Intelligent linkage control system for ventilation and drainage of underground pipe gallery

Abstract

The invention relates to the field of computer program control, and discloses an underground pipe gallery ventilation and drainage intelligent linkage control system which comprises a sensing module, a dynamic causal chain engine, a linkage strategy generation module and an execution module. A dynamic causal graph model based on multi-dimensional environmental parameters is constructed, a logic dependency relationship between ventilation and drainage is identified, and a multi-stage linkage control strategy is generated in combination with event driving, so that real-time cooperative control of ventilation and drainage is realized; the problems of resource waste and response lag caused by control path splitting and response logic static state in a traditional system are solved. Particularly, when scenes such as sudden water accumulation or gas accumulation are processed, the system can actively predict the environment evolution trend, the operation states of the draught fan and the water pump are regulated and controlled in advance, and the response intelligence and the environment adaptability of the control system are remarkably improved.

Description

Technical Field

[0001] The present invention relates to an intelligent linkage control system for underground pipe gallery ventilation and drainage, and belongs to the technical field of computer program control. Background Art

[0002] As a key component of urban infrastructure, the underground pipe gallery is widely used in scenarios such as the centralized layout and operation and maintenance of various pipelines including electricity, communication, water supply and drainage, and gas. To ensure its operation safety and environmental stability, a ventilation system and a drainage system are usually equipped to regulate air quality, handle leakage and waterlogging, etc. Under the traditional control mode, these two systems mostly operate independently. Each is based on sensor data and controls the start and stop of devices such as fans and pumps according to set thresholds, forming a typical passive trigger mechanism.

[0003] Existing control schemes mostly use PLC or SCADA platforms as carriers and rely on basic parameters such as temperature, humidity, and water level to trigger device responses. However, with the increase in the service life of the pipe gallery and the uncertainty of environmental changes, the traditional threshold control mode gradually exposes problems such as untimely response to complex coupling events and lack of strategy coordination. To improve the system intelligence, the industry has tried to introduce multi-sensor data fusion and simple strategy linkage logic, but there are still the following deficiencies in practice: S1: The ventilation system and the drainage system are independent of each other in control logic and lack the ability to conduct linkage analysis of environmental data; S2: Most control strategies rely on fixed thresholds and are difficult to adapt to instantaneous events or environmental changes. For example, when a leakage suddenly occurs, the system may delay the response because the threshold is not reached, missing the best control opportunity. Especially in old pipe galleries, the problems of sensor drift and parameter false alarms are more prominent, resulting in frequent false alarms and missed alarms; S3: The pipe gallery control system is usually built by different manufacturers in stages, the data platforms are fragmented, and the information between subsystems such as ventilation and drainage, fire protection and electricity is isolated, making it impossible to achieve global situation awareness and cross-system collaborative decision-making.

[0004] To alleviate the above problems, some systems have tried to set the response order for multiple systems through a rule engine or introduce a scenario judgment library based on expert knowledge. However, such methods generally rely on static rules and empirical presets, are difficult to cover complex coupling events and spatial distribution differences, and have high costs for later maintenance and updates, with a low implementation rate in engineering applications and limited practicality; Therefore, there is an urgent need for an intelligent control system for underground pipe gallery operation scenarios with logical self-adaptive ability, which can fuse multi-source perception data, dynamically model the causal relationship between ventilation and drainage, and generate forward-looking linkage control strategies in combination with the event evolution trend, so as to achieve real-time collaborative response to complex working conditions and improve the overall intelligence level of the control system. Summary of the Invention

[0005] The present invention provides an intelligent linkage control system for underground pipe gallery ventilation and drainage, and its main purpose is to solve the problems of separated ventilation and drainage control, response delay, and system data islands.

[0006] To achieve the above object, an intelligent linkage control system for underground pipe gallery ventilation and drainage provided by the present invention includes:

[0007] A sensing module for real-time collecting multi-dimensional environmental parameters in the underground pipe gallery, where the multi-dimensional environmental parameters at least include temperature T, humidity H, water level L, and gas concentration C g ;

[0008] A dynamic causal chain engine, communicatively connected to the sensing module, for receiving real-time data streams of the multi-dimensional environmental parameters and constructing and dynamically updating a dynamic causal graph model based on the real-time data streams. The dynamic causal graph model identifies and characterizes multiple associations between the ventilation and drainage based on logical dependency relationships. Among them, for the target ventilation volume V of the ventilation equipment tar get , it is at least partially based on the current humidity H and gas concentration C g and is dynamically determined by the following formula:

[0009] V tar get = α×H + β×C g ,

[0010] where α and β are weight coefficients dynamically adjusted according to historical environmental parameter data and a lightweight transfer learning algorithm, respectively representing the influence degrees of humidity and gas concentration on the target ventilation volume;

[0011] A linkage strategy generation module, communicatively connected to the dynamic causal chain engine, for real-time monitoring environmental events triggered by changes in the multi-dimensional environmental parameters, and generating an adaptive multi-stage linkage control strategy according to the dynamic causal graph model and the currently triggered environmental events. The linkage control strategy can drive the execution module to coordinately control the ventilation equipment and the drainage equipment according to the types of different environmental events and the logical dependency relationships characterized in the dynamic causal graph model, so as to solve the problem of cooperative failure easily caused by the existing independent control system in the multi-parameter coupling scenario;

[0012] An execution module, communicatively connected to the linkage strategy generation module, for receiving and executing control instructions in the linkage control strategy to adjust the operating states of the ventilation equipment and the drainage equipment, and feedbacking the execution states to the dynamic causal chain engine to form a closed-loop control.

[0013] Preferably, the dynamic causal chain engine further includes a lightweight transfer learning module. The lightweight transfer learning module trains a lightweight neural network using historical environmental parameter data and dynamically optimizes the weights of the logical dependencies in the dynamic causal graph model based on the training results, so that the system can automatically adapt to the environmental characteristics of different underground utility tunnel sections and reduce the need for manual parameter adjustment.

[0014] Preferably, the linkage strategy generation module further includes a digital twin model. After generating the linkage control strategy, the digital twin model previews the execution effect of the linkage control strategy in the virtual utility tunnel environment based on the current utility tunnel environmental parameters and equipment status, and dynamically corrects the control instructions according to the preview results to improve the control accuracy and reduce the risk of misoperation caused by static threshold triggering.

[0015] Preferably, the sensing module includes multiple edge computing nodes. The edge computing nodes are deployed in different areas of the underground utility tunnel and are used to perform real-time compression processing on the multi-dimensional environmental parameters collected nearby. After extracting the event features representing environmental changes, the event features rather than the original data are sent to the dynamic causal chain engine to reduce the data transmission load and improve the system response speed.

[0016] Preferably, the dynamic causal chain engine is also configured to support cross-system data fusion, receive and analyze relevant data from the fire monitoring system and the power monitoring system, and synchronously push the generated linkage control instructions to the fire monitoring system and the power monitoring system to achieve multi-system collaborative emergency response. For example, when draining water, the high-risk circuits are synchronously shut down.

[0017] Preferably, the dynamic causal chain engine further includes an event dynamic priority mapping mechanism. When multiple concurrent instantaneous environmental events are detected, the event dynamic priority mapping mechanism dynamically assigns priorities to each event based on preset risk assessment rules and the weights of the logical dependencies in the dynamic causal graph model, and preferentially triggers the linkage control strategies related to high-priority events to cope with complex and sudden changes in the utility tunnel environment.

[0018] Preferably, when assigning event priorities, the event dynamic priority mapping mechanism also considers the physical topology of the underground utility tunnel and the spatio-temporal correlation of the environmental parameters collected by the sensing module, and constructs an event energy field model. Among them, for any environmental event E, its energy field intensity I E is at least partially related to the degree ΔS of deviation of the environmental parameters measured by each sensor within the influence range of the event from the normal value i and the distance d from the event source i For example, it is calculated by the following formula:

[0019]

[0020] Among them, w i is a weight coefficient related to the sensor type and location, used to characterize the contribution degree of different sensor data to the intensity of the event energy field.

[0021] Compared with the problems described in the background technology, the beneficial effects of the present invention are as follows:

[0022] 1. By constructing a dynamic causal chain model, the deep linkage between ventilation control and drainage control at the strategy generation level is realized, effectively solving the problem that the ventilation and drainage control paths in the traditional system are independent of each other and cannot respond collaboratively based on the environmental evolution trend. During the actual operation process, when the system monitors humidity fluctuations caused by, for example, rising water levels, it no longer only triggers a single drainage operation, but based on the logical relationship between environmental parameters, dynamically predicts subsequent environmental risks such as possible gas accumulation and oxygen concentration decrease, and drives the ventilation equipment to intervene in advance, thereby avoiding equipment corrosion or abnormal air quality caused by the failure to adjust ventilation in time after drainage. This mechanism strengthens the response ability of the control system to multi-parameter coupling scenarios, reflects the evolution of the control strategy from static response to intelligent prediction linkage, and has high control intelligence and overall system optimization ability.

[0023] 2. Through the lightweight transfer learning mechanism, it can automatically adapt to the environmental characteristics of different pipe gallery areas, effectively alleviating the problem of control failure caused by fixed thresholds and changing environments. For example, in old pipe galleries, the traditional system often misjudges humidity fluctuations as occasional situations, while this system can dynamically identify the coupling trend of continuous humidity increase and water level change by combining historical environmental evolution characteristics, and timely adjust the dependence weights in the causal chain model to generate more targeted control strategies, thereby improving the adaptability of the system in diverse environments. At the same time, the digital twin model introduced in the system endows the control strategy with the ability of virtual pre-play before execution, enabling it to judge the applicability of the strategy without relying on static preset rules. After the strategy is generated, the system can pre-play the influence path of the control action in the virtual pipe gallery environment, combine the current environmental state and equipment parameters, dynamically simulate subsequent changes such as humidity diffusion and gas distribution, and then correct the strategy execution logic.

[0024] 3. The system introduces edge computing nodes at the perception layer, which can preliminarily compress sensor data and extract event features nearby, effectively reducing the burden of redundant data transmission on the network and improving the overall control response speed. During actual operation, in case of local sudden water accumulation, gas leakage and other events, the edge node can real-time identify key change patterns and quickly report to the decision-making layer, enabling the system to complete local response without relying on the operation of the central node. This mechanism not only improves the response efficiency of the control system to emergencies, but also improves the operation stability of the system under weak signal or complex deployment conditions, showing good engineering deployment adaptability. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Figure 1 FIG. 1 is a schematic structural diagram of the dynamic causality diagram model of the present invention.

[0026] Figure 2 FIG. 2 is a flowchart of the coordinated control of the drainage task execution and power monitoring of the present invention.

[0027] Figure 3 FIG. 3 is a diagram of the dynamic priority mapping and policy conflict resolution mechanism of the present invention.

[0028] The implementation, functional features and advantages of the present invention will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0029] It should be understood that the specific embodiments described herein are only for explaining the present invention and are not used to limit the present invention.

[0030] An embodiment of the present application provides an intelligent linkage control system for underground pipe gallery ventilation and drainage, including:

[0031] A sensing module for real-time collecting multi-dimensional environmental parameters in the underground pipe gallery, where the multi-dimensional environmental parameters at least include temperature T, humidity H, water level L, and gas concentration C g ;

[0032] A dynamic causality chain engine, communicatively connected to the sensing module, for receiving the real-time data stream of the multi-dimensional environmental parameters and constructing and dynamically updating a dynamic causality diagram model based on the real-time data stream. The dynamic causality diagram model identifies and characterizes multiple associations between the ventilation and drainage based on logical dependency relationships. Among them, for the target ventilation volume V of the ventilation equipment target , it is at least partially based on the current humidity H and gas concentration C g and is dynamically determined by the following formula:

[0033] V target = α×H + β×C g ,

[0034] where α and β are weight coefficients dynamically adjusted according to historical environmental parameter data and a lightweight transfer learning algorithm, respectively representing the influence degrees of humidity and gas concentration on the target ventilation volume;

[0035] The linkage strategy generation module is communicatively connected to the dynamic causal chain engine, and is configured to monitor in real time the environmental events triggered by the changes in the multi-dimensional environmental parameters, and generate an adaptive multi-stage linkage control strategy according to the dynamic causal graph model and the currently triggered environmental event. The linkage control strategy can drive the execution module to jointly control the ventilation equipment and the drainage equipment according to the types of different environmental events and the logical dependency relationships characterized in the dynamic causal graph model, so as to solve the problem of collaborative failure easily caused by the existing independent control system in the multi-parameter coupling scenario;

[0036] The execution module is communicatively connected to the linkage strategy generation module, and is configured to receive and execute the control instructions in the linkage control strategy, so as to adjust the operating states of the ventilation equipment and the drainage equipment, and feed back the execution state to the dynamic causal chain engine to form a closed-loop control.

[0037] Preferably, the dynamic causal chain engine further includes a lightweight transfer learning module. The lightweight transfer learning module trains a lightweight neural network using historical environmental parameter data, and dynamically optimizes the weights of the logical dependency relationships in the dynamic causal graph model based on the training results, so that the system can automatically adapt to the environmental characteristics of different underground utility tunnel sections and reduce the need for manual parameter adjustment.

[0038] Preferably, the linkage strategy generation module further includes a digital twin model. The digital twin model is configured to, after generating the linkage control strategy, preview the execution effect of the linkage control strategy in the virtual utility tunnel environment based on the current utility tunnel environmental parameters and equipment states, and dynamically correct the control instructions according to the preview results, so as to improve the control accuracy and reduce the risk of misoperation caused by static threshold triggering.

[0039] Preferably, the sensing module includes a plurality of edge computing nodes. The edge computing nodes are deployed in different areas of the underground utility tunnel, and are configured to perform real-time compression processing on the multi-dimensional environmental parameters collected nearby, and extract the event features characterizing the environmental changes, and then send the event features instead of the original data to the dynamic causal chain engine, so as to reduce the data transmission load and improve the system response speed.

[0040] Preferably, the dynamic causal chain engine is further configured to support cross-system data fusion, receive and analyze the relevant data from the fire monitoring system and the power monitoring system, and synchronously push the generated linkage control instructions to the fire monitoring system and the power monitoring system, so as to achieve multi-system collaborative emergency response, such as synchronously shutting down high-risk circuits during drainage.

[0041] Preferably, the dynamic causal chain engine further includes an event dynamic priority mapping mechanism, which is used to, when detecting multiple concurrent instantaneous environmental events, dynamically assign priorities to each event based on preset risk assessment rules and the weights of various logical dependencies in the dynamic causal graph model, and preferentially trigger linkage control strategies related to high-priority events to cope with complex and sudden changes in the utility tunnel environment.

[0042] Preferably, when assigning event priorities, the event dynamic priority mapping mechanism also considers the physical topology of the underground utility tunnel and the spatio-temporal correlation of the environmental parameters collected by the sensing module, and constructs an event energy field model. Among them, for any environmental event E, its energy field intensity I E is at least partially related to the degree ΔS of deviation of the environmental parameters measured by each sensor within the influence range of the event from the normal value i and the distance d from the event source i For example, it is calculated by the following formula:

[0043]

[0044] where w i is a weight coefficient related to the sensor type and location, and is used to characterize the contribution degree of different sensor data to the event energy field intensity.

[0045] Preferably, it further includes an adaptive resolution strategy module, which is used to, when detecting that the energy fields of multiple concurrent events overlap and may cause control conflicts or resource competition, automatically adjust the linkage control strategy or allocate execution resources according to the types of different events and preset resolution rules, for example, preferentially process events with stronger physical destructiveness to ensure the stable operation and safety of the system.

[0046] Preferably, when training and optimizing the dynamic causal graph model by using historical environmental parameter data, the lightweight transfer learning module also takes the topological features of the underground utility tunnel as important input parameters, such as the number of branch pipelines and the distribution density of equipment, so that the lightweight neural network can adapt to the physical characteristics of different underground utility tunnel sections faster and more accurately, and improve the generalization ability of the model.

[0047] Preferably, when constructing the event energy field model and calculating the event energy field intensity, the dynamic causal chain engine further considers the spatial distance between key equipment and potential risk sources, such as the distance between the leakage point and electrical equipment. The closer the distance, the higher the priority of the event related to the risk is increased, and the linkage control strategy is adjusted. For example, when leakage occurs in the pipe section near the electrical equipment, the system automatically raises the priority of drainage in this area and synchronously reduces the ventilation intensity to avoid water vapor diffusion to the equipment area.

[0048] Example 1: During operation, the sensing module collects multi-dimensional environmental parameters in the current utility tunnel environment, including humidity (denoted as H) and gas concentration (denoted as C g ), and this data is transmitted to the dynamic causal chain engine in real time. To achieve the logical coupling of ventilation control and drainage control, the engine dynamically maintains a causal graph model constructed based on logical dependencies.

[0049] The target ventilation volume V target is calculated based on the following relationship:

[0050] V target = α·H + β·C g ,

[0051] where the variable V target represents the target air volume that the ventilation equipment should reach at the current moment, with the unit of cubic meters per minute (m 3 / min), and its value determines the start-stop state and operating power of the fan. This target air volume is the direct output of the control strategy and is used to drive the speed control module of the fan in the execution module.

[0052] The variable H represents the humidity collected by the sensing module, with the unit of relative humidity percentage (%RH), and its source is the humidity sensors deployed at key nodes of the utility tunnel; the variable C g represents the gas concentration. Generally, multiple groups of gas sensors can be used to detect different types of suspicious gases (such as hydrogen sulfide, methane, etc.), and after normalization processing, a unified concentration evaluation index is formed, with the unit of ppm.

[0053] The parameters α and β are the influence weights corresponding to humidity and gas concentration, with dimensionless characteristics, and the value range is positive real numbers. Their values are derived from the learning and modeling process of historical environmental parameters and air volume control records. To achieve the adaptability of the model under different underground utility tunnel structures and environmental characteristics, this example introduces a lightweight transfer learning mechanism: the system periodically collects historical operation data from edge computing nodes, trains a shallow neural network model locally, and combines the existing weight parameters for transfer and update. Finally, the characteristics of the current utility tunnel section are used as prior inputs to dynamically adjust α and β.

[0054] For example, in an old area with a slow response of the ventilation system, the system will automatically increase the relative value of β to enhance the driving ability of gas concentration changes on the target air volume; while in an area with long-term high humidity, the proportion of α will be increased to enhance the response to humidity fluctuations.

[0055] After calculating V targetAfterward, the linkage strategy generation module matches the target air volume with the current equipment operating status. If the difference exceeds a threshold, a control instruction is generated, and the execution module adjusts the fan operating intensity. Furthermore, if the drainage equipment is currently operating at high load, the system will also integrate strategies based on the current drainage flow rate, water level change rate, and event type to coordinate fan and pump operations to prevent water vapor diffusion and energy waste.

[0056] To further enhance the system's ability to respond to sudden environmental events, this embodiment also clarifies the calculation mechanism of the energy field model. When detecting multiple concurrent events, the dynamic causal chain engine calculates the event energy field intensity according to the following formula:

[0057]

[0058] Among them, the variable ΔS i Indicates the deviation between the parameter currently measured by the i-th sensor and its historical stable interval; variable d i Indicates the spatial distance between the sensor and the initial point of the event (in meters); variable w i It is a weight factor defined according to the sensor type and deployment location, which is used to characterize the importance of the sensor data to the overall judgment.

[0059] The formula aims to quantify the spatial intensity of current environmental events to determine their control priority. The calculation process is initially completed by the edge node and uploaded to the central node, where it is used to update the event scheduling priority mapping table and drive the priority adjustment of the control strategy. For example, if a region experiences both a sharp increase in gas concentration and a rapid rise in water levels, and that region is located at the intersection of multiple trunk lines, the system will automatically assign a higher linkage response level due to the higher spatial energy field intensity, prioritizing the combined ventilation and drainage control strategy.

[0060] To verify the adaptability and safety of the strategy, the digital twin model performs a real-time preview in a simulation environment based on the current environmental parameters after generating control instructions. This process includes the following steps: (1) reading the latest environmental status snapshot; (2) inputting the control instructions into the virtual environment model; (3) simulating the parameter evolution path after the strategy is executed, such as humidity diffusion trend and wind direction adjustment; (4) comparing it with the expected control target; and (5) if potential deviation or execution conflict is found, rolling back and optimizing the strategy in a timely manner.

[0061] Through the mechanism arrangement of this embodiment, the ventilation and drainage control strategy has realized the transition from a single static threshold to a multi-dimensional data-driven linkage decision-making. More importantly, the system no longer relies on artificially preset rules for emergency response, but through four levels of data perception, causal deduction, dynamic adjustment, and feedback closed-loop, the system realizes the active recognition and adaptive adjustment of the environmental situation, all of which belong to the extended implementation methods known to those of ordinary skill in the art.

[0062] Embodiment 2: To further illustrate the relationship and operation mechanism between the key modules of the intelligent linkage control system for underground pipe gallery ventilation and drainage described in the present invention, in combination with Figures 1 to 3 , the core structure and control process of the system are further described.

[0063] As Figure 1 shown, the dynamic causal graph model in the system is the core module for realizing multi-dimensional environmental parameter-driven strategy control. This model includes environmental parameter nodes, equipment control nodes, logical dependency edges, and a logical optimization mechanism. Among the environmental parameter nodes, there are two dimensions of temperature and humidity, which can continuously model the real-time changing environmental state in the underground pipe gallery. The equipment control nodes include the water pump power and the exhaust fan rate, which are used to receive strategy instructions and control the execution behavior of specific equipment. The logical dependency edges include the dependency relationship between temperature and the fan, and the causal relationship between water level and the water pump. These dependency edges reflect the dynamic linkage logic between environmental changes and equipment responses. The logical optimization part includes the linkage optimization between the water level and the water pump and the dynamic weight adjustment of the transfer learning mechanism to ensure that the system strategy model can adaptively adjust the weights according to the historical operation characteristics of different regions. During operation, the event dynamic priority mapping module re-prioritizes each task according to the event type and risk level, and transmits the updated priority to the dynamic causal graph model to adjust the calculation weights of the causal nodes in real time. Finally, the dynamic causal graph model provides simulation parameters to the digital twin model to support the preview and evaluation of the control strategy in the virtual environment.

[0064] As Figure 2As shown, when faced with the execution of the drainage task, the linkage system will first send a request to the power monitoring module to query the status of high-risk circuits, and determine whether the drainage behavior may affect the safe operation of the current high-voltage power line. After the power monitoring module completes the judgment, it will return a power-off permission to the linkage system. If the drainage operation is allowed, the linkage system will send an instruction to start the water pump (with power P) to the drainage equipment, where the power P is an operating parameter dynamically determined according to the current environmental status and strategy output. After receiving the instruction, the drainage equipment starts the water pump and feeds back the execution status to the power monitoring module for the system to complete the control closed-loop. Subsequently, the linkage system will synchronize the drainage area according to the water level change trend in the pipe gallery and the actual pumping range of the water pump, realizing precise zoning control to ensure that the drainage behavior is only carried out within the electrical safety area and avoiding risks to other systems.

[0065] As Figure 3 shown, to address the policy conflict problem brought about by the concurrent triggering of multiple events, the system designs an event dynamic mapping mechanism. This mechanism takes risk assessment as the core, integrates two types of emergency events: abnormal water level and gas leakage, and quantifies the impact degree of the events through energy field calculation and spatial attenuation coefficient analysis methods respectively. In the risk assessment module, the system uniformly measures the influence of the two types of events and calculates the risk level, and determines the priority ranking and parameter weight adjustment according to the results. The event with a higher risk level will be responded to by the system first. Among them, the abnormal water level will drive the drainage strategy generation module to generate a specific drainage plan, and the gas leakage will affect the ventilation strategy correction module through weight adjustment, thereby dynamically adjusting the air volume, direction and execution logic. Through this mapping mechanism, the system can take into account both response efficiency and operation safety when facing emergencies, and improve the accuracy and adaptability of the overall control strategy.

[0066] Example 3: During the specific operation process, the sensing module receives multi-dimensional environmental parameters such as humidity (denoted as H) and gas concentration (denoted as C g ) in real time from edge computing nodes deployed in different regions. The dynamic causal chain engine calculates the target ventilation volume V of the ventilation equipment based on these real-time data streams target . This calculation follows the following relational expression:

[0067] V target = α·H + β·C g ,

[0068] where: The unit of V target is m 3 / min, indicating the target ventilation volume that the fan should output currently; H is the humidity value collected in real time, with the unit of percentage (%RH), sourced from distributed humidity sensors; C gIt is synthesized after normalizing the detection results of gases such as hydrogen sulfide and methane to unify the concentration index, with the unit of ppm; α and β are weight parameters adaptively adjusted by the system, which are used to dynamically quantify the contribution degree of each environmental factor to the ventilation demand, and the unit is dimensionless.

[0069] To achieve a control strategy with strong adaptability and rapid response, the system continuously adjusts the above weight parameters through a lightweight transfer learning algorithm. The training process relies on the historical environmental data cached locally at the edge node, combines the current regional physical topology (such as sensor density, duct distribution), and periodically updates the neural network model. Position information is added as an auxiliary input variable to the model, enabling the system to not only adjust α and β based on historical trends but also consider the equipment response differences caused by spatial structure differences, enhancing the adaptation ability of the causal graph between physical pipe sections.

[0070] For example, in the pipe section where the ventilation facilities are aging and gas accumulation occurs frequently, the system can automatically increase the weight of β, so that the gas concentration change obtains a higher priority in the target air volume calculation. In the area that has been in a humid environment for a long time, α is dynamically increased to respond more quickly to the impact of humidity fluctuations on the ventilation system.

[0071] Before executing the control strategy, the linkage strategy generation module compares the currently planned output V target with the current operating state of the equipment. If the deviation exceeds the preset response threshold, the fan speed regulation module will be triggered to issue a control instruction; in addition, to avoid possible resource conflicts or execution errors during the execution of the control strategy, the system conducts a rehearsal through a digital twin model. This model is established based on the real tunnel environment parameters and includes the following operation steps: taking the currently collected environmental parameters (such as H, C g , L, etc.) and the current power of the fan and water pump as inputs; calling the physical behavior modeling module to simulate the dynamic trends of air volume change, humidity diffusion path, and gas concentration after executing the control strategy based on empirical rules; comparing the output fitting curve with the threshold setting interval to determine whether the strategy may cause overshoot, resource competition, or control delay; if there are risk points, the current control path will be rolled back through the strategy reconstruction logic, and a better solution will be re-matched and the execution instruction will be issued. This mechanism ensures that the system not only has the ability to output strategies but also has the ability of forward-looking adjustment and feedback correction, strengthening its operation stability in scenarios of multi-parameter coupling and equipment collaboration.

[0072] When dealing with multiple concurrent emergencies (such as water accumulation and gas leakage occurring simultaneously), the event dynamic priority mapping mechanism will calculate the energy field intensity I E . In the calculation method, each parameter deviation value ΔS iDerived from the difference between the sensor and its historical steady-state range; position parameter d i Obtained by calculating the real-time distance between the sensor coordinates reported by the edge node and the location of the initial event point; weight coefficient w i Preset in the system initialization stage and regularly corrected according to the sensor accuracy and regional importance during actual operation. The calculation result of the event energy field is not only used to sort the event response priorities, but also directly participates in the weight distribution in the process of generating the linkage strategy. For example, the implementation details such as improving the response time window of the fan in the risk area and delaying the startup of equipment in the non-critical area all belong to the extended implementation methods that can be known to those of ordinary skill in the art.

[0073] Example 4: When calculating the target ventilation volume Vtarget of the ventilation equipment in this example, humidity and gas concentration are used as key influencing factors, and the system dynamically adjusts the weight coefficients α and β through a lightweight transfer learning algorithm. Specifically, the system uses the edge computing node to periodically train a shallow neural network model based on historical environmental data and regional physical characteristics. This model dynamically adjusts the weight coefficients according to the climate change trend, sensor data changes, and equipment response characteristics in the current region. For example, in areas with relatively drastic humidity changes, the system will automatically increase the value of the weight coefficient α, making the impact of humidity changes on the ventilation volume more sensitive; in areas with large gas concentration changes, the system will increase the weight coefficient β to enhance the driving ability of gas concentration on the air volume adjustment. This dynamic adjustment mechanism ensures that the system can flexibly respond under different environmental conditions, improving the control accuracy and response speed.

[0074] In the target ventilation volume V target = α·H + β·C g , in the calculation, the variables H and C g respectively represent the humidity and gas concentration in the current corridor environment. α and β are dynamically adjusted weight coefficients, and their values are derived from the training results of historical environmental parameter data. The humidity H is collected by the humidity sensors arranged in the corridor and is expressed as a percentage; the gas concentration C g is obtained by normalizing the detection results of multiple groups of gas sensors for gases such as hydrogen sulfide and methane, and the unit is ppm. The values of α and β are dynamically optimized by the transfer learning model and adaptively adjusted based on historical data and environmental characteristics. This calculation path ensures that under different environmental conditions, the ventilation volume can be accurately adjusted in real time according to actual needs, avoiding the response delay or inaccuracy problems caused by fixed thresholds in traditional systems.

[0075] To improve the accuracy of the control strategy and the feasibility of execution, in this embodiment, a digital twin model is used to virtually pre-evaluate the execution effect of the control strategy. The specific operation steps include: First, the current environmental parameters collected from the sensing module are combined with the operating state of the device and input into the digital twin model; then, the model simulates the changes in various environmental parameters after the execution of the strategy according to empirical rules and physical modeling methods, such as the diffusion path of humidity, the change trend of gas concentration, etc.; finally, the pre-evaluation result is compared with the control target. If potential deviations or resource conflicts are found in the strategy, the system will automatically roll back and optimize the control strategy to ensure the adaptability and safety before the implementation of the plan.

[0076] In a complex environment with multiple concurrent events, in this embodiment, an event energy field model and a dynamic priority mapping mechanism are used to achieve the priority management of different environmental events. The system first calculates the energy field intensity IE of each event, and the formula is:

[0077]

[0078] where ΔS i represents the deviation value of the parameter currently measured by the i-th sensor from its historical stable interval, d i is the distance between the sensor and the event source, and w i is the weight coefficient related to the sensor type and location. Through this calculation, the system can quantify the spatial intensity of each event and dynamically assign the priority of each event according to the calculation result. High-priority events will trigger the linkage control strategy first, so as to ensure that the system can respond in a timely manner in case of emergencies and avoid control conflicts or resource waste.

[0079] When multiple events occur, the system will sort the events according to the energy field intensity and give priority to handling the events with greater impact. For example, when the water level suddenly rises and the gas concentration changes violently, the system will give priority to starting the drainage strategy and adjusting the operation of the ventilation system to prevent water vapor from diffusing to the key equipment area. Based on this, when multiple events are detected and the energy fields overlap, the system will automatically adjust the linkage control strategy or reallocate the execution resources according to the event type and the preset resolution rules. For example, when the drainage event and the fire alarm event occur concurrently, the system will automatically reduce the operation priority of the ventilation system to ensure the proper priority and resource allocation of the drainage operation, thus avoiding potential equipment conflicts or operation errors.

[0080] For those skilled in the art, it is obvious that the present invention is not limited to the details of the above exemplary embodiments, and without departing from the spirit or basic characteristics of the present invention, the present invention can be implemented in other specific forms.

[0081] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An intelligent linkage control system for underground pipe gallery ventilation and drainage, characterized in that, Including: A sensing module, which is used to collect multi-dimensional environmental parameters in the underground pipe gallery in real time, and the multi-dimensional environmental parameters at least include temperature T, humidity H, water level L, and gas concentration C g ; A dynamic causal chain engine, communicatively connected to the perception module, for receiving a real-time data stream of the multi-dimensional environmental parameters and constructing and dynamically updating a dynamic causal graph model based on the real-time data stream, the dynamic causal graph model identifying and characterizing multiple associations between the ventilation and the drainage based on logical dependencies, wherein, for a target ventilation volume V of the ventilation device target , which is at least partially based on the current humidity H and the gas concentration C g and is dynamically determined by the following formula: V targeet = α × H + β × C g , Wherein, α and β are weight coefficients dynamically adjusted according to historical environmental parameter data and lightweight transfer learning algorithms, respectively representing the influence degrees of humidity and gas concentration on the target ventilation volume; A linkage strategy generation module, communicatively connected to the dynamic causal chain engine, for real-time monitoring of environmental events triggered by changes in the multi-dimensional environmental parameters, and generating an adaptive multi-stage linkage control strategy according to the dynamic causal graph model and the currently triggered environmental event. The linkage control strategy can drive the execution module to cooperate in controlling ventilation equipment and drainage equipment according to the types of different environmental events and the logical dependency relationships characterized in the dynamic causal graph model; An execution module, communicatively connected to the linkage strategy generation module, for receiving and executing the control instructions in the linkage control strategy to adjust the operating states of the ventilation equipment and the drainage equipment, and feeding back the execution states to the dynamic causal chain engine to form a closed-loop control.

2. The intelligent linkage control system for underground pipe gallery ventilation and drainage according to claim 1, characterized in that The dynamic causal chain engine further includes a lightweight transfer learning module. The lightweight transfer learning module trains a lightweight neural network using historical environmental parameter data and dynamically optimizes the weights of the logical dependency relationships in the dynamic causal graph model based on the training results, so that the system can automatically adapt to the environmental characteristics of different underground utility tunnel sections.

3. The intelligent linkage control system for underground pipe gallery ventilation and drainage according to claim 1 or 2, characterized in that, The linkage strategy generation module further includes a digital twin model. The digital twin model is used to, after generating the linkage control strategy, preview the execution effect of the linkage control strategy in a virtual utility tunnel environment based on the current utility tunnel environmental parameters and equipment states, and dynamically correct the control instructions according to the preview results.

4. The intelligent linkage control system for underground pipe gallery ventilation and drainage according to any one of claims 1-3, characterized in that, The sensing module includes multiple edge computing nodes deployed in different areas of the underground utility tunnel. The edge computing nodes are used for real-time compression processing of the multi-dimensional environmental parameters collected nearby, and after extracting the event features characterizing environmental changes, sending the event features rather than the original data to the dynamic causal chain engine.

5. The intelligent linkage control system for underground pipe gallery ventilation and drainage according to any one of claims 1-4, characterized in that, The dynamic causal chain engine is further configured to support cross-system data fusion, receive and analyze relevant data from the fire monitoring system and the power monitoring system, and synchronously push the generated linkage control instructions to the fire monitoring system and the power monitoring system.

6. The intelligent linkage control system for underground pipe gallery ventilation and drainage according to claim 1, wherein, The dynamic causal chain engine further includes an event dynamic priority mapping mechanism. The event dynamic priority mapping mechanism is used to, when detecting multiple concurrent instantaneous environmental events, dynamically assign priorities to each event based on preset risk assessment rules and the weights of the logical dependency relationships in the dynamic causal graph model, and preferentially trigger the linkage control strategies related to high-priority events.

7. The intelligent linkage control system for underground pipe gallery ventilation and drainage according to claim 6, characterized in that, When allocating event priorities, the event dynamic priority mapping mechanism also considers the physical topology of the underground utility tunnel and the spatio-temporal correlation of the environmental parameters collected by the sensing module, and constructs an event energy field model. Among them, for any environmental event E, its energy field intensity I E is at least partially related to the degree ΔS by which the environmental parameters measured by each sensor within the influence range of the event deviate from the normal value i and the distance d from the event source i are related. For example, it is calculated by the following formula: where, w i is a weight coefficient related to the sensor type and location, and is used to characterize the contribution degree of different sensor data to the intensity of the event energy field.

Images