A water conservancy project humanoid robot inspection and hidden trouble investigation system based on polymorphic fusion

Abstract

The application discloses a kind of based on polymorphism fusion's water conservancy engineering humanoid robot inspection and hidden danger investigation system, including water conservancy engineering body robot, based on polymorphism fusion's water conservancy information architecture unit, inspection and hidden danger investigation unit;Based on polymorphism fusion's water conservancy information architecture unit is used to collect the sensor data of water conservancy engineering and water conservancy hidden danger knowledge graph data, carries out data operation and early warning;Inspection and hidden danger investigation unit contains several inspection paths, collects corresponding path information, carries out hidden danger intelligent diagnosis.The application is based on the module setting of water conservancy engineering body robot, combines the model prediction control MPC of layered reinforcement decision framework and optimizes the inspection path algorithm of optimization inspection path;Integrate structured, semi-structured, unstructured data tease out potential risk elements and its correlation in water conservancy engineering, realize hidden danger traceability analysis, risk prediction and cross-domain knowledge fusion, provide intelligent decision support for water conservancy engineering whole life cycle risk management.

Description

Technical Field

[0001] This invention relates to the field of intelligent operation and maintenance technology for water conservancy projects, and in particular to a humanoid robot inspection and hidden danger investigation system for water conservancy projects based on multi-state fusion. Background Technology

[0002] Traditional water conservancy project inspections rely on manual methods such as visual inspection, touching, and sound, which have the following problems:

[0003] 1. High-risk scenario limitations: Manual inspection of the backwater surface of dams, deep-water gates, and narrow tunnels is inefficient and highly dangerous;

[0004] 2. Poor adaptability to complex environments: Existing underwater robots have difficulty simultaneously detecting structural leaks (such as piping and cracks) and hydrological parameters (flow velocity and sediment deposition);

[0005] 3. Fragmented multi-source data: The lack of deep integration between structural health monitoring (strain, displacement), hydrological sensing (water level, flow rate) and visual data leads to insufficient analysis of the correlation between potential hazards.

[0006] Therefore, it is necessary to research and design a more intelligent multi-mode integrated inspection and hidden danger investigation system for water conservancy projects. Summary of the Invention

[0007] To address the aforementioned problems, this invention provides a humanoid robot inspection and hazard investigation system for water conservancy projects based on multi-state fusion, comprising a water conservancy engineering embodied robot, a water conservancy information architecture unit based on multi-state fusion, and an inspection and hazard investigation unit.

[0008] The amphibious robot for water conservancy projects includes an amphibious control module, a multimodal perception kit, and a communication-decision module. The amphibious control module is used to check and control the movement of the amphibious robot for water conservancy projects. The multimodal perception kit is used for the amphibious robot for water conservancy projects to detect safety data of water conservancy projects. The communication-decision module is used for the amphibious robot for water conservancy projects to transmit data in real time with the shore-based fog gateway and perform hierarchical enhanced decision-making calculations.

[0009] The water conservancy information architecture unit based on multi-state fusion is used to collect sensor data and water conservancy hidden danger knowledge graph data of water conservancy projects, and to perform data calculation and early warning.

[0010] The inspection and hazard investigation unit collects relevant path information through several inspection paths, and performs intelligent hazard diagnosis based on the inspection results and information collection.

[0011] As a further explanation of the present invention, the amphibious body control module includes a biomimetic adaptive body and a body perception sensor;

[0012] The biomimetic adaptive body adopts a modular joint design, including waterproof servo motors and magnetically attached feet, and its posture stability is monitored in real time by an inertial measurement unit (IMU).

[0013] The body sensing sensors are embedded with strain gauges and torque sensors in the robot joints to monitor the load status of the robotic arm in real time; the feet integrate a pressure sensor array to dynamically sense the ground hardness and slip risk.

[0014] Furthermore, the multimodal sensing kit includes an underwater kit, a detection kit, and a problem-solving kit;

[0015] The underwater kit includes a waterproof sonar array and a miniature water quality sensor;

[0016] The detection kit includes a microscopic camera and a conductivity probe mounted on the end of a robotic arm, which detects the depth of concrete carbonation and steel corrosion through contact.

[0017] The problem handling kit includes a flexible gripper based on tactile feedback for grasping underwater foreign objects or collecting sediment samples.

[0018] Furthermore, the communication-decision module includes a basic communication module and a hierarchical enhanced decision-making framework;

[0019] The basic communication module uses an underwater acoustic communication module to transmit data in real time with a shore-based fog gateway.

[0020] The hierarchical enhanced decision-making framework optimizes inspection paths based on model predictive control (MPC), avoids high-flow areas or structurally vulnerable areas, and dynamically switches tasks according to the priority of potential hazards.

[0021] Furthermore, the water conservancy information architecture unit based on multi-modal fusion includes a water conservancy multi-modal sensor array composed of a detection module, a monitoring module, and an environmental module.

[0022] Furthermore, the inspection path includes a land mode path, an underwater mode path, and a climbing mode path;

[0023] The land-based route detects potential structural hazards by walking along the dam crest, horse trail, and sluice gate crest.

[0024] The underwater mode path involves diving to the bottom of the reservoir to detect seepage and siltation in the dam foundation;

[0025] The climbing mode path involves adhering to the back of the gate to check for hinge corrosion and sealing failure; or adhering to the culvert to check for culvert corrosion, leakage, and the flatness and smoothness of the culvert surface.

[0026] Furthermore, the water conservancy information architecture unit based on polymorphic fusion includes a scenario-monitoring data-knowledge base analysis model, and the water conservancy hidden danger knowledge graph is obtained by analysis and reasoning from the scenario-monitoring data-knowledge base analysis model.

[0027] Furthermore, the hierarchical reinforcement decision operation includes the following steps:

[0028] Step 1: Initial State Acquisition. Input the initial state x0 of the embodied robot and acquire the state value y from the sensor. k ;

[0029] Step 2: Set up state estimation, and use Kalman filtering to fuse multi-sensor data to estimate the robot's position and state at time k;

[0030]

[0031] Where: μ k-1 To control the input, Q noise External and system interference noise;

[0032] Step 3: Construct a prediction model. Based on the robot's original patrol route and current position, obtain the robot's motion reference trajectory position x at time point k+i. ref According to the state x at time k k and μ k+ik Predict the robot's position at the next moment based on the control input.

[0033]

[0034] In the formula: A i Let A be the system matrix at time i. j Let B be the system matrix at time j, and let B be the adjustment matrix. The state equations are discretized using the Euler method.

[0035] Step 4: Rolling optimization solution: Based on the project monitoring situation, set constraints and solve the optimization function simultaneously;

[0036]

[0037] U * =argminJ(x k ,U)ST(constraints)

[0038] In the formula: r is the control penalty, to prevent drastic changes in the input control of state N-1; μ N-1 The control input is defined at time N-1, where U* represents the optimal input control and U represents the initial input control.

[0039] Step 5: Feedback and correction to determine if the objective has been achieved. If so, proceed to the next step; otherwise, recalculate.

[0040] Furthermore, the intelligent diagnosis of hidden dangers adopts a dynamic evaluation method with fuzzy priority, which includes evaluation of the risk level of the hidden danger and evaluation of the urgency of its treatment.

[0041] Furthermore, the calculation formula for the risk level assessment of the hazard is R = α × hazard type + β × monitoring status.

[0042] In the formula: α and β are weighting coefficients, α+β=1, α≥0, β≥0;

[0043] The formula for calculating the urgency assessment of the hazard handling is as follows:

[0044] In the formula: t is the current time, t d The statutory time limit for hazard remediation;

[0045] Fuzzy dynamic weight adjustment:

[0046]

[0047] In the formula, μ i (t) is an expert score assigned based on the potential risks and the urgency of their handling, and is trained and stored in a software library using machine learning;

[0048] Priority calculation

[0049] Fuzzy priority calculation:

[0050] x i =ω R ·R+ω T ·T

[0051] Normalized sorting calculation:

[0052]

[0053] After normalization calculation, priority ranking is performed, and the calculation results are used as constraints for optimization calculation in the MPC optimization inspection path of model predictive control.

[0054] The beneficial effects of this invention are:

[0055] This invention is based on the modular setup of a water conservancy engineering embodied robot. It combines a hierarchical enhanced decision-making framework with a model predictive control (MPC) algorithm to optimize inspection paths, avoiding rapid flow areas or structurally vulnerable areas. Tasks are dynamically switched according to the priority of potential hazards, and multi-state data of water conservancy projects are collected. By integrating structured, semi-structured, and unstructured data, and utilizing natural language processing, data mining, and knowledge representation learning techniques, the invention systematically sorts out the potential risk factors and their relationships in water conservancy projects. This enables hazard source analysis, risk prediction, and cross-domain knowledge fusion, providing intelligent decision support for the full life-cycle risk management of water conservancy projects. Attached Figure Description

[0056] Figure 1 This is an architecture diagram of the humanoid robot inspection and hidden danger investigation system for water conservancy projects based on multi-state fusion, as described in this invention.

[0057] Figure 2 This is a knowledge graph architecture diagram of the present invention;

[0058] Figure 3 This is a flowchart of the knowledge graph construction process for this invention;

[0059] Figure 4 This is a flowchart of the hierarchical reinforcement decision-making algorithm of the present invention. Detailed Implementation

[0060] The embodiments of the present invention will now be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0061] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", "first", "second", etc., indicate the orientation, position or order relationship based on the orientation, position or order relationship shown in the drawings, and are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention.

[0062] See appendix Figure 1 As shown, the invention discloses a water conservancy engineering humanoid robot inspection and hidden danger investigation system based on multi-state fusion, which includes a water conservancy engineering embodied robot, a water conservancy information architecture unit based on multi-state fusion, and an inspection and hidden danger investigation unit.

[0063] The amphibious robot for water conservancy projects includes an amphibious control module, a multimodal perception kit, and a communication-decision module. The amphibious control module is used to check and control the movement of the amphibious robot for water conservancy projects. The multimodal perception kit is used for the amphibious robot for water conservancy projects to detect safety data of water conservancy projects. The communication-decision module is used for the amphibious robot for water conservancy projects to transmit data in real time with the shore-based fog gateway and perform hierarchical enhanced decision-making calculations.

[0064] The water conservancy information architecture unit based on multi-state fusion is used to collect sensor data and water conservancy hidden danger knowledge graph data of water conservancy projects, and to perform data calculation and early warning.

[0065] The inspection and hazard investigation unit includes several inspection paths, collects corresponding path information, and performs intelligent hazard diagnosis based on the inspection results and information collection.

[0066] This invention is based on the modular setup of a water conservancy engineering embodied robot. It combines a hierarchical enhanced decision-making framework with a model predictive control (MPC) algorithm to optimize inspection paths, avoiding rapid flow areas or structurally vulnerable areas. Tasks are dynamically switched according to the priority of potential hazards, and multi-state data of water conservancy projects are collected. By integrating structured, semi-structured, and unstructured data, and utilizing natural language processing, data mining, and knowledge representation learning techniques, the invention systematically sorts out the potential risk factors and their relationships in water conservancy projects. This enables hazard source analysis, risk prediction, and cross-domain knowledge fusion, providing intelligent decision support for the full life-cycle risk management of water conservancy projects.

[0067] Specifically, in this embodiment, the amphibious body control module includes a biomimetic adaptive body and a body perception sensor. The biomimetic adaptive body adopts a modular joint design, is equipped with a waterproof servo motor and magnetically adsorbed feet, and is adapted to walking on land (slope ≤ 60°), crawling along underwater walls (resistance to current velocity ≥ 2m / s), and peristalsis in silt environments. It can monitor attitude stability in real time through an inertial measurement unit (IMU) based on a dynamic center of gravity adjustment algorithm, and optimize underwater motion energy consumption by combining a fluid dynamics model.

[0068] The body sensing sensors are embedded with strain gauges and torque sensors in the robot joints to monitor the load status of the robotic arm in real time; the feet integrate a pressure sensor array to dynamically sense the ground hardness and slip risk.

[0069] In this embodiment, the multimodal sensing kit includes an underwater-mountable kit: a waterproof sonar array (for scanning cavities in dam foundations, dikes, etc.) and a miniature water quality sensor (for measuring water temperature, conductivity, etc.);

[0070] The detection kit includes a microscope camera and a conductivity probe at the end of a robotic arm for contact detection of concrete carbonation depth and steel corrosion.

[0071] Problem-solving kits include flexible grippers based on haptic feedback, which can grab underwater foreign objects (such as those entangled in fishing nets) or collect sediment samples.

[0072] In this embodiment, the communication-decision module includes a basic communication module: using an underwater acoustic communication module (frequency band 10-50kHz) to transmit data in real time with a shore-based fog gateway.

[0073] A layered, enhanced decision-making framework: Model-based predictive control (MPC) optimizes inspection paths, avoiding high-flow areas or structurally vulnerable zones. Tasks are dynamically switched based on hazard priority (e.g., leak point re-inspection takes precedence over routine inspection).

[0074] In this embodiment, the water conservancy information architecture unit based on multi-modal fusion includes a water conservancy multi-modal sensor array composed of a detection module, a monitoring module, and an environmental module. The detection module includes ground-penetrating radar (to detect dam cavities), a sonar array (to scan underwater conditions), and a multispectral underwater camera (to identify biological erosion and underwater damage). The monitoring module includes a high-precision inclinometer (to measure dam displacement), GNSS sensors, and sensors for stress, strain, seepage pressure, and seepage flow. The environmental module includes environmental monitoring equipment such as surveillance cameras, drones, water level gauges, rainfall radar, and air temperature and pressure sensors.

[0075] In this embodiment, the inspection path includes a land-based path: along walkable surfaces such as the dam crest, horse trail, and sluice gate crest to detect potential structural hazards;

[0076] Underwater mode path: Submerge to the bottom of the reservoir to detect seepage and siltation in the dam foundation;

[0077] Climbing mode path: Adhere to the back of the gate to check for hinge corrosion and seal failure; or adhere to the culvert to check for culvert corrosion, leakage, and the flatness and smoothness of the culvert surface.

[0078] In this embodiment, see Appendix Figure 2-3 As shown, the water conservancy information architecture unit based on multi-morphic fusion integrates structured, semi-structured, and unstructured data, utilizing natural language processing, data mining, and knowledge representation learning technologies to systematically analyze the potential risk factors and their relationships in water conservancy projects. It integrates water conservancy industry standards, hazard assessment experience, and historical incident cases to construct a scenario-monitoring data-knowledge base analysis model. The graph uses hazard entities (such as reservoirs, dams, and canals) as core nodes, linking hazard types (leakage, cracks, settlement, etc.), analyzing related attributes such as formation mechanisms (geological structure, material aging, extreme weather, etc.), impact range, monitoring indicators, and mitigation measures, forming a dynamic and reasonable networked knowledge system. This enables hazard source analysis, risk prediction, and cross-domain knowledge fusion, providing intelligent decision support for the full life-cycle risk management of water conservancy projects.

[0079] The water conservancy information architecture unit based on multi-state fusion deploys a lightweight hazard identification model on the robot end and a simulation calculation, hydrological prediction and analysis model, and comprehensive risk early warning model on the cloud.

[0080] It enables timely and tiered early warning. High-risk hazards (such as a sudden increase in seepage flow) trigger emergency broadcasts and link with the gate control system; long-term risks (such as slow settlement) generate reinforcement scheme suggestions and synchronize them to the water conservancy management platform.

[0081] In this embodiment, please refer to the appendix for details. Figure 4 As shown, the hierarchical reinforcement decision operation includes the following steps:

[0082] Step 1: Initial State Acquisition. Input the initial state x0 of the embodied robot and acquire the state value y from the sensor. k ;

[0083] Step 2: Set up state estimation, and use Kalman filtering to fuse multi-sensor data to estimate the robot's position and state at time k;

[0084]

[0085] Where: μ k-1 For control inputs (acceleration, steering angle, etc.), Q noise External and system interference noise.

[0086] Step 3: Construct a prediction model. Based on the robot's original patrol route and current position, obtain the robot's motion reference trajectory position x at time point k+i. ref According to the state x at time k k and μ k+ik Predict the robot's position at the next moment based on the control input.

[0087]

[0088] In the formula: A i Let A be the system matrix at time i. j Let B be the system matrix at time j, and let B be the adjustment matrix. The state equations are discretized using the Euler method.

[0089] Step 4: Rolling optimization solution: Based on the project monitoring situation, set constraints and solve the optimization function simultaneously;

[0090]

[0091] U * =argminJ(x k ,U)ST(constraints)

[0092] In the formula: r is the control penalty, to prevent drastic changes in the input control of state N-1; μ N-1 To control the input at time N-1, U * For optimal input control, U is the initial input control.

[0093] Step 5: Feedback and correction. Determine whether the objective has been achieved based on the actual situation. If yes, proceed to the next step; otherwise, recalculate.

[0094] In this embodiment, the intelligent diagnosis of hidden dangers adopts a dynamic evaluation method of fuzzy priority of hidden dangers, which includes evaluation of the risk level of hidden dangers and evaluation of the urgency of handling hidden dangers.

[0095] In this embodiment, the calculation formula for the risk level assessment of the hidden danger is R = α × hidden danger type + β × monitoring status;

[0096] In the formula: α and β are weighting coefficients, α + β = 1, α ≥ 0, β ≥ 0

[0097] Hazard types include: floating debris; gates unable to open or close; weeds encroaching on the spillway; reservoir bank collapse; sidewall deformation; abnormal deformation of structures; abnormal leakage; bubbling on the water surface; whirlpools; pits and depressions; ant nests and rodent burrows, etc. These hazard types are quantified into specific numerical values, for example:

[0098] Major hidden dangers (such as a dam being assessed as a Class III dam and not having been rectified). 5 Significant hidden danger (floodgates cannot be opened or closed normally and there is no backup). 4 General hidden danger (lack of monitoring facilities, but the consequences are controllable) 3 Minor hazard (localized corrosion of the metal structure, not affecting functionality) 2 No obvious hidden dangers 1

[0099] The monitoring results are quantified into specific numerical values ​​based on the specific monitoring conditions, for example:

[0100] Continuous automated monitoring, data integrity, and timely handling of anomalies 0–20 Regular manual and automatic monitoring, complete data 21–40 Insufficient monitoring frequency or missing data (<30%) 41–60 Monitoring facilities missing or data missing ≥30% 61–80 No monitoring or monitoring failure 81–100

[0101] The formula for calculating the urgency assessment of the hazard handling is as follows:

[0102] In the formula: t is the current time, t d The statutory time limit for addressing potential hazards.

[0103] Fuzzy dynamic weight adjustment:

[0104]

[0105] In the formula, μ i (t) is an expert score assigned based on the potential risks and the urgency of their handling, and is stored in a software library through machine learning training.

[0106] Priority calculation

[0107] Fuzzy priority calculation:

[0108] x i =ω R ·R+ω T·T

[0109] Normalized sorting calculation:

[0110]

[0111] After normalization calculation, priority ranking is performed, and the calculation results are used as constraints for optimization calculation in the MPC optimization inspection path of model predictive control.

[0112] The above description only illustrates preferred embodiments of the present invention and should not be construed as limiting the scope of the claims. The present invention is not limited to the above embodiments, and variations in its specific structure are permitted. In short, all variations made within the scope of the independent claims of the present invention are within the scope of protection of the present invention.

Claims

1. A humanoid robot inspection and hidden danger investigation system for water conservancy projects based on multi-state fusion, characterized in that: This includes embodied robots for water conservancy projects, water conservancy information architecture units based on multi-state fusion, and inspection and hidden danger investigation units; The amphibious robot for water conservancy projects includes an amphibious control module, a multimodal perception kit, and a communication-decision module. The amphibious control module is used to check and control the movement of the amphibious robot for water conservancy projects. The multimodal perception kit is used for the amphibious robot for water conservancy projects to detect safety data of water conservancy projects. The communication-decision module is used for the amphibious robot for water conservancy projects to transmit data in real time with the shore-based fog gateway and perform hierarchical enhanced decision-making calculations. The water conservancy information architecture unit based on multi-state fusion is used to collect sensor data and water conservancy hidden danger knowledge graph data of water conservancy projects, and to perform data calculation and early warning. The inspection and hazard investigation unit collects relevant path information through several inspection paths, and performs intelligent hazard diagnosis based on the inspection results and information collection. The amphibious avatar control module includes a biomimetic adaptive body and a body perception sensor. The biomimetic adaptive body adopts a modular joint design, including waterproof servo motors and magnetically attached feet, and its posture stability is monitored in real time by an inertial measurement unit (IMU). The body sensing sensor is embedded with strain gauges and torque sensors in the robot joints to monitor the load status of the robotic arm in real time; the foot integrates a pressure sensor array to dynamically sense the ground hardness and slippage risk. The multimodal sensing suite includes an underwater suite, a detection suite, and a problem-solving suite; The underwater kit includes a waterproof sonar array and a miniature water quality sensor; The detection kit includes a microscopic camera and a conductivity probe mounted on the end of a robotic arm, which detects the depth of concrete carbonation and steel corrosion through contact. The problem handling kit includes a flexible gripper based on tactile feedback for grasping underwater foreign objects or collecting sediment samples; The communication-decision module includes a basic communication module and a hierarchical enhanced decision-making framework; The basic communication module uses an underwater acoustic communication module to transmit data in real time with a shore-based fog gateway. The hierarchical enhanced decision-making framework optimizes inspection paths based on model predictive control (MPC), avoids high-flow areas or structurally vulnerable areas, and dynamically switches tasks according to the priority of potential hazards.

2. The water conservancy engineering humanoid robot inspection and hidden danger investigation system based on multi-state fusion according to claim 1, characterized in that: The water conservancy information architecture unit based on multi-mode fusion includes a water conservancy multi-modal sensor array composed of a detection module, a monitoring module, and an environmental module.

3. The humanoid robot inspection and hidden danger investigation system for water conservancy projects based on multi-state fusion as described in claim 1, characterized in that: The inspection path includes a land-based path, an underwater path, and a climbing path. The land-based route detects potential structural hazards by walking along the dam crest, horse trail, and sluice gate crest. The underwater mode path involves diving to the bottom of the reservoir to detect seepage and siltation in the dam foundation; The climbing mode path involves adhering to the back of the gate to check for hinge corrosion and sealing failure; or adhering to the culvert to check for culvert corrosion, leakage, and the flatness and smoothness of the culvert surface.

4. The humanoid robot inspection and hidden danger investigation system for water conservancy projects based on multi-state fusion as described in claim 1, characterized in that: The water conservancy information architecture unit based on polymorphic fusion includes a scenario-monitoring data-knowledge base analysis model, and the water conservancy hidden danger knowledge graph is obtained by analysis and reasoning from the scenario-monitoring data-knowledge base analysis model.

5. The humanoid robot inspection and hidden danger investigation system for water conservancy projects based on multi-state fusion as described in claim 1, characterized in that: The hierarchical reinforcement decision computation includes the following steps: Step 1: Initial State Acquisition, input the initial state of the embodied robot. x 0, and obtain status values ​​from the sensors. y k ; Step 2: Set up state estimation and use Kalman filtering to fuse multi-sensor data for estimation. k The robot's position and status at any given time; ； In the formula: To control the input, Q noise External and system interference noise; Step 3: Build a predictive model. Based on the robot's original patrol route and current position, obtain the robot's... k + i Time point motion reference trajectory position x ref ,according to k Moment State x k and Predict the robot's position at the next moment based on the control input. In the formula: A i Let A be the system matrix at time i. j Let B be the system matrix at time j, and let B be the adjustment matrix. The state equations are discretized using the Euler method. Step 4: Rolling optimization solution: Based on the project monitoring situation, set constraints and solve the optimization function simultaneously; ； In the formula: r To control the input control and prevent drastic changes in the N-1 state; The control input is defined at time N-1, where U* represents the optimal input control and U represents the initial input control. Step 5: Feedback and correction to determine if the objective has been achieved. If so, proceed to the next step; otherwise, recalculate.

6. The humanoid robot inspection and hidden danger investigation system for water conservancy projects based on multi-state fusion as described in claim 1, characterized in that: The intelligent diagnosis of hidden dangers adopts a dynamic evaluation method with fuzzy priority of hidden dangers, which includes evaluation of the risk level of hidden dangers and evaluation of the urgency of hidden danger treatment.

7. The humanoid robot inspection and hidden danger investigation system for water conservancy projects based on multi-state fusion as described in claim 6, characterized in that: The calculation formula for the risk level assessment of the hazard is R = α × hazard type + β × monitoring status. In the formula: α and β are weighting coefficients, α+β=1, α≥0, β≥0; The formula for calculating the urgency assessment of the hazard handling is as follows: ; In the formula: t is the current time, t d The statutory time limit for hazard remediation; Fuzzy dynamic weight adjustment: ; In the formula, Experts are assigned scores based on potential risks and the urgency of their handling, and the scores are trained and stored in a software library using machine learning. Priority calculation Fuzzy priority calculation: ; Normalized sorting calculation: ; After normalization calculation, priority ranking is performed, and the calculation results are used as constraints for optimization calculation in the MPC optimization inspection path of model predictive control.

Images