions: a low-load region, a linear control region, and a saturation region. The neighborhood influence weight table is configured as a multi-dimensional gain scheduling matrix, which stores the values ​​corresponding to the source node being in each region. And the local node is in the range Coupling coefficient under specific operating conditions in combination The transmitting control unit collects the current basic valve opening value, maps this basic opening value to the corresponding discretized execution interval identifier, and encapsulates this identifier in the action warning signal. The receiving control unit parses this identifier as the first index dimension, obtains the current local discretized execution interval identifier as the second index dimension, and retrieves the matching specific operating condition coupling coefficient in the multidimensional gain scheduling matrix. The control unit calculates the feedforward compensation value using this coefficient; it executes passive parameter self-calibration logic based on a silent window to correct model mismatch caused by drift in the physical characteristics of the pipeline network; the control unit continuously monitors the communication network and identifies the silent judgment time window, which is defined as: within a preset time window Within this time window, only a single associated node's action warning signal is received, and the target adjustment increment of the local control unit is lower than the action dead zone threshold. Set as Within the silent judgment time window, the control unit collects the actual response data sequence of the local pressure sensor. Based on the current coupling coefficient, the theoretically expected pressure response data sequence is calculated. The control unit calculates the integral residual ratio of the two within the time window. The calculation formula is: When the integral residual ratio In continuous Within a valid verification period, it is consistently greater than or less At that time, the control unit utilizes microsteps The corresponding coefficients in the weight table are corrected using the following formula: ,in Set as , Set as The control unit executes the security degradation logic. The control unit monitors the heartbeat signal of the associated node through the communication network. When the heartbeat signal of a specific associated node is not received within a preset period, the control unit determines that the communication link with the associated node has failed and temporarily sets the coupling coefficient corresponding to the associated node in the neighborhood influence weight table to zero. At this time, the control unit generates control commands only based on the feedback signal of the local sensor until the communication link is restored.

[0023] Example 1: In the ring-networked operation of a large-scale district heating network including three heat source stations and twelve distributed booster pump stations, the fluid medium exhibits transient hydraulic coupling characteristics under full-load and high-pressure conditions in winter. Furthermore, the pressure wave generated by the emergency flow cut-off action of a single heat source station propagates in the network at the speed of sound. When the control unit of the heat source station (as the source node) detects that the absolute value of the target adjustment increment of the local valve exceeds the set full stroke... When the broadcast trigger threshold is reached, the control unit generates an action warning signal containing a source node identifier, a shutdown direction identifier, and a discretized action level identifier within the same clock cycle before driving the physical actuator, and broadcasts it to the communication network. The pressure regulating valve, a critical area downstream of the pressure wave propagation path and thus the disturbed node, receives the action warning signal using the time difference between light speed transmission and medium sound speed transmission. Its control unit indexes the corresponding basic coupling coefficient in the local neighborhood influence weight table based on the source node identifier, and combines this with the current state of the local valve. The linear control region state of the opening degree is determined by calling the matching high-gain specific operating condition coupling coefficient in the multi-dimensional gain scheduling matrix. The disturbed node control unit is then configured according to the formula. Calculate feedforward compensation value And superimpose it onto the local PID control command. Generate synthesis driving instructions The local valve is driven to perform reverse pre-regulation before the physical pressure wave arrives. The hydraulic impedance change generated by this pre-regulation action cancels out the pressure impact energy in the time domain, so that the controlled pressure variable of the disturbed node remains stable without triggering the local feedback regulation dead zone.

[0024] Example 2: This example constructs a closed-loop hydraulic testing platform to verify the actual control efficiency of the technical solution of the present invention in a real industrial pipe network environment with complex noise interference, and to determine the effective working boundary of key control parameters. The platform simulates the physical architecture of a secondary urban heating pipe network, consisting of three variable frequency centrifugal pumps and five sets of pipes with diameters of [missing information]. The system consists of steel pipelines and electrically operated control valves distributed at key nodes, forming a ring topology with multi-source drive characteristics. To reproduce the real signal environment of the industrial site, the data acquisition system directly receives data from pressure transmitters (measuring range) installed at each node. to Precision level (Level) to To verify the system's anti-interference capability under non-ideal conditions, a signal-to-noise ratio of [value missing] was artificially superimposed onto the original acquired signal. Gaussian white noise was used to simulate the mechanical vibration interference during pump station operation, and a frequency of was introduced. Power frequency interference components.

[0025] In this experiment, the broadcast trigger threshold With coupling coefficient The settings follow the following engineering decision-making logic: for broadcast trigger thresholds Its settings need to balance the contradiction between the communication bus load rate and the ability to suppress minor disturbances. If the amplitude is too large, the system will be insensitive to oscillations in the low and medium frequency ranges; if... If the value is too small, it will lead to channel congestion. Based on Shannon's sampling theorem and the characteristic frequency analysis of pipeline pressure waves, this experiment will... Set to full valve stroke This value is greater than the equivalent opening fluctuation corresponding to the sensor's noise floor, ensuring that only valid actions trigger the broadcast, and addressing the coupling coefficient. The values ​​are not chosen empirically, but determined based on open-loop step tests of the action response. Under steady-state conditions, the driving source node generates... The step action is measured to determine the amplitude of the pressure response at the disturbed node. Based on the flow characteristic curve of the valve at the disturbed node, the maintenance... Required theoretical reverse opening Take the ratio As The baseline value.

[0026] The test procedure was set to verify the lateral decoupling capability under simulated emergency conditions. Initially, the pipeline network was in a state of... Under steady-state operating pressure, all node control units are in automatic mode. After the test starts, the execution amplitude of the source node (node ​​A) is... The rapid shutdown action (simulating fault clearance) will generate a strong positive pressure water hammer wave in the pipeline network and propagate to the downstream node (node ​​B). Three comparative sample groups were set up to verify the response differences of different control strategies: the control group used traditional independent PID control with no communication between nodes; the sample group of this invention enabled cooperative control logic based on the aforementioned calibration parameters; the boundary verification group used coupling coefficients... Set as calibration value To test the boundary effects of over-optimized parameters, each group of experiments was repeated. To eliminate random errors, the statistical results of the key process data are shown in Table 1.

[0027] Table 1: Comparison of Pressure Response Characteristics of Node B under Sudden Shutdown Conditions

[0028] Experimental data showed that, in the control group, node B could only initiate regulation after detecting a pressure deviation, resulting in a pressure peak deviation as high as [missing data]. Furthermore, the delayed feedback has led to continuous [problems / issues]. In contrast to the multiple rounds of oscillations, the sample group of this invention operates simultaneously with the source node (earlier than the arrival of the pressure wave). That is, by superimposing a reverse feedforward compensation at node B, the pressure peak deviation is reduced to The decline reached And the stabilization time is shortened to To verify the effect of the side-feedback mechanism on suppressing physical disturbances, the boundary validation group data shows that when the coupling coefficient... Exceeding the calibration value At times, excessive feedforward compensation can artificially create a reverse pressure drop (peak deviation). This causes the system to enter an overcompensated oscillation state.

[0029] Example 3: This example combines Figures 1 to 3 An adaptive cooperative control system for valve nodes in an industrial fluid pipeline network is described, such as... Figure 1As shown, the collaborative control logic of this system is jointly executed by the source node control unit and the associated node control unit. The source node control unit identifies the action requirements of the controlled unit by monitoring the target adjustment increment. At the same time or before the physical valve of the local controlled unit performs the adjustment action, it broadcasts a warning signal containing the source node ID and the action amplitude to the communication network. Taking advantage of the characteristic that the transmission speed of electrical signals is much greater than the transmission speed of physical disturbances, the associated node control unit receives the warning signal and obtains the source node action information before the physical disturbance arrives with a delay of the medium's sound speed. Based on the neighborhood influence weight table of locally stored coupling coefficient and gain scheduling, it multiplies the action amplitude with the coupling coefficient to calculate the reverse feedforward compensation amount, and superimposes it into the local control command to generate a synthetic drive signal. Finally, it drives the controlled unit to complete the preventive execution adjustment before the physical disturbance arrives.

[0030] like Figure 2 As shown in the figure, the horizontal axis represents the three discretized execution intervals of the local node: low load region, linear control region, and saturation region, respectively, and the vertical axis represents the coupling coefficient under specific operating conditions. Dimensionless values ​​between 0.7 and 1.4, where dashed lines represent the coefficient distribution of the source node in the low-load region, solid lines represent the coefficient distribution of the source node in the linear control region, and dotted lines represent the coefficient distribution of the source node in the saturation region. Each data point corresponds to nine independent coupling coefficients between the source node and the local node under different logical interval combinations; for example... Figure 3 As shown, the overall architecture of the system is divided into a communication network layer and a physical pipeline layer. The communication network layer uses industrial fieldbus or industrial Ethernet to realize bus connection between nodes. The physical pipeline layer includes fluid pipeline profiles and distributed source node regulating valves, disturbed node regulating valves and pressure sensors. Both the source node control unit and the disturbed node control unit integrate microcontrollers (MCUs), neighborhood influence weight tables, communication modules and I / O interfaces, and are respectively connected to actuators to drive physical valves. In the figure, the dashed path indicates the prediction signal transmission path with faster electrical signal transmission speed, and the solid arrow indicates the pressure wave physical disturbance transmission path with slower transmission speed.

[0031] Example 4: This example addresses the black-box problem of model physical parameter drift caused by long-term operation in traditional control schemes by constructing parameter self-calibration logic based on passive residual comparison and micro-step iteration. This scheme does not rely on additional active test stimuli but utilizes silent window data from the natural operation of the pipeline network to achieve calibration of core coupled parameters. The online transparent calibration system defines the logical capture conditions for the silent window: that is, within the time window. Set as to Within the communication network, only a single associated node (source node)'s action prediction signal is detected, and the target adjustment increment of the local control unit is lower than the action dead zone threshold. This condition is set based on statistical principles, selecting single-variable excitation samples with high signal-to-noise ratio from the complex pipeline network dynamic data stream to eliminate the impact of multi-source interference on parameter identification accuracy. Once an effective silent window is locked, the control unit initiates the residual feature extraction process, and the system records two data curves in parallel: one is the actual response curve collected by the local pressure sensor. The other is the coupling coefficient based on the current storage. and the theoretical expected response curve calculated from the received source node action amplitude. The system calculates the integral ratio of the two within the time window. ,Right now The ratio Reflects the degree of deviation of the current model parameters from the actual physical properties: If This indicates that the actual physical response is stronger than the model prediction. Too small; conversely, too large.

[0032] Based on the above ratio, the system executes micro-step size iterative correction logic, and the system maintains the calibration confidence counter. ,like In continuous One (e.g.) All (number) effective silent windows exhibit the same deviation characteristic, for example, consistently greater than [number]. or less If the problem is determined to be a systematic physical parameter drift rather than random noise interference, the control unit will trigger a parameter update command based on the formula. The coupling coefficient is corrected, where For microstep size factor such as This microstep design follows the principle of conservative iteration to ensure that the parameter adjustment process is smooth and convergent, and avoids oscillation of control logic due to single misjudgment.

[0033] Example 5: This example describes a standardized offline calibration and data filling procedure, constructs and verifies a multi-dimensional gain scheduling matrix, ensuring that the matrix has accurate physical mapping capabilities before actual operation. This procedure is executed on a controlled hydraulic experimental platform. By traversing preset operating condition combinations, real coupling coefficient data points are obtained. Discrete execution intervals for the source node and the disturbed node are set. The system divides the entire valve stroke into three logical intervals: low load, linear control, and saturation. For each pair of source and disturbed node combinations, nine possible interval combinations are simulated one by one on the experimental platform. For each combination, the source node is positioned at the center opening of its corresponding interval and driven to perform a standard amplitude step action; the disturbed node is positioned at the center opening of its corresponding interval, and the open-loop pressure response data of this node without feedforward compensation is recorded. Then, based on the collected open-loop response data, the coupling coefficient for a specific operating condition is calculated. For each combination of operating conditions, the system extracts the peak pressure response of the disturbed node. By combining this with the current flow characteristic curve of the node, the theoretical opening adjustment amount required to offset the pressure fluctuation can be calculated. Calculate the ratio ,in The step amplitude of the source node is calculated by repeating the process multiple times and averaging the results to eliminate random errors. Finally, the nine calculated values ​​are... The values ​​are filled into the corresponding cells of the multidimensional gain scheduling matrix to complete the parameterized mapping of the static topology.

[0034] In addition, a pre-deployment calibration procedure is performed on-site to adapt to the specific physical environment of the actual pipeline network. After the system is first connected to the target pipeline network, it does not immediately start fully automatic closed-loop operation, but enters passive monitoring mode. In this mode, the control unit only listens to and records the action warning signals in the network and the actual response of the local sensors, but does not perform physical actions. The system uses the built-in statistical analysis module to compare the theoretical response calculated by the preset matrix parameters with the actual response measured on-site, and calculates the deviation coefficient. If the deviation exceeds the preset threshold, the system automatically performs overall scaling correction on the basic gain in the matrix based on the measured data until the deviation converges to the allowable range before switching to active collaborative control mode.

[0035] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0036] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An adaptive cooperative control system for valve nodes in an industrial fluid pipeline network, characterized in that, This includes control units distributed across various nodes of the pipeline network. Each control unit stores a neighborhood influence weight table, which records the identifiers of associated nodes within the logical neighborhood and their corresponding coupling coefficients. The control units operate as follows: Monitor the target adjustment increment of the local controlled unit, and when the absolute value of the target adjustment increment exceeds the broadcast trigger threshold, generate a warning signal containing the action amplitude value and action direction identifier and broadcast it to the associated node; Receive the warning signal from the associated node and index the corresponding coupling coefficient in the neighborhood influence weight table according to the identifier; The feedforward compensation value is obtained by multiplying the motion amplitude value by the coupling coefficient, and then the feedforward compensation value is superimposed on the local control command to drive the local controlled unit to perform state evolution. Among them, the warning signal is issued before or simultaneously with the local control unit outputting the command to drive the physical action, so as to establish the time advance of digital signal transmission before the physical medium transmits the disturbance. The superposition operation of feedforward compensation values ​​is limited to the time between the moment the warning signal is received and the moment when the controlled physical parameter fluctuations caused by the physical adjustment action of the associated node are transmitted to the local node; The control unit generates a compensation component that is opposite to the direction of the upcoming parameter fluctuation based on the coupling coefficient, so as to achieve lateral decoupling before the physical disturbance takes effect.

2. The adaptive cooperative control system for valve nodes in an industrial fluid pipeline network according to claim 1, characterized in that, The neighborhood influence weight table is constructed as a multi-dimensional gain scheduling matrix; the control unit performs gain scheduling according to the following rules: obtain the current basic state value of the controlled unit and convert it into a discretized execution interval identifier; encapsulate the discretized execution interval identifier in the warning signal for broadcast; parse the discretized execution interval identifier in the received warning signal as the first index dimension, and use the local current discretized execution interval identifier as the second index dimension; The coupling coefficient for a specific operating condition that matches the combination of the first and second index dimensions is retrieved from the multidimensional gain scheduling matrix and used as the basis for calculating the feedforward compensation value.

3. The adaptive cooperative control system for valve nodes in an industrial fluid pipeline network according to claim 2, characterized in that, The control unit executes parameter self-calibration logic: continuously monitors the broadcast data stream in the communication network, and when only a single associated node's warning signal is received within the preset silent judgment time window and the local controlled unit is in a steady state, it collects the actual response characteristic data of the local controlled physical parameters; Calculate the residual ratio between the actual response feature data and the theoretical response data derived based on the current coupling coefficient; establish a confidence counter, and when the residual ratio shows a unidirectional deviation characteristic in multiple consecutive valid verification periods, correct the corresponding coupling coefficient in the multidimensional gain scheduling matrix according to the preset micro-step update rule.

4. The adaptive cooperative control system for valve nodes in an industrial fluid pipeline network according to claim 1, characterized in that, The broadcast trigger threshold is limited to the dead zone range used to eliminate noise fluctuations during steady-state operation; the data structure of the warning signal consists of the source node identifier, discretized action level, and action direction identifier, and it is strictly prohibited to include the absolute physical state value of the node, so as to reduce the load occupancy of the communication bus and ensure the real-time transmission of control commands.

5. The adaptive cooperative control system for valve nodes in an industrial fluid pipeline network according to claim 1, characterized in that, The coupling coefficient is a dimensionless constant pre-calibrated based on the pipeline hydraulic model. Its value represents the ratio between the parameter change caused by the physical fluctuations of the associated node when the unit adjustment action is transmitted to the local node and the parameter change caused by the unit adjustment action of the local node. The calculation process of the feedforward compensation value consists only of table lookup index operation and linear algebra multiplication operation.

6. The adaptive cooperative control system for valve nodes in an industrial fluid pipeline network according to claim 1, characterized in that, The control unit also executes security degradation logic: it establishes a neighbor node heartbeat monitoring mechanism, and determines that the communication link with the associated node is failed when no heartbeat signal is received from a specific associated node within a preset period; it temporarily sets the coupling coefficient corresponding to the failed associated node in the neighborhood influence weight table to zero, and generates local control commands only based on the controlled physical parameters fed back by the local sensing unit.

7. The adaptive cooperative control system for valve nodes in an industrial fluid pipeline network according to claim 2, characterized in that, The discretized execution interval identifier divides the entire stroke of the controlled unit into three logical intervals: low load region, linear control region, and saturation region. The multidimensional gain scheduling matrix stores nine independent coupling coefficients corresponding to the source node and local node under the above different logical interval combinations, in order to compensate for the nonlinear flow gain characteristics of the actuator under different adjustment intervals.

8. The adaptive cooperative control system for valve nodes in an industrial fluid pipeline network according to claim 3, characterized in that, Residual ratio Calculated according to the following rules: ,in, The residual ratio, and These are the start and end times of the valid verification period, respectively. The actual change in the controlled physical parameters collected by the local sensing unit. This refers to the theoretically expected parameter changes calculated based on the forecast signal and the current coupling coefficient.

9. The adaptive cooperative control system for valve nodes in an industrial fluid pipeline network according to claim 1, characterized in that, The control unit includes a microcontroller unit, and the neighborhood influence weight table is stored in the internal memory of the microcontroller unit; the forecast signal is broadcast through the industrial network; the superposition operation of the feedforward compensation value is performed at the output register level of the microcontroller unit.

10. An adaptive cooperative control system for valve nodes in an industrial fluid pipeline network according to claim 1, characterized in that, The control unit executes a broadcast priority management strategy and marks the generated warning signal as a priority processing packet when the absolute value of the local target adjustment increment exceeds a preset mutation level threshold; the collaborative control system also includes a network switching device, which is configured to prioritize forwarding priority processing packets.

Images