Multi-target collaborative optimization control method for ultra-long tunnel electromechanical equipment

Abstract

The invention discloses a multi-objective collaborative optimization control method for ultra-long tunnel electromechanical equipment, and belongs to the technical field of intelligent traffic infrastructures. The method comprises the following steps: (a) preparing a distributed sensor array by adopting a multi-layer graphene-boron nitride heterostructure composite sensing material; (b) performing spatial registration based on tunnel three-dimensional deformation data obtained by laser radar scanning and fiber grating sensing data, and constructing a dynamic partition model; (c) establishing a multi-objective optimization function including three dimensions of energy consumption efficiency, equipment life and safety coefficient, and introducing an equipment maintenance cost weight factor; and (d), before collaborative optimization calculation is carried out through an improved NSGA-III algorithm, an equipment working mode pre-screening step is executed. The graphene-boron nitride heterostructure composite material is adopted, the salt spray corrosion resistance of the sensor is improved by 3 times through the plasma activation and atomic layer deposition technology, and the service life of the sensor in the 95% RH humidity environment is prolonged to 8 years or above.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent transportation infrastructure, more particularly, to a multi-target cooperative optimization control method for electromechanical equipment in an ultra-long tunnel. BACKGROUND

[0002] With the continuous expansion of ultra-long tunnel projects, the complexity and control difficulty of electromechanical equipment systems have significantly increased. The traditional control method mainly has the following technical bottlenecks: 1. The sensitivity of the sensor network is seriously attenuated in harsh environments such as humidity and vibration, and the service life of conventional materials is shortened due to chemical corrosion; 2. The spatial registration accuracy of multi-source heterogeneous data (such as laser point cloud and fiber strain data) is insufficient, making it difficult to achieve dynamic partitioning and accurate control; 3. Equipment maintenance relies on regular manual inspection, which cannot predict sudden failures in real time, and lacks a knowledge sharing mechanism across tunnel systems; 4. The existing digital twin model parameter calibration and sensor self-maintenance process are separated, and data drift problems are likely to occur during long-term operation. Especially for tunnels over 10 km long, existing problems such as control response lag, low equipment coordination efficiency, and significant energy waste make it difficult to meet the needs of modern smart tunnel construction. SUMMARY

[0003] 1. Problems to be solved

[0004] In view of the defects and deficiencies of the prior art, the present application provides a multi-target cooperative optimization control method for electromechanical equipment in an ultra-long tunnel, which uses a graphene-boron nitride heterostructure composite material. Through plasma activation and atomic layer deposition process, the salt mist corrosion resistance of the sensor is improved by 3 times, and the service life is extended to more than 8 years in a 95% RH humidity environment, solving the performance degradation problem of traditional sensors in harsh environments.

[0005] 2. Technical solutions

[0006] To achieve the above-mentioned purpose, the technical solutions provided by the present application are as follows:

[0007] A multi-target cooperative optimization control method for electromechanical equipment in an ultra-long tunnel, comprising the following steps:

[0008] Step (a) uses a multi-layer graphene-boron nitride heterostructure composite sensing material to prepare a distributed sensor array. The material is made of graphene, carbon nanotubes, polyimide resin and boron nitride nanosheets in a mass ratio of 3:1:6:2, and is processed by two-step high-temperature pressing.

[0009] Step (b) performs spatial registration based on the tunnel three-dimensional deformation data obtained by laser radar scanning and the fiber Bragg sensor data, and constructs a dynamic partitioning model.

[0010] Step (c) establishing a multi-objective optimization function including three dimensions: energy efficiency, equipment life, and safety factor, and introducing a weight factor for equipment maintenance cost;

[0011] Step (d) before performing collaborative optimization calculations using the improved NSGA-III algorithm, a device operating mode pre-screening step is performed;

[0012] Step (e) synchronously performs a sensor array self-cleaning operation during the system full parameter dynamic calibration process.

[0013] Furthermore, the preparation of the multilayer graphene-boron nitride heterostructure composite material in step (a) includes: after the graphene and carbon nanotubes are subjected to plasma activation treatment, a boron nitride isolation layer with a thickness of 5-8 nm is generated by atomic layer deposition technology, and then a step-by-step temperature curing treatment is performed, specifically, the temperature is increased to 180°C at a rate of 10°C / min and maintained for 30 minutes, and then the temperature is increased to 260°C at a rate of 5°C / min and maintained for 45 minutes.

[0014] Furthermore, the construction of the dynamic partition model in step (b) specifically includes:

[0015] (b1) Perform wavelet packet transform denoising on the original sensor data, retaining the detail components in the 4-6 layer decomposition coefficients;

[0016] (b2) The improved DBSCAN algorithm is used to detect abnormal data points, with the neighborhood radius set to 0.5 meters and the minimum number of included points set to 15.

[0017] Furthermore, the improved NSGA-III algorithm in step (d) includes:

[0018] (d1) Introducing a case matching mechanism based on the equipment historical operation database during the population initialization phase, prioritizing the retention of historical excellent solutions with a similarity of more than 80% with the current operating conditions;

[0019] (d2) A dynamic mutation operator is used in the evolutionary operation, and the mutation probability decreases linearly from 0.15 to 0.05 with the number of iterations.

[0020] Furthermore, the self-cleaning operation in step (e) is specifically: removing deposits on the sensor surface by generating 20-40kHz ultrasonic waves through piezoelectric ceramics, and this operation is performed synchronously with the parameter calibration process of the digital twin system.

[0021] Furthermore, when the digital twin system constructs a virtual image of the device, it compares and analyzes the real-time sensor data with the virtual model output, and activates the parameter compensation mechanism when the deviation exceeds 5%.

[0022] Furthermore, the method also includes step (f): an equipment life prediction model is constructed based on accelerated aging test data under salt spray environment, high humidity environment, and vibration conditions, and transfer learning technology is used to migrate the parameters of the general equipment degradation model to the tunnel-specific model.

[0023] Furthermore, the equipment working mode pre-screening step in step (d) includes: when it is detected that the remaining life prediction value of a certain equipment is less than 72 hours, automatically generating an equipment switching instruction and triggering the standby equipment preheating program, and the preheating temperature is controlled in the range of 40±5℃.

[0024] Furthermore, the optimal solution set of the multi-objective optimization function in step (c) is shared among multiple tunnel systems through a federated learning framework. While each local system retains private data, the global model performance is improved through encrypted gradient exchange.

[0025] Furthermore, it also includes steps for optimizing the dynamic balance of energy: solar energy is collected in real time through the cadmium telluride photovoltaic power generation film laid on the top of the tunnel. When the remaining power of the energy storage system is less than 30%, the equipment's graded power reduction mode is activated, prioritizing the reduction of the operating power of non-critical equipment to 60%-75% of the rated value, and simultaneously adjusting the ventilation system to intermittent operation mode.

[0026] 3. Beneficial effects

[0027] Compared with the prior art, the present invention has the following beneficial effects:

[0028] This invention utilizes a graphene-boron nitride heterostructure composite material, using plasma activation and atomic layer deposition processes. This allows the sensor to maintain a conductivity of 2.3×10^4 S / m while improving its salt spray corrosion resistance by three times. This extends its service life to over eight years in a 95% relative humidity environment, addressing the performance degradation issue of traditional sensors in harsh environments. By improving the DBSCAN algorithm and wavelet packet noise reduction technology, the spatial registration error between lidar and fiber optic data is reduced from ±15 cm to ±3 cm, compared to traditional methods. The dynamic partitioning model reconstruction response time is shortened to less than 5 seconds, significantly improving the efficiency of handling abnormal conditions. The improved NSGA-III algorithm, combined with a case matching mechanism, accelerates optimization calculation convergence by 40%. The equipment life prediction model, based on domain adaptation technology using transfer learning, reduces prediction error in salt spray environments from 22% to 7%. A federated learning framework, using encrypted gradient exchange, improves energy consumption optimization by 18% after collaborative training of the global model across 10 tunnel systems. The equipment pre-screening mechanism combined with a 72-hour life warning reduces the unplanned downtime rate by 65%; the self-cleaning system uses 35kHz ultrasonic waves and digital twin calibration to synchronize operations, reducing the frequency of manual maintenance by 80%, saving annual maintenance costs by more than 2 million yuan / km. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a flow chart of the method of the present invention. DETAILED DESCRIPTION

[0030] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0031] Example 1

[0032] See also Figure 1 This embodiment provides a technical solution: a multi-objective collaborative optimization control method for electromechanical equipment in ultra-long tunnels, which aims to achieve collaborative optimization of the working mode of electromechanical equipment through the improved NSGA-III algorithm described in claim 4. The specific steps are as follows:

[0033] Excellent solutions under similar operating conditions are extracted from the equipment's historical operation database as part of the initial population. A case matching mechanism is used to select solutions with at least 80% similarity to the current operating conditions and prioritize these solutions to improve the quality of the initial population. The remaining population individuals are randomly generated to ensure diversity.

[0034] A dynamic mutation operator is introduced during the evolution process, which causes the mutation probability to decrease linearly with the number of iterations. Initially, a higher mutation probability (0.15) is set to explore the solution space; as the number of iterations increases, the mutation probability is gradually reduced to 0.05 to enhance local search capabilities.

[0035] A multi-objective optimization function was established, encompassing energy efficiency, equipment lifespan, and safety factor, and a weighting factor for equipment maintenance cost was introduced. A collaborative optimization calculation was performed using the improved NSGA-III algorithm, ultimately yielding a Pareto optimal solution set.

[0036] Before formal optimization calculations are performed, equipment is pre-screened based on the predicted remaining lifespan. If a device's remaining lifespan is less than 72 hours, a switchover command is automatically generated, triggering the preheating process for the backup device, with the preheating temperature controlled within the range of 40±5°C.

[0037] The improved NSGA-III algorithm, combining historical data with a dynamic mutation operator, significantly improves optimization efficiency and solution quality, enabling multi-objective optimization functions to find the global optimal solution set in a shorter time. A pre-screening step for equipment operating modes effectively avoids system interruptions caused by equipment failures, improving system reliability and safety. This method simultaneously addresses energy efficiency, equipment lifespan, and safety factors, providing technical support for the efficient management of electromechanical equipment in ultra-long tunnels.

[0038] Embodiment 2

[0039] This embodiment provides a technical solution: a multi-objective collaborative optimization control method for super-long tunnel mechanical and electrical equipment, including sensor array self-cleaning and digital twin system synchronous calibration: according to the provisions of claims 5 and 6, the process of sensor array self-cleaning operation and digital twin system synchronous calibration is described in detail:

[0040] High-frequency ultrasonic waves of 20-40 kHz are generated by piezoelectric ceramics to remove deposits on the surface of the sensor through vibration. Self-cleaning operation is performed regularly, and the frequency can be dynamically adjusted according to environmental conditions. For example, in high humidity or dusty environments, the cleaning frequency can be appropriately increased.

[0041] The digital twin system compares real-time sensor data with virtual model output and initiates parameter compensation mechanism when the deviation exceeds 5%. During the construction of the virtual mirror, historical operation data, real-time monitoring data and external environmental variables (such as temperature and humidity, vibration, etc.) of the equipment are included in the model to ensure that it can accurately reflect the actual state of the equipment.

[0042] During the system full-parameter dynamic calibration process, the sensor array self-cleaning operation is performed synchronously. For example, when the digital twin system detects that the sensor data in a certain area deviates greatly, the self-cleaning operation is first triggered, and then the related parameters are recalibrated.

[0043] The sensor array self-cleaning function effectively avoids the problem of signal attenuation caused by the accumulation of pollutants, ensuring the long-term stability and high sensitivity of the sensor. The synchronous calibration of the digital twin system further improves the monitoring accuracy and reduces the error caused by environmental changes or equipment aging. The combination of the two significantly improves the reliability and accuracy of the entire monitoring system and reduces maintenance costs.

[0044] Embodiment 3

[0045] This embodiment provides a technical solution: a multi-objective collaborative optimization control method for super-long tunnel mechanical and electrical equipment, including energy dynamic balance optimization: according to the provisions of claim 10, the specific implementation process of energy dynamic balance optimization is described in detail:

[0046] A cadmium telluride photovoltaic power generation film is laid on the top of the tunnel to collect solar energy in real time and store it in the energy storage system. When the remaining power of the energy storage system is less than 30%, the hierarchical energy consumption reduction mode is started.

[0047] The running power of non-critical equipment is reduced to 60%-75% of the rated value, such as the lighting system and part of the monitoring equipment. The ventilation system is adjusted to intermittent operation mode to reduce unnecessary energy consumption.

[0048] Dynamically adjust operating strategies based on equipment load, prioritizing the normal operation of critical equipment (such as fire protection systems and emergency lighting). For non-critical equipment, adopt a rotational operation mode to avoid prolonged high-load operation of a single device.

[0049] The application of cadmium telluride photovoltaic thin films achieves energy self-sufficiency, reducing dependence on the external power grid. A graded power reduction mode and intermittent operation strategy effectively extend battery life while ensuring the normal operation of critical equipment. This approach demonstrates significant advantages in addressing power shortages and is of great significance for energy conservation, emission reduction, and improving system stability.

[0050] Comparative Example 1

[0051] Traditional optimization methods that do not use the improved NSGA-III algorithm include the following:

[0052] Description: Traditional optimization methods typically rely on a single objective or a simple weighted multi-objective optimization strategy, lacking the ability to effectively handle multiple constraints under complex operating conditions. For example, focusing solely on energy efficiency while ignoring equipment lifespan and safety factors can lead to premature equipment failure or safety hazards.

[0053] Comparison Results: By comparison, the improved NSGA-III algorithm in Example 1 was able to simultaneously address multiple objectives and provide a more optimal solution. The introduction of a dynamic mutation operator and historical data significantly improved optimization efficiency and solution quality, demonstrating the superiority of this approach in complex scenarios.

[0054] Comparative Example 2

[0055] Sensor array applications without self-cleaning functions include the following:

[0056] Description: Without a self-cleaning function, sensors are susceptible to contaminants such as dust and oil, leading to signal attenuation and inaccurate measurements. Furthermore, traditional manual cleaning is not only time-consuming and labor-intensive, but can also cause data errors due to untimely cleaning.

[0057] Comparison Results: In contrast, the self-cleaning function in Example 2 automatically removes deposits from the sensor surface, maintaining long-term stable sensor operation. Simultaneous digital twin system calibration further improves monitoring accuracy and significantly reduces maintenance costs and the need for human intervention.

[0058] In specific applications, step (a) accurately weighs the raw materials of graphene, carbon nanotubes, polyimide resin, and boron nitride nanosheets in a mass ratio of 3:1:6:2. The graphene and carbon nanotubes are first placed in a vacuum plasma reaction chamber and activated for 60 minutes by introducing argon gas at a power setting of 300W. Subsequently, atomic layer deposition technology is used to alternately introduce BCl3 and NH3 precursors onto the surface of the activated material, and the cycle is repeated 120 times to form a 6nm thick boron nitride isolation layer. The composite material is placed in a high-pressure mold for step-by-step temperature curing: the temperature is raised to 180°C at 10°C / min, then maintained at pressure for 30 minutes, and then raised to 260°C at 5°C / min and maintained at pressure for 45 minutes, ultimately forming a sensing material sheet with a heterogeneous structure. The sheet is cut into 5×5cm² units and arranged at intervals of 20 meters to form a distributed sensor array.

[0059] In step (b), a RIEGL VZ-4000i lidar was used to scan the tunnel structure, simultaneously collecting strain data from fiber Bragg grating sensors. A six-layer wavelet packet transform was applied to the raw data for noise reduction, retaining detail components from layers 4 to 6. Spatial registration was performed using an improved DBSCAN algorithm: a neighborhood radius of ε = 0.5 m, a minimum number of included points (MinPts) = 15, and density clustering using the Mahalanobis distance metric. When three consecutive data points were detected to exceed a threshold, a partition model reconstruction was automatically triggered, generating dynamic control zones encompassing the ventilation, lighting, and drainage subsystems.

[0060] Step (c) establishes the objective function F = αE + βL + γS, where:

[0061] Energy efficiency E = Σ(P_i×t_i) / E_total, where P_i is the real-time power of device i

[0062] Equipment life L=1 / (Σ(1 / L_i)), L_i is calculated based on accelerated aging test data

[0063] Safety factor S=min(S_1,S_2,...,S_n), S_i is the safety score of each monitoring point

[0064] The weight factors α, β, and γ are dynamically adjusted according to the equipment maintenance cost. When the maintenance cost is higher than the preset threshold, the β value is automatically increased by 30%.

[0065] Step (d) Before the NSGA-III algorithm iteration, perform device pre-screening:

[0066] Call the equipment life prediction model and generate a switching instruction when the remaining life is less than 72 hours;

[0067] Start the preheating process of the standby equipment and control the temperature in the range of 35-45℃;

[0068] The improved NSGA-III implementation includes:

[0069] During initialization, excellent solutions with working condition similarity ≥ 80% are matched from the historical database as the initial population;

[0070] Using dynamic mutation operator, the mutation probability p_m=0.15-0.05×(t / T), t is the current iteration number, T is the total iteration number;

[0071] Step (e) is performed synchronously during the calibration process:

[0072] The piezoelectric ceramic drive module generates 35kHz ultrasonic waves for 30 seconds to remove deposits on the sensor surface;

[0073] The digital twin system compares sensor data with virtual model output in real time. When the deviation is greater than 5%, the following occurs:

[0074] Start the parameter compensation mechanism and update the model stiffness coefficient K_new=K_old×(1+Δ / 10);

[0075] Trigger the calibration signal to recalibrate the sensor zero point.

[0076] The above is a schematic description of the present invention and its embodiments, which is not restrictive. The drawings show only one embodiment of the present invention, and the actual method is not limited thereto. Therefore, if a person skilled in the art is inspired by the above and, without departing from the purpose of the present invention, designs structures and embodiments similar to the technical solution without inventive means, they shall fall within the scope of protection of the present invention.

Claims

1. A multi-objective collaborative optimization control method for electromechanical equipment in ultra-long tunnels, characterized by: The following steps are involved: Step (a) preparing a distributed sensor array using a multilayer graphene-boron nitride heterostructure composite sensing material, wherein the material is prepared by a two-step high-temperature pressurization molding process of graphene, carbon nanotubes, polyimide resin, and boron nitride nanosheets in a mass ratio of 3:1:6:2; Step (b) spatially registering the three-dimensional deformation data of the tunnel acquired by the laser radar scanning with the fiber Bragg grating sensing data to construct a dynamic partition model; Step (c) establishing a multi-objective optimization function including three dimensions: energy efficiency, equipment life, and safety factor, and introducing a weight factor for equipment maintenance cost; Step (d) before performing collaborative optimization calculations using the improved NSGA-III algorithm, a device operating mode pre-screening step is performed; Step (e) synchronously performs a sensor array self-cleaning operation during the system full parameter dynamic calibration process.

2. The multi-objective collaborative optimization control method for electromechanical equipment in an ultra-long tunnel according to claim 1 is characterized by: The preparation of the multilayer graphene-boron nitride heterostructure composite material in step (a) includes: after the graphene and carbon nanotubes are subjected to plasma activation treatment, a 5-8 nm thick boron nitride isolation layer is formed by atomic layer deposition technology, and then a step-by-step temperature curing treatment is performed, specifically, the temperature is increased to 180°C at a rate of 10°C / min and maintained for 30 minutes, and then increased to 260°C at a rate of 5°C / min and maintained for 45 minutes.

3. The multi-objective collaborative optimization control method for electromechanical equipment in an ultra-long tunnel according to claim 1 is characterized by: The construction of the dynamic partition model in step (b) specifically includes: (b1) Perform wavelet packet transform denoising on the original sensor data, retaining the detail components in the 4-6 layer decomposition coefficients; (b2) The improved DBSCAN algorithm is used to detect abnormal data points, with the neighborhood radius set to 0.5 meters and the minimum number of included points set to 15.

4. The multi-objective collaborative optimization control method for electromechanical equipment in an ultra-long tunnel according to claim 1 is characterized by: The improved NSGA-III algorithm in step (d) includes: (d1) Introducing a case matching mechanism based on the equipment historical operation database during the population initialization phase, prioritizing the retention of historical excellent solutions with a similarity of more than 80% with the current operating conditions; (d2) A dynamic mutation operator is used in the evolutionary operation, and the mutation probability decreases linearly from 0.15 to 0.05 with the number of iterations.

5. The multi-objective collaborative optimization control method for electromechanical equipment in an ultra-long tunnel according to claim 1 is characterized by: The self-cleaning operation described in step (e) is specifically: using piezoelectric ceramics to generate 20-40kHz ultrasonic waves to remove deposits on the sensor surface, and this operation is performed synchronously with the parameter calibration process of the digital twin system.

6. The multi-objective collaborative optimization control method for electromechanical equipment in an ultra-long tunnel according to claim 5, characterized in that: When the digital twin system builds a virtual image of the device, it compares and analyzes the real-time sensor data with the virtual model output, and activates the parameter compensation mechanism when the deviation exceeds 5%.

7. The multi-objective collaborative optimization control method for electromechanical equipment in an ultra-long tunnel according to claim 1 is characterized by: The method also includes step (f): an equipment life prediction model is constructed based on accelerated aging test data under salt spray environment, high humidity environment, and vibration conditions, and transfer learning technology is used to migrate the parameters of the general equipment degradation model to the tunnel-specific model.

8. The multi-objective collaborative optimization control method for electromechanical equipment in an ultra-long tunnel according to claim 1 is characterized by: The equipment working mode pre-screening step in step (d) includes: when it is detected that the remaining life prediction value of a certain equipment is less than 72 hours, automatically generating an equipment switching instruction and triggering the standby equipment preheating program, and the preheating temperature is controlled in the range of 40±5℃.

9. The multi-objective collaborative optimization control method for electromechanical equipment in an ultra-long tunnel according to claim 1, characterized in that: The optimal solution set of the multi-objective optimization function in step (c) is used to share knowledge among multiple tunnel systems through a federated learning framework. While each local system retains private data, it improves the global model performance through encrypted gradient exchange.

10. The multi-objective collaborative optimization control method for electromechanical equipment in an ultra-long tunnel according to claim 1, characterized in that: It also includes steps for optimizing dynamic energy balance: solar energy is collected in real time through the cadmium telluride photovoltaic thin film laid on the top of the tunnel. When the remaining power of the energy storage system is less than 30%, the equipment's graded power reduction mode is activated, prioritizing the reduction of the operating power of non-critical equipment to 60%-75% of the rated value, and simultaneously adjusting the ventilation system to intermittent operation mode.

Images