Tunnel oxygen supply control method based on multi-parameter coordination

Abstract

The invention discloses a tunnel oxygen supply control method based on multi-parameter coordination, and relates to the technical field of tunnel construction, and the method comprises the steps: constructing a multi-source environment monitoring data set, training a target ventilation quantity prediction model fused with a high-altitude correction coefficient based on the data set, and outputting an initial target ventilation quantity and an oxygen supply parameter; generating a pre-regulation instruction based on the initial target ventilation quantity and the oxygen supply parameters, simulating an instruction execution process through a digital twin virtual simulation model, and outputting oxygen supply evaluation parameters of oxygen content distribution, wind pressure loss and energy consumption of each block of the tunnel; constructing a tunnel oxygen supply evaluation model, and comparing oxygen supply evaluation parameters obtained by simulation with a safety threshold; and executing the formal adjustment instruction. Through multi-dimensional sensing treatment, high-altitude self-adaptive calculation, digital twinborn simulation verification, multi-scene parameter optimization, multi-device collaborative linkage and model dynamic evolution technologies, the breakthrough of tunnel oxygen supply from static experience control to dynamic precise intelligent regulation is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel construction, and in particular to a tunnel oxygen supply control method based on multi-parameter coordination. Background Art

[0002] In tunnel construction, especially in high-altitude and long-distance tunnel projects, the oxygen content, air quality and ventilation efficiency in the tunnel directly affect the safety and work efficiency of construction workers. Existing tunnel oxygen supply control methods generally have the following problems: single parameter monitoring. Traditional systems mostly rely on single-point oxygen sensor data and ignore related parameters such as CO2 concentration, wind pressure loss, and equipment energy consumption, resulting in a lag in oxygen supply adjustment; insufficient adaptability to high altitudes. The air pressure is low and the air is thin in high altitude areas. The existing ventilation model does not fully consider the altitude correction factor, which can easily lead to excessive or insufficient oxygen supply; there is a lack of virtual verification links, and adjustment instructions are directly executed without simulation verification. Local hypoxia, energy consumption surge and other problems may occur due to improper parameter matching; equipment coordination is poor. Equipment such as fans, oxygen generators and air silos are mostly controlled independently, lacking a dynamic coordination mechanism, making it difficult to adapt to dynamic environmental changes during tunnel excavation; model iteration is slow, and the control model has not established a real-time feedback update mechanism. After long-term operation, the prediction accuracy decreases, and manual intervention is required to adjust parameters.

[0003] As tunnel projects expand into areas with complex geology and high altitudes, the traditional experience-based static oxygen supply mode can no longer meet the construction requirements of "safety, efficiency, and energy saving". There is an urgent need to build an intelligent control system that coordinates multiple parameters and combines virtual and real elements. Summary of the Invention

[0004] In order to solve the above technical problems, a tunnel oxygen supply control method based on multi-parameter coordination is provided. This technical solution solves the above problems.

[0005] In order to achieve the above objects, the technical solution adopted by the present invention is:

[0006] A tunnel oxygen supply control method based on multi-parameter coordination, comprising:

[0007] Based on sensors within the tunnel, an array collects multi-dimensional environmental parameters and equipment operation data. This data is then subjected to noise reduction, spatiotemporal alignment, and standardization to construct a multi-source environmental monitoring dataset. A target ventilation volume prediction model, incorporating a high-altitude correction factor, is trained based on this dataset to output the initial target ventilation volume and oxygen supply parameters.

[0008] Generate pre-adjustment instructions based on the initial target ventilation volume and oxygen supply parameters. Use the digital twin virtual simulation model to simulate the instruction execution process and output oxygen supply assessment parameters such as oxygen content distribution, wind pressure loss, and energy consumption in each block of the tunnel.

[0009] A tunnel oxygen supply assessment model was constructed, and the simulated oxygen supply assessment parameters were compared with safety thresholds. If all assessment parameters met the standards, formal ventilation and oxygen supply dynamic adjustment instructions were generated. If any parameters did not meet the standards, a parameter optimization mechanism was triggered. The fan speed ratio, air silo supply volume, and oxygen generator power were adjusted using a fuzzy PID algorithm, and simulation verification was repeated.

[0010] Execute formal adjustment instructions, realize joint operation control of fans, air silos and oxygen generators through the equipment collaboration module, and collect data after execution in real time to feed back to the target ventilation volume prediction model to complete the dynamic update of the model.

[0011] Preferably, the array of sensors in the tunnel collects multi-dimensional environmental parameters and equipment operation data of the tunnel, performs noise reduction, spatiotemporal alignment and standardization on the data to construct a multi-source environmental monitoring data set, and trains a target ventilation volume prediction model integrating a high altitude correction coefficient based on the data set to output the initial target ventilation volume and oxygen supply parameters. Specifically, the method includes:

[0012] Based on the sensor array in the tunnel, oxygen sensors, gas analyzers, anemometers, and temperature sensors are deployed every 50 meters according to the excavation zone to collect real-time data on oxygen content, CO2 concentration, wind speed, and ambient temperature;

[0013] Through the sensor array, the speed of the forced-in fan, the pressure of the wind silo, the oxygen production of the oxygen concentrator and the operating power parameters of the construction equipment are synchronously collected to establish a real-time database of equipment operation;

[0014] The original monitoring data collected by the sensor array were subjected to 3σ criterion outlier removal and wavelet transform noise reduction processing, and then the spatiotemporal coordinates were aligned. The min-max normalization method was used to map the data to the [0,1] interval to construct a multi-source environmental monitoring dataset.

[0015] A target ventilation rate prediction model was trained based on a multi-source environmental monitoring dataset, incorporating a high-altitude correction factor. The model inputs included excavation length, tunnel cross-sectional area, and equipment oxygen consumption characteristics, and output the initial target ventilation rate and oxygen concentrator power parameters.

[0016] The five-fold cross-validation method was used to evaluate the model performance. Model training was completed when the prediction error was less than 8%. If it did not meet the standard, the model was retrained by increasing the sample size of similar working conditions.

[0017] Preferably, the target ventilation volume prediction model integrated with the high altitude correction coefficient is trained based on the multi-source environmental monitoring data set, and the model input includes the excavation length, tunnel cross-sectional area, and equipment oxygen consumption characteristics, and the output initial target ventilation volume and oxygen concentrator power parameters specifically include:

[0018] Among them, the core calculation formula of the target ventilation volume prediction model is:

[0019]

[0020] Where Q is the target ventilation volume, K is the high altitude correction factor, L is the tunnel excavation length, S is the tunnel cross-sectional area, and C is the target ventilation volume. E is the total oxygen consumption of the equipment, C T is the target oxygen content, C M Measure oxygen content for the sensor array;

[0021] The calculation formula for the high altitude correction factor K is:

[0022] K=1+0.001×(H-3000)

[0023] Where H is the tunnel entrance elevation. When H is greater than or equal to 4000, the upper limit of K is 1.5, and K is the high altitude correction factor.

[0024] Preferably, the pre-adjustment instructions are generated based on the initial target ventilation volume and oxygen supply parameters, and the instruction execution process is simulated by the digital twin virtual simulation model to output the oxygen supply evaluation parameters of the oxygen content distribution, wind pressure loss and energy consumption of each block of the tunnel, which specifically include:

[0025] Calculate the fan speed, duct pressure, and oxygen concentrator output power parameters based on the initial target ventilation volume, and generate pre-adjustment instructions containing multi-device coordination parameters;

[0026] Build a digital twin virtual model with a 1:1 scale of the actual tunnel, embedding tunnel structural parameters, ventilation duct roughness, and equipment characteristic curves;

[0027] Input pre-adjustment instructions into the digital twin virtual model to simulate the oxygen diffusion process, wind pressure attenuation along the tunneling process, and energy consumption distribution at different tunneling stages;

[0028] Output simulation results, including oxygen content distribution thermodynamic map of each block, wind pressure loss value (error ≤ 5%) and energy consumption per unit length data;

[0029] Perform statistical analysis on the simulation data, extract key oxygen supply evaluation parameters and store them in the simulation result database.

[0030] Preferably, the construction of a digital twin virtual model with a 1:1 ratio to the actual tunnel, and the implantation of tunnel structural parameters, ventilation duct roughness and equipment characteristic curves specifically include:

[0031] The digital twin virtual model consists of three core sub-models:

[0032] Oxygen diffusion sub-model: Based on computational fluid dynamics, the turbulent diffusion process of oxygen in the tunnel is simulated. The diffusion rate formula is:

[0033] v=v0×(1+0.02×ΔT)

[0034] Where v is the actual diffusion rate, v0 is the diffusion rate at standard temperature, and ΔT is the difference between ambient temperature and standard temperature;

[0035] Wind pressure loss sub-model: The wind pressure loss calculation formula is:

[0036]

[0037] Where ΔP is the wind pressure loss, λ is the local resistance coefficient, L is the pipe length, D is the pipe diameter, ρ is the air density, and v is the wind speed;

[0038] Energy consumption calculation sub-model: The total energy consumption formula of the system is:

[0039] E=∑(P i ×t i )

[0040] Where, E is the total energy consumption, P i is the power of the i-th device, t i The running time of the device.

[0041] Preferably, the tunnel oxygen supply evaluation model is constructed, and the oxygen supply evaluation parameters obtained by simulation are compared with the safety threshold. If all the evaluation parameters meet the standards, a formal ventilation and oxygen supply dynamic adjustment instruction is generated. If there are parameters that do not meet the standards, a parameter optimization mechanism is triggered, and the fan speed ratio, air silo supply and oxygen generator power are adjusted through a fuzzy PID algorithm. The simulation verification is then re-performed, specifically including:

[0042] Set safety threshold standards, which are oxygen content ≥ 19.5%, CO2 concentration ≤ 0.5%, wind pressure deviation ≤ 50Pa, and system energy consumption ≤ 110% of the design value;

[0043] Compare the oxygen supply assessment parameters obtained by simulation with the safety thresholds item by item to generate a parameter compliance matrix;

[0044] If all parameters in the parameter compliance matrix meet the safety threshold requirements, the pre-adjustment instruction will be confirmed as the formal ventilation and oxygen supply dynamic adjustment instruction;

[0045] If there are any parameters that do not meet the standards, the parameter optimization mechanism is triggered and the adjustment amount of each device parameter is generated through the fuzzy PID algorithm;

[0046] Update the adjusted parameters to the pre-adjustment instructions, return to S2 and re-simulate and verify, and the number of single optimization iterations shall not exceed 5 times.

[0047] Preferably, if there are substandard parameters, triggering the parameter optimization mechanism and generating the parameter adjustment amount of each device through the fuzzy PID algorithm specifically includes:

[0048] The adjustment rules of the parameter optimization mechanism are:

[0049] When the oxygen content is lower than the threshold, the ventilation volume adjustment value calculation formula is used to obtain the ventilation volume adjustment value, where the ventilation volume adjustment value calculation formula is:

[0050] ΔQ=0.1×(19.5%-C 实测 )×Q

[0051] Where ΔQ is the ventilation adjustment value, 19.5% is the minimum safe oxygen content threshold, C 实测 is the measured oxygen content, Q is the current ventilation volume, and the oxygen concentrator power is proportionally increased by 10% to 20%;

[0052] When the wind pressure deviation exceeds the limit, the fuzzy PID control output formula is used for adjustment. The fuzzy PID control output formula is:

[0053] u(k)=K p ×e(k)+K i ×∑e(j)+K d ×[e(k)-e(k-1)]

[0054] Where u(k) is the fan speed correction value, K p is the proportionality coefficient, K i is the integral coefficient, K d is the differential coefficient, e(k) is the current wind pressure deviation, e(k-1) is the wind pressure deviation at the previous moment, and K is dynamically adjusted through the fuzzy rule base p , K i and K d , until the wind pressure stabilizes for less than 30s;

[0055] When energy consumption exceeds the limit, optimize the fan operation combination and adopt the main fan plus relay fan segmented operation mode. The energy consumption optimization target per unit length is:

[0056] E′=E×(1-0.15×η)

[0057] Where E′ is the optimized unit energy consumption, E is the original unit energy consumption, and η is the fan combination matching degree.

[0058] Preferably, the execution of the formal adjustment instruction realizes the joint operation control of the fan, the air silo and the oxygen concentrator through the equipment collaboration module, and at the same time collects the data after the execution in real time and feeds it back to the target ventilation volume prediction model. The dynamic update of the model is completed specifically including:

[0059] The dynamic adjustment instructions for ventilation and oxygen supply are sent to the controllers of each device via the LoRa wireless communication protocol. The instructions include the fan speed, air bin valve opening, and oxygen concentrator power parameters.

[0060] The equipment coordination module coordinates the operating status of the fan, air silo and oxygen concentrator in real time to ensure that the parameters of each device are adjusted synchronously;

[0061] After the adjustment command is executed, the sensor array continuously collects 48 hours of environmental parameters and equipment operation data, and stores them in the historical database in time series;

[0062] The collected feedback data is marked as samples and added to the training set of the target ventilation volume prediction model at a ratio of 20%;

[0063] The target ventilation rate prediction model is updated using an incremental learning approach, freezing the underlying feature extraction parameters and only optimizing the top-level output layer weights.

[0064] Preferably, the equipment coordination module coordinates the operating status of the blower, air silo and oxygen concentrator in real time to ensure synchronous adjustment of the parameters of each device, specifically including:

[0065] The control logic of the device collaboration module is:

[0066] The fan coordination, the speed ratio formula of the main fan and the relay fan is:

[0067]

[0068] Where n1 is the main fan speed, n2 is the relay fan speed, P1 is the main fan rated power, P2 is the relay fan rated power, L1 is the distance from the main fan to the relay fan, and L2 is the distance from the relay fan to the working surface.

[0069] Air silo scheduling, the air supply volume calculation formula is:

[0070] Q f =k×ΔC×Q

[0071] Where Q f is the air supply volume of the wind silo, k is the air supply coefficient, ΔC is the oxygen content fluctuation value, Q is the current ventilation volume, and when the sensor array detects that the oxygen content fluctuation is greater than 0.5% / min, the air supply will be automatically turned on;

[0072] Oxygen production linkage, the oxygen production adjustment formula of the oxygen concentrator is:

[0073]

[0074] In the formula, O is the target oxygen production, O0 is the current oxygen production, C 目标 is the target oxygen content, C 实测 The actual oxygen content.

[0075] Preferably, the S405 specifically includes:

[0076] A dual evaluation mechanism is set up during incremental learning:

[0077] Performance evaluation, perform error prediction calculation, where the prediction error calculation formula is:

[0078]

[0079] Where, e is the prediction error, Q p The ventilation rate predicted by the model, Q a is the actual required ventilation volume; when e is greater than or equal to 5%, the full retraining process is triggered;

[0080] Adaptability assessment: The prediction accuracy of the model at different tunneling distances is tested every 10 days. If the prediction error of a certain section is greater than 10%, the weight of the samples in the section in the training set is increased. The weight adjustment formula is:

[0081] w i ′=w i ×(1+0.2×e i )

[0082] Where w i ′ is the adjusted weight, w i is the original weight, e i is the prediction error for this section;

[0083] At the same time, feedback samples were stratified and sampled according to the excavation stage, seasonal temperature and equipment type to ensure the uniform distribution of training samples.

[0084] Compared with the prior art, the present invention has the following beneficial effects:

[0085] 1. By collecting multi-dimensional environmental parameters through a distributed sensor array and combining it with a high-altitude correction coefficient model, we can accurately predict the target ventilation volume, keeping the prediction error within 8%. This solves the problem of local hypoxia caused by traditional single-point monitoring. Digital twin virtual simulation technology is introduced to simulate and verify sub-models such as oxygen diffusion and wind pressure loss before command execution. This ensures that the oxygen content in each block is ≥19.5% and the wind pressure deviation is ≤50Pa, thus avoiding safety risks caused by parameter mismatch.

[0086] 2. Through the fuzzy PID algorithm and the equipment collaboration module, dynamic adjustment of the fan speed ratio, wind silo air supply volume and oxygen generator power is achieved. The response time of the main fan and relay fan speed linkage is less than 30s. When the oxygen content fluctuates beyond the limit, the wind silo automatically replenishes air to ensure the stability of oxygen supply. An energy consumption optimization target model is established, and the "main fan + relay fan" segmented operation mode is adopted. Combined with incremental learning to dynamically optimize equipment operating parameters, the system energy consumption is reduced by more than 15% compared with traditional methods, and the energy consumption per unit length is controlled within 110% of the design value.

[0087] 3. The target ventilation volume prediction model is dynamically updated through a dual evaluation mechanism. When the prediction error is greater than 5%, full retraining is triggered to ensure the model's continued adaptability in different tunneling stages and altitude environments, reducing manual maintenance costs. The distribution of training samples is optimized through stratified sampling and weight adjustment mechanisms, making the method adaptable to tunnel projects with different geological conditions, tunneling lengths, and equipment configurations. It shows significant advantages, especially in high-altitude tunnels above 3,000 meters above sea level. BRIEF DESCRIPTION OF THE DRAWINGS

[0088] Figure 1 It is a step flow chart of the present invention;

[0089] Figure 2 This is a flowchart of the steps of S1 in the present invention;

[0090] Figure 3 This is a step flow chart of S2 in the present invention. DETAILED DESCRIPTION

[0091] The following description is intended to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are merely examples, and those skilled in the art may conceive of other obvious variations.

[0092] Reference Figure 1 As shown, a tunnel oxygen supply control method based on multi-parameter coordination includes:

[0093] S1: Based on sensors within the tunnel, an array collects multi-dimensional environmental parameters and equipment operation data from the tunnel. This data is then subjected to noise reduction, spatiotemporal alignment, and standardization to construct a multi-source environmental monitoring dataset. A target ventilation volume prediction model, incorporating a high-altitude correction factor, is trained based on this dataset to output the initial target ventilation volume and oxygen supply parameters.

[0094] S2: Generate pre-adjustment instructions based on the initial target ventilation volume and oxygen supply parameters. Use the digital twin virtual simulation model to simulate the instruction execution process and output oxygen supply assessment parameters such as oxygen content distribution, wind pressure loss, and energy consumption in each block of the tunnel.

[0095] S3: Construct a tunnel oxygen supply evaluation model and compare the simulated oxygen supply evaluation parameters with the safety threshold. If all evaluation parameters meet the standards, generate a formal ventilation and oxygen supply dynamic adjustment instruction. If any parameters do not meet the standards, trigger the parameter optimization mechanism and adjust the fan speed ratio, air silo supply volume, and oxygen generator power through the fuzzy PID algorithm. Return to S2 and re-simulate and verify.

[0096] S4: Execute the formal adjustment instructions, realize the joint operation control of the fan, air silo and oxygen generator through the equipment collaboration module, and collect the data after execution in real time to feed it back to the target ventilation volume prediction model to complete the dynamic update of the model.

[0097] Reference Figure 2 As shown, S1 specifically includes:

[0098] S101: Based on the sensor array in the tunnel, oxygen sensors, gas analyzers, anemometers, and temperature sensors are deployed every 50 meters according to the excavation zone to collect real-time data on oxygen content, CO2 concentration, wind speed, and ambient temperature;

[0099] S102: synchronously collecting the speed of the forced-in blower, the wind silo pressure, the oxygen production of the oxygen concentrator, and the operating power parameters of the construction equipment through the sensor array to establish a real-time database of equipment operation;

[0100] S103: The original monitoring data collected by the sensor array is subjected to 3σ criterion outlier removal and wavelet transform noise reduction processing, and then the time and space coordinates are aligned. The min-max normalization method is used to map the data to the [0, 1] interval to construct a multi-source environmental monitoring dataset;

[0101] S104: A target ventilation rate prediction model is trained based on a multi-source environmental monitoring dataset, incorporating a high-altitude correction factor. The model inputs include excavation length, tunnel cross-sectional area, and equipment oxygen consumption characteristics, and outputs initial target ventilation rate and oxygen concentrator power parameters.

[0102] S105: Use the five-fold cross-validation method to evaluate model performance. When the prediction error is less than 8%, the model training is completed. If it does not meet the standard, increase the sample size of similar working conditions and retrain;

[0103] A distributed multi-dimensional sensor array and data preprocessing integration solution is proposed. The data quality is improved through the 3σ criterion, wavelet denoising and time-space alignment technology. Combined with the model optimization mechanism of 50-fold cross-validation, the problem of high noise in traditional single-parameter data acquisition and insufficient model generalization ability is solved.

[0104] S104 specifically includes:

[0105] Among them, the core calculation formula of the target ventilation volume prediction model is:

[0106]

[0107] Where Q is the target ventilation volume, K is the high altitude correction factor, L is the tunnel excavation length, S is the tunnel cross-sectional area, and C is the target ventilation volume. E is the total oxygen consumption of the equipment, C T is the target oxygen content, C M Measure oxygen content for the sensor array;

[0108] The calculation formula for the high altitude correction factor K is:

[0109] K=1+0.001×(H-3000)

[0110] Where H is the tunnel entrance elevation. When H is greater than or equal to 4000, the upper limit of K is 1.5, and K is the high altitude correction factor.

[0111] The first ventilation volume calculation model to integrate a high-altitude correction coefficient quantifies the impact of altitude on oxygen demand through a dynamic altitude correction formula, solving the problem of insufficient oxygen supply accuracy of traditional models in high-altitude environments and achieving accurate oxygen supply calculation in different altitude scenarios.

[0112] Reference Figure 3 As shown, S2 specifically includes:

[0113] S201: Calculating fan speed, air duct pressure, and oxygen concentrator output power parameters based on the initial target ventilation volume, and generating a pre-adjustment instruction including multi-device coordination parameters;

[0114] S202: Build a digital twin virtual model with a 1:1 scale of the actual tunnel, embedding tunnel structural parameters, ventilation duct roughness, and equipment characteristic curves;

[0115] S203: Inputting the pre-adjustment instructions into the digital twin virtual model to simulate the oxygen diffusion process, wind pressure attenuation law along the tunneling process, and energy consumption distribution at different tunneling stages;

[0116] S204: Output simulation results, including a thermodynamic diagram of oxygen content distribution in each block, wind pressure loss value (error ≤ 5%), and energy consumption per unit length;

[0117] S205: Performing statistical analysis on the simulation data, extracting key oxygen supply evaluation parameters and storing them in a simulation result database;

[0118] By introducing 1:1 digital twin virtual simulation technology, the effects of adjustment instructions are verified in advance by implanting a virtual model with physical characteristic parameters, and multi-dimensional evaluation parameters are output to solve the safety risks and parameter mismatch problems existing in the direct execution of traditional control instructions, thus realizing a safe closed loop of "simulation optimization-instruction execution".

[0119] S202 specifically includes:

[0120] The digital twin virtual model consists of three core sub-models:

[0121] Oxygen diffusion sub-model: Based on computational fluid dynamics, the turbulent diffusion process of oxygen in the tunnel is simulated. The diffusion rate formula is:

[0122] v=v0×(1+0.02×ΔT)

[0123] Where v is the actual diffusion rate, v0 is the diffusion rate at standard temperature, and ΔT is the difference between ambient temperature and standard temperature;

[0124] Wind pressure loss sub-model: The wind pressure loss calculation formula is:

[0125]

[0126] Where ΔP is the wind pressure loss, λ is the local resistance coefficient, L is the pipe length, D is the pipe diameter, ρ is the air density, and v is the wind speed;

[0127] Energy consumption calculation sub-model: The total energy consumption formula of the system is:

[0128] E=∑(P i ×t i )

[0129] Where, E is the total energy consumption, P i is the power of the i-th device, t i How long the device has been running;

[0130] A digital twin model system with multi-physical field coupling is constructed. Through refined modeling of oxygen diffusion, wind pressure loss and energy consumption calculation, accurate simulation of multiple physical parameters in the tunnel oxygen supply process is achieved, solving the problem that traditional simulation models ignore the influence of multi-factor coupling.

[0131] S3 specifically includes:

[0132] S301: Set safety threshold standards, which are oxygen content ≥ 19.5%, CO2 concentration ≤ 0.5%, wind pressure deviation ≤ 50Pa, and system energy consumption ≤ 110% of the design value;

[0133] S302: Compare the oxygen supply assessment parameters obtained by simulation with the safety thresholds item by item to generate a parameter compliance matrix;

[0134] S303: If all parameters in the parameter compliance matrix meet the safety threshold requirements, the pre-adjustment instruction is confirmed as a formal ventilation and oxygen supply dynamic adjustment instruction;

[0135] S304: If there are any parameters that do not meet the standards, the parameter optimization mechanism is triggered and the adjustment amount of each device parameter is generated through the fuzzy PID algorithm;

[0136] S305: Update the adjusted parameters to the pre-adjustment instructions, return to S2 and re-simulate and verify, and the number of single optimization iterations shall not exceed 5;

[0137] A multi-dimensional evaluation mechanism based on the parameter compliance matrix is ​​proposed. Combined with a finite iterative optimization strategy, it solves the problems of single evaluation criteria and unbounded optimization in traditional control methods through safety threshold comparison and dynamic parameter adjustment, and achieves a balance between safety and efficiency.

[0138] S304 specifically includes:

[0139] The adjustment rules of the parameter optimization mechanism are:

[0140] When the oxygen content is lower than the threshold, the ventilation volume adjustment value calculation formula is used to obtain the ventilation volume adjustment value, where the ventilation volume adjustment value calculation formula is:

[0141] ΔQ=0.1×(19.5%-C 实测 )×Q

[0142] Where ΔQ is the ventilation adjustment value, 19.5% is the minimum safe oxygen content threshold, C 实测 is the measured oxygen content, Q is the current ventilation volume, and the oxygen concentrator power is proportionally increased by 10% to 20%;

[0143] When the wind pressure deviation exceeds the limit, the fuzzy PID control output formula is used for adjustment. The fuzzy PID control output formula is:

[0144] u(k)=K p ×e(k)+K i ×Σe(j)+K d ×[e(k)-e(k-1)]

[0145] Where u(k) is the fan speed correction value, K p is the proportionality coefficient, K i is the integral coefficient, K d is the differential coefficient, e(k) is the current wind pressure deviation, e(k-1) is the wind pressure deviation at the previous moment, and K is dynamically adjusted through the fuzzy rule base p , K i and K d , until the wind pressure stabilizes for less than 30s;

[0146] When energy consumption exceeds the limit, optimize the fan operation combination and adopt the main fan plus relay fan segmented operation mode. The energy consumption optimization target per unit length is:

[0147] E′=E×(1-0.15×η)

[0148] Where, E′ is the optimized unit energy consumption, E is the original unit energy consumption, and η is the fan combination matching degree;

[0149] Design multi-scenario adaptive parameter optimization rules, establish quantitative adjustment formulas for oxygen content, wind pressure, and energy consumption anomalies, and combine fuzzy PID dynamic parameter adjustment to solve the problem of poor adaptability of traditional single adjustment methods and achieve multi-parameter collaborative optimization.

[0150] S4 specifically includes:

[0151] S401: Sending dynamic ventilation and oxygen supply adjustment instructions to each device controller via the LoRa wireless communication protocol. The instructions include fan speed, air bin valve opening, and oxygen concentrator power parameters.

[0152] S402: The equipment coordination module coordinates the operating status of the fan, air silo, and oxygen concentrator in real time to ensure that the parameters of each device are adjusted synchronously;

[0153] S403: After the adjustment instruction is executed, the sensor array continuously collects 48 hours of environmental parameters and equipment operation data, and stores them in a historical database in time series;

[0154] S404: Mark the collected feedback data as samples and add them to the training set of the target ventilation volume prediction model at a ratio of 20%;

[0155] S405: Using incremental learning to update the target ventilation rate prediction model, freezing the bottom-level feature extraction parameters and only optimizing the top-level output layer weights;

[0156] Build a device collaboration and model evolution mechanism based on LoRa communication, realize dynamic model update through real-time data feedback and incremental learning, solve the problem of traditional models with declining accuracy and requiring manual intervention during long-term operation, and realize the system's self-evolution capability.

[0157] S402 specifically includes:

[0158] The control logic of the device collaboration module is:

[0159] The fan coordination, the speed ratio formula of the main fan and the relay fan is:

[0160]

[0161] Where n1 is the main fan speed, n2 is the relay fan speed, P1 is the main fan rated power, P2 is the relay fan rated power, L1 is the distance from the main fan to the relay fan, and L2 is the distance from the relay fan to the working surface.

[0162] Air silo scheduling, the air supply volume calculation formula is:

[0163] Q f =k×ΔC×Q

[0164] Where Q f is the air supply volume of the wind silo, k is the air supply coefficient, ΔC is the oxygen content fluctuation value, Q is the current ventilation volume, and when the sensor array detects that the oxygen content fluctuation is greater than 0.5% / min, the air supply will be automatically turned on;

[0165] Oxygen production linkage, the oxygen production adjustment formula of the oxygen concentrator is:

[0166]

[0167] In the formula, O is the target oxygen production, O0 is the current oxygen production, C 目标 is the target oxygen content, C 实测 is the measured oxygen content;

[0168] Establish multi-device collaborative control logic, and realize dynamic linkage of fans, air silos, and oxygen generators through quantified speed ratios, air supply volume, and oxygen production adjustment formulas, solve the problem of delayed response of independent control of traditional equipment, and improve the overall oxygen supply stability of the system.

[0169] S405 specifically includes:

[0170] A dual evaluation mechanism is set up during incremental learning:

[0171] Performance evaluation, perform error prediction calculation, where the prediction error calculation formula is:

[0172]

[0173] Where, e is the prediction error, Q p The ventilation rate predicted by the model, Q a is the actual required ventilation volume; when e is greater than or equal to 5%, the full retraining process is triggered;

[0174] Adaptability assessment: The prediction accuracy of the model at different tunneling distances is tested every 10 days. If the prediction error of a certain section is greater than 10%, the weight of the samples in the section in the training set is increased. The weight adjustment formula is:

[0175] w i ′=w i ×(1+0.2×e i )

[0176] Where w i ′ is the adjusted weight, w i is the original weight, e i is the prediction error for this section;

[0177] At the same time, feedback samples were stratified and sampled according to the excavation stage, seasonal temperature and equipment type to ensure the uniform distribution of training samples;

[0178] An incremental learning mechanism of dual evaluation and stratified sampling is proposed. Through dynamic weight adjustment and full-weight training trigger conditions, the problems of uneven sample distribution and low iteration efficiency in traditional model update are solved, and the continuous adaptation of the model in different excavation stages is achieved.

[0179] In summary, the advantages of the present invention are:

[0180] The pioneering "data acquisition - virtual simulation - evaluation and optimization - execution feedback" closed-loop control architecture for tunnel oxygen supply enables a transition from static empirical control to dynamic intelligent regulation through dynamic linkage of multiple links. This addresses the technical pain points of traditional control methods, which lack global optimization and real-time adaptation, and enables precise control of the oxygen supply system throughout its entire life cycle.

[0181] A distributed sensor array deployment scheme was proposed, with multiple sensors deployed at high density according to tunneling zones. Preprocessing techniques such as 3σ outlier removal, wavelet transform noise reduction, and spatial-temporal coordinate alignment were combined to construct a high-quality multi-source environmental monitoring dataset. A model optimization mechanism using 50% cross-validation was introduced to effectively improve data reliability and model generalization, addressing the issues of high noise and insufficient model accuracy associated with traditional single-parameter data collection.

[0182] The first target ventilation volume prediction formula to incorporate a high-altitude correction factor uses a dynamic altitude correction algorithm to quantify the impact of altitude on oxygen demand. An upper limit for the correction factor is set for altitudes ≥4000m, enabling accurate oxygen supply calculations at different altitudes. This overcomes the limitations of traditional models in high-altitude areas.

[0183] A digital twin model with a 1:1 scale of the actual tunnel was constructed, incorporating physical parameters such as tunnel structure and equipment characteristics, and developing three core sub-models for oxygen diffusion, wind pressure loss, and energy consumption calculation. Virtual simulation was used to pre-verify the effects of adjustment commands, output multi-dimensional evaluation parameters, and form a secure closed loop of "simulation optimization-command execution," avoiding the safety risks and parameter mismatch issues of traditional direct execution models.

[0184] A parameter optimization mechanism based on the fuzzy PID algorithm is designed, and a quantitative adjustment formula is established for scenarios such as insufficient oxygen content, wind pressure deviation, and excessive energy consumption. The ventilation volume is dynamically adjusted when the oxygen content is lower than the threshold; the PID parameters are optimized in real time through the fuzzy rule library when the wind pressure is abnormal; and when energy consumption exceeds the standard, a "main fan + relay fan" segmented operation mode is adopted to achieve multi-parameter coordinated optimization and efficient energy saving.

[0185] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions merely illustrate the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A tunnel oxygen supply control method based on multi-parameter coordination, characterized in that: include: S1: Based on sensors within the tunnel, an array collects multi-dimensional environmental parameters and equipment operation data from the tunnel. This data is then subjected to noise reduction, spatiotemporal alignment, and standardization to construct a multi-source environmental monitoring dataset. A target ventilation volume prediction model, incorporating a high-altitude correction factor, is trained based on this dataset to output the initial target ventilation volume and oxygen supply parameters. S2: Generate pre-adjustment instructions based on the initial target ventilation volume and oxygen supply parameters. Use the digital twin virtual simulation model to simulate the instruction execution process and output oxygen supply assessment parameters such as oxygen content distribution, wind pressure loss, and energy consumption in each block of the tunnel. S3: Construct a tunnel oxygen supply evaluation model and compare the simulated oxygen supply evaluation parameters with the safety threshold. If all evaluation parameters meet the standards, generate a formal ventilation and oxygen supply dynamic adjustment instruction. If any parameters do not meet the standards, trigger the parameter optimization mechanism and adjust the fan speed ratio, air silo supply volume, and oxygen generator power through the fuzzy PID algorithm. Return to S2 and re-simulate and verify. S4: Execute the formal adjustment instructions, realize the joint operation control of the fan, air silo and oxygen generator through the equipment collaboration module, and collect the data after execution in real time to feed it back to the target ventilation volume prediction model to complete the dynamic update of the model.

2. The tunnel oxygen supply control method based on multi-parameter coordination according to claim 1 is characterized in that: Said S1 specifically includes: S101: Based on the sensor array in the tunnel, oxygen sensors, gas analyzers, anemometers, and temperature sensors are deployed every 50 meters according to the excavation zone to collect real-time data on oxygen content, CO2 concentration, wind speed, and ambient temperature; S102: synchronously collecting the speed of the forced-in blower, the wind silo pressure, the oxygen production of the oxygen concentrator, and the operating power parameters of the construction equipment through the sensor array to establish a real-time database of equipment operation; S103: The original monitoring data collected by the sensor array is subjected to 3σ criterion outlier removal and wavelet transform noise reduction processing, and then the time and space coordinates are aligned. The min-max normalization method is used to map the data to the [0, 1] interval to construct a multi-source environmental monitoring dataset; S104: A target ventilation rate prediction model is trained based on a multi-source environmental monitoring dataset, incorporating a high-altitude correction factor. The model inputs include excavation length, tunnel cross-sectional area, and equipment oxygen consumption characteristics, and outputs initial target ventilation rate and oxygen concentrator power parameters. S105: Use the five-fold cross-validation method to evaluate the model performance. When the prediction error is less than 8%, the model training is completed. If it does not meet the standard, increase the sample size of similar working conditions and retrain.

3. The tunnel oxygen supply control method based on multi-parameter coordination according to claim 2 is characterized in that: The S104 specifically include: Among them, the core calculation formula of the target ventilation volume prediction model is: Where Q is the target ventilation volume, K is the high altitude correction factor, L is the tunnel excavation length, S is the tunnel cross-sectional area, and C is the target ventilation volume. E is the total oxygen consumption of the equipment, C T is the target oxygen content, C M Measure oxygen content for the sensor array; The calculation formula for the high altitude correction factor K is: K=1+0.001×(H-3000) Where H is the tunnel entrance elevation. When H is greater than or equal to 4000, the upper limit of K is 1.5, and K is the high altitude correction factor.

4. The tunnel oxygen supply control method based on multi-parameter coordination according to claim 3 is characterized in that: The S2 specifically includes: S201: Calculating fan speed, air duct pressure, and oxygen concentrator output power parameters based on the initial target ventilation volume, and generating a pre-adjustment instruction including multi-device coordination parameters; S202: Build a digital twin virtual model with a 1:1 scale of the actual tunnel, embedding tunnel structural parameters, ventilation duct roughness, and equipment characteristic curves; S203: Inputting the pre-adjustment instructions into the digital twin virtual model to simulate the oxygen diffusion process, wind pressure attenuation law along the tunneling process, and energy consumption distribution at different tunneling stages; S204: Output simulation results, including oxygen content distribution thermodynamic map of each block, wind pressure loss value and energy consumption per unit length data; S205: Perform statistical analysis on the simulation data, extract key oxygen supply evaluation parameters and store them in a simulation result database.

5. The tunnel oxygen supply control method based on multi-parameter coordination according to claim 4 is characterized in that: The S202 specifically includes: The digital twin virtual model consists of three core sub-models: Oxygen diffusion sub-model: Based on computational fluid dynamics, the turbulent diffusion process of oxygen in the tunnel is simulated. The diffusion rate formula is: v=v0×(1+0.02×ΔT) Where v is the actual diffusion rate, v0 is the diffusion rate at standard temperature, and ΔT is the difference between ambient temperature and standard temperature; Wind pressure loss sub-model: The wind pressure loss calculation formula is: Where ΔP is the wind pressure loss, λ is the local resistance coefficient, L is the pipe length, D is the pipe diameter, ρ is the air density, and v is the wind speed; Energy consumption calculation sub-model: The total energy consumption formula of the system is: E=∑(P i ×t i ) Where, E is the total energy consumption, P i is the power of the i-th device, t i The running time of the device.

6. The tunnel oxygen supply control method based on multi-parameter coordination according to claim 5 is characterized in that: The S3 specifically includes: S301: Set safety threshold standards, which are oxygen content ≥ 19.5%, CO2 concentration ≤ 0.5%, wind pressure deviation ≤ 50Pa, and system energy consumption ≤ 110% of the design value; S302: Compare the oxygen supply assessment parameters obtained by simulation with the safety thresholds item by item to generate a parameter compliance matrix; S303: If all parameters in the parameter compliance matrix meet the safety threshold requirements, the pre-adjustment instruction is confirmed as a formal ventilation and oxygen supply dynamic adjustment instruction; S304: If there are any parameters that do not meet the standards, the parameter optimization mechanism is triggered and the adjustment amount of each device parameter is generated through the fuzzy PID algorithm; S305: Update the adjusted parameters to the pre-adjustment instructions, return to S2 and re-simulate and verify, and the number of single optimization iterations shall not exceed 5 times.

7. The tunnel oxygen supply control method based on multi-parameter coordination according to claim 6, characterized in that: The S304 specifically includes: The adjustment rules of the parameter optimization mechanism are: When the oxygen content is lower than the threshold, the ventilation volume adjustment value calculation formula is used to obtain the ventilation volume adjustment value, where the ventilation volume adjustment value calculation formula is: ΔQ=0.1×(19.5%-C 实测 )×Q Where ΔQ is the ventilation adjustment value, 19.5% is the minimum safe oxygen content threshold, C 实测 is the measured oxygen content, Q is the current ventilation volume, and the oxygen concentrator power is proportionally increased by 10% to 20%; When the wind pressure deviation exceeds the limit, the fuzzy PID control output formula is used for adjustment. The fuzzy PID control output formula is: u(k)=K p ×e(k)+K i ×∑e(j)+K d ×[e(k)-e(k-1)] Where u(k) is the fan speed correction value, K p is the proportionality coefficient, K i is the integral coefficient, K d is the differential coefficient, e(k) is the current wind pressure deviation, e(k-1) is the wind pressure deviation at the previous moment, and K is dynamically adjusted through the fuzzy rule base p , K i and K d , until the wind pressure stabilizes for less than 30s; When energy consumption exceeds the limit, optimize the fan operation combination and adopt the main fan plus relay fan segmented operation mode. The energy consumption optimization target per unit length is: E′=E×(1-0.15×η) Where E′ is the optimized unit energy consumption, E is the original unit energy consumption, and η is the fan combination matching degree.

8. The tunnel oxygen supply control method based on multi-parameter coordination according to claim 1 is characterized in that: The S4 specifically includes: S401: Sending dynamic ventilation and oxygen supply adjustment instructions to each device controller via the LoRa wireless communication protocol. The instructions include fan speed, air bin valve opening, and oxygen concentrator power parameters. S402: The equipment coordination module coordinates the operating status of the fan, air silo, and oxygen concentrator in real time to ensure that the parameters of each device are adjusted synchronously; S403: After the adjustment instruction is executed, the sensor array continuously collects 48 hours of environmental parameters and equipment operation data, and stores them in a historical database in time series; S404: Mark the collected feedback data as samples and add them to the training set of the target ventilation volume prediction model at a ratio of 20%; S405: Update the target ventilation volume prediction model using incremental learning, freeze the bottom feature extraction parameters, and only optimize the top output layer weights.

9. The tunnel oxygen supply control method based on multi-parameter coordination according to claim 8, characterized in that: The S402 specifically includes: The control logic of the device collaboration module is: The fan coordination, the speed ratio formula of the main fan and the relay fan is: Where n1 is the main fan speed, n2 is the relay fan speed, P1 is the main fan rated power, P2 is the relay fan rated power, L1 is the distance from the main fan to the relay fan, and L2 is the distance from the relay fan to the working surface. Air silo scheduling, the air supply volume calculation formula is: Q f =k×ΔC×Q Where Q f is the air supply volume of the wind silo, k is the air supply coefficient, ΔC is the oxygen content fluctuation value, Q is the current ventilation volume, and when the sensor array detects that the oxygen content fluctuation is greater than 0.5% / min, the air supply will be automatically turned on; Oxygen production linkage, the oxygen production adjustment formula of the oxygen concentrator is: In the formula, O is the target oxygen production, O0 is the current oxygen production, C 目标 is the target oxygen content, C 实测 The actual oxygen content.

10. The tunnel oxygen supply control method based on multi-parameter coordination according to claim 9, characterized in that: The S405 specifically includes: A dual evaluation mechanism is set up during incremental learning: Performance evaluation, perform error prediction calculation, where the prediction error calculation formula is: Where, e is the prediction error, Q p The ventilation rate predicted by the model, Q a is the actual required ventilation volume; when e is greater than or equal to 5%, the full retraining process is triggered; Adaptability assessment: The prediction accuracy of the model at different tunneling distances is tested every 10 days. If the prediction error of a certain section is greater than 10%, the weight of the samples in the section in the training set is increased. The weight adjustment formula is: In i ′=w i ×(1+0.2×e i ) Where w i ′ is the adjusted weight, w i is the original weight, e i is the prediction error for this section; At the same time, feedback samples were stratified and sampled according to the excavation stage, seasonal temperature and equipment type to ensure the uniform distribution of training samples.

Images