Tunnel waterproof bonding layer spraying setting self-adaptive control method based on Kolmogorov-Arnold network and PID

Abstract

The invention provides a tunnel waterproof bonding layer spraying setting self-adaptive control method based on a Kolmogorov-Arnold network and PID (Proportion Integration Differentiation). The method comprises the following steps: dividing spraying areas according to spraying width, and collecting multi-point water content, temperature and humidity, vehicle speed and historical film-forming deviation; calculating a wet point proportion and a wet-dry difference according to the water content distribution, performing temperature and humidity correction to obtain a film forming water influence value, and determining a spraying amount jump limit and spraying amount upper and lower limits; inputting the film forming demand characteristic quantity into a Kolmogorov-Arnold network to obtain the target flow of each spraying area, and correcting the difference of adjacent areas under the constraint; the PID forms actual flow according to the flow and the opening degree of the pressure feedback regulating valve; calculating a film-forming continuity index according to the appearance after spraying to obtain a current deviation, and updating a historical deviation for a next construction section; according to the method, the consistency of spraying quantity setting and execution can be improved, and the film forming continuity is improved.

Description

Technical Field

[0001] This invention relates to the field of tunnel waterproof bonding layer spraying control technology, and in particular to an adaptive control method for tunnel waterproof bonding layer spraying settings based on Kolmogorov-Arnold network and PID. Background Technology

[0002] The spraying of waterproof bonding layers in tunnels typically employs a multi-nozzle, multi-channel parallel approach to form a continuous film on the substrate, achieving adhesion and waterproofing between the subsequent waterproofing layer and the substrate. During construction, the moisture content of the substrate often exhibits uneven distribution, such as localized wet patches and wet-dry transitions, along the spraying width and travel direction. Furthermore, changes in tunnel temperature, relative humidity, and vehicle speed over time collectively affect the material spreading, evaporation, and film formation processes, thus posing requirements for the stability of spray flow rate settings and execution.

[0003] In existing technologies, spraying systems are generally divided into multiple spraying channels or spraying zones according to the spraying width, with each channel supplied with material through a regulating valve. The control side is often equipped with flow sensors and pressure sensors, and closed-loop methods such as PID are used to ensure that the actual flow rate tracks the given set value. The set value is usually generated based on the construction formula, empirical parameters, or lookup / mapping rules, and can be further corrected by combining base surface moisture content detection, tunnel temperature and humidity detection, and vehicle speed information. Some systems use cameras or reflective inspection after spraying to record the appearance and conduct quality inspections, which serve as a basis for manual verification or post-event evaluation.

[0004] In the above schemes, the setpoint calculations are mostly based on single-point or average moisture content, which makes it difficult to express the wet-dry transition and abrupt changes within the same spraying area. Furthermore, the settings between different spraying channels lack jump constraints for differences between adjacent areas, easily leading to discontinuous changes in lateral spray volume. Post-spray appearance inspections mostly remain at the level of result recording or manual adjustment, without forming a deviation update mechanism linked to the spraying settings. As a result, when operating conditions change or the substrate condition remains consistent, the spraying settings cannot be iteratively and adaptively corrected with the construction section, thus causing local film discontinuity.

[0005] Therefore, an adaptive control method for setting the spraying of tunnel waterproof bonding layer that can overcome the shortcomings of the existing technology is a problem that needs to be solved by those skilled in the art. Summary of the Invention

[0006] One objective of this invention is to propose an adaptive control method for the spraying setting of tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID. The core technical problem to be solved is: in the construction of tunnel waterproof bonding layer spraying, facing conditions such as uneven distribution of base surface water content and changes in temperature, humidity and vehicle speed, how to achieve adaptive generation of spraying flow rate setting according to spraying area / channel and close-loop update with the film formation result after spraying, so as to ensure the continuity of film formation and the stability of the spraying process.

[0007] The adaptive control method for setting the spraying of tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID according to an embodiment of the present invention includes the following steps:

[0008] S1. Divide the tunnel base surface into spraying zones according to the spraying width, and obtain the base surface moisture content, temperature and humidity, vehicle speed and historical film formation deviation values ​​at multiple points in the spraying zone to form the spraying zone measurement values.

[0009] S2. Based on the multi-point base surface moisture content measured in the spraying area, the moisture content distribution is obtained by converging the spraying area. The moisture content distribution characterizes the wet and dry areas and reflects the wet-dry transition. The wet point ratio and wet-dry difference are calculated from the moisture content distribution. Then, the film formation water influence value is calculated in combination with temperature and humidity. Based on the film formation water influence value, the spray volume jump limit value, the upper limit value and the lower limit value of the spray volume are determined. The film formation demand characteristic quantity of the spraying area is generated by integrating the moisture content distribution, wet point ratio, wet-dry difference, film formation water influence value, vehicle speed and historical film formation deviation value.

[0010] S3. Input the film formation demand characteristics of the spraying area, including the water content distribution and the influence of water on film formation, into the Kolmogorov-Arnold network. The network includes a univariate mapping subnetwork and a combination layer that correspond to each of the film formation demand characteristics of the spraying area. The univariate mapping subnetwork maps each item to obtain the film formation contribution value. The combination layer weights and fuses the film formation contribution value. Under the constraint of the spray volume jump limit value, the spray volume difference between adjacent spraying areas is corrected. Under the constraint of the upper limit value and the lower limit value of the spray volume, the correction result is limited to form the target spray flow rate value of the spraying area.

[0011] S4. Using the target spray flow rate of the spraying area as the set value, the proportional-integral-derivative controller adjusts the valve opening based on the flow feedback value and the pressure feedback value to obtain the actual spray flow rate;

[0012] S5. Based on the actual spraying flow rate, collect the post-spray appearance characteristics of the sprayed area, calculate the film formation continuity index value and compare it with the film formation continuity threshold to form the current film formation deviation value.

[0013] S6. Update the historical film formation deviation value according to the spraying area based on the current film formation deviation value. The historical film formation deviation value is used as the historical film formation deviation value in the measurement value of the spraying area of ​​the next construction section.

[0014] Optionally, step S1 specifically includes:

[0015] Based on the spraying width of the spraying device, the number of spraying execution channels, and the width of the tunnel base, a spraying width coordinate system is established and the spraying area boundary is set along the spraying width direction. A spraying area number corresponding to the spraying area boundary is generated so that each spraying area corresponds to a spraying execution channel.

[0016] In each spraying area, the positions of water content measuring points are set according to the preset number of water content measuring points and the preset spacing between measuring points. The water content of the base surface at multiple points within the preset detection distance before spraying is collected, and the water content of the base surface at multiple points is assigned to the corresponding spraying area according to the position of the water content measuring points to ensure that each spraying area obtains the same number of measuring point values.

[0017] The moisture content of multiple base surfaces belonging to the same spraying area is sorted according to the vehicle's direction of travel to form a moisture content sequence. Temperature, humidity and vehicle speed are collected simultaneously. Temperature and humidity include tunnel gas temperature and tunnel gas relative humidity. The temperature and humidity are assigned to each spraying area under the spraying width according to the collection time.

[0018] Read the historical film formation deviation value corresponding to each spraying zone number, combine the water content sequence, temperature and humidity, vehicle speed and historical film formation deviation value according to the spraying zone number to form the spraying zone measurement value, and arrange the spraying zone measurement value in the order of the spraying zone number to form a spraying zone measurement matrix for parallel processing.

[0019] Optionally, step S2 specifically includes:

[0020] The multi-point base surface moisture content in the spraying area measurement values ​​is aligned with the spraying area boundary according to the location of the moisture content measurement points. The multi-point base surface moisture content falling into the same spraying area is aggregated into a set of moisture content measurement points for that spraying area. The moisture content measurement points located at the spraying area boundary are allocated to a single spraying area according to the preset boundary assignment rules.

[0021] The set of water content measurement points is sorted according to the vehicle travel direction to obtain the water content sequence. The water content sequence is used as the water content distribution and the water content distribution includes the number of water content measurement points in the spraying area. The maximum water content, minimum water content, and mean water content are calculated from the water content sequence as statistical quantities.

[0022] Using the mean water content as a comparison benchmark, each point in the water content sequence is judged to determine whether the water content is greater than the mean water content. The number of measuring points that meet the judgment is counted and compared with the number of water content measuring points in the spraying area to obtain the wet point ratio value. The wet point ratio value is associated with the spraying area number.

[0023] The wet-dry difference is determined by the difference between the maximum and minimum water content, and the wet point ratio and wet-dry difference are fused according to the preset weight fusion rule to obtain the water content change characteristics, so that the water content change characteristics can simultaneously reflect the change in water film coverage area and the water content change amplitude.

[0024] The temperature and relative humidity of the tunnel gas in the temperature and humidity are input into the temperature and humidity correction mapping. The correction amount is applied to the water content change characteristics and the value of the influence of water on film formation is output. The value of the influence of water on film formation is limited to a preset range as the calculation input for the spray volume constraint.

[0025] The influence level is determined by comparing the film-forming water-affected value with the preset influence level threshold. Based on the influence level, the spray volume jump limit value is selected and used to limit the spray volume difference between adjacent spraying areas. Based on the influence level, the film-forming range of the material is corrected to obtain the upper limit value and lower limit value of the spray volume, so that the spray volume jump limit value, the upper limit value and the lower limit value of the spray volume correspond to the spraying area number.

[0026] The water content components of the water content distribution are sequentially used as the preceding dimensions. The wet point ratio, wet-dry difference, film formation affected by water, vehicle speed, and historical film formation deviation are added in a fixed order as scalar dimensions to generate the film formation demand feature quantity of the spraying area. The number of dimensions of the film formation demand feature quantity of the spraying area is equal to the number of water content measurement points in the spraying area plus five. Each dimension is made to correspond to the univariate mapping subnetwork of the Kolmogorov Arnold network. The feature matrix of film formation demand of the spraying area is formed by arranging the feature matrix of film formation demand of the spraying area in the order of the spraying area number. The spray volume jump limit value, spray volume upper limit value, and spray volume lower limit value are output and aligned with the rows of the feature matrix of film formation demand of the spraying area.

[0027] Optionally, when inputting the tunnel gas temperature and relative humidity from the temperature and humidity data into the temperature and humidity correction mapping, applying a correction to the water content abrupt change characteristics, and outputting the film formation water influence value, the temperature and humidity correction mapping adopts a water influence function, wherein the water influence function is specifically:

[0028] ;

[0029] in, Number the spraying area The effect of water on film formation This is the lower limit of the preset value range. To preset the upper limit of the value range, Number the spraying area Temperature and humidity correction factor, As the first weighting coefficient, This is the second weighting coefficient. Number the spraying area The wet point ratio value, Number the spraying area The wet-dry difference, Number the spraying area The average moisture content, To perform the minimum value operation, This is for calculating the maximum value.

[0030] Optionally, step S3 specifically includes:

[0031] The film formation demand feature of the spraying area is arranged in the order of the spraying area number to form the film formation demand feature matrix of the spraying area. The film formation demand feature matrix of the spraying area is input into the input layer of the Kolmogorov-Arnold network at once, so that the number of rows of the input tensor is the number of spraying areas and each row corresponds to a film formation demand feature of a spraying area. At the same time, the spray volume jump limit value, spray volume upper limit value and spray volume lower limit value are aligned according to the spraying area number and input into the combination layer as constraint parameters.

[0032] Based on the fact that the dimension of the input layer is the sum of the number of water-containing measurement points in the spraying area and the five scalar feature dimensions, a corresponding univariate mapping subnetwork is set for each dimension of the input layer in the Kolmogorov-Arnold network. In each univariate mapping subnetwork, input neurons, piecewise function neurons and output neurons are set in sequence to fix the mapping path from a single input feature to the film-forming contribution value.

[0033] Each water content component of the water content distribution is input into the corresponding univariate mapping subnetwork to obtain the water content film formation contribution value component. The wet point ratio value, wet-dry difference value, film formation affected by water value, vehicle speed and historical film formation deviation value are input into the corresponding univariate mapping subnetwork to obtain the corresponding film formation contribution value. The piecewise function neuron performs piecewise function mapping on the input features according to the preset piecewise nodes and outputs the nonlinear film formation contribution value.

[0034] All film-forming contribution values ​​of the same spraying area are aggregated into the weighted fusion neuron of the composite layer. The film-forming contribution values ​​are weighted and summed using preset weight parameters to obtain the initial spraying flow rate value, so that the initial spraying flow rate value corresponds to the spraying area number.

[0035] The initial spray flow rate value is input into the jump correction unit of the combined layer. The difference between the initial spray flow rate values ​​of adjacent spray areas is calculated according to the spray area number order. The magnitude of the difference between the initial spray flow rate values ​​is compared with the corresponding spray volume jump limit value and a jump judgment is made. If the spray volume jump limit value is met, the initial spray flow rate value is retained. If the spray volume jump limit value is not met, the difference between the initial spray flow rate values ​​is truncated according to the spray volume jump limit value and the direction of the difference is maintained to generate the jump correction spray flow rate value.

[0036] The jump-corrected spray flow rate value is input into the boundary limiting unit of the combined layer. The jump-corrected spray flow rate value is compared with the corresponding upper and lower limits of the spray volume and boundary limiting is performed. When the jump-corrected spray flow rate value exceeds the upper limit of the spray volume, the upper limit of the spray volume is output. When the jump-corrected spray flow rate value is lower than the lower limit of the spray volume, the lower limit of the spray volume is output. The target spray flow rate value of the spray area is obtained and output in the order of the spray area number.

[0037] Optionally, when the combination layer performs jump correction and boundary limitation on the initial spray flow rate value based on the spray volume jump limit value, the spray volume upper limit value, and the spray volume lower limit value to obtain the target spray flow rate value of the spray area, the target spray flow rate value of the spray area is determined by the target spray function, wherein the target spray function is specifically:

[0038] ;

[0039] in, Number the spraying area The target spray flow rate value for the spraying area. Number the spraying area The upper limit of the spray volume, Number the spraying area The lower limit of the spray volume, Number the spraying area The jump correction spray flow rate value, Number the spraying area The limit value for the jump in injection volume, Number the spraying area The initial spray flow rate value, Number the spraying area The initial spray flow rate value, Number the spraying area. To perform the minimum value operation, This is for calculating the maximum value.

[0040] Optionally, step S4 specifically includes:

[0041] The target spray flow rate value for each spraying zone is issued to the spraying execution channel corresponding to the spraying zone according to the spraying zone number, and used as the flow rate set value for the proportional-integral-derivative controller;

[0042] Flow feedback and pressure feedback values ​​are collected in the material supply pipeline of the spraying execution channel. The flow feedback value is compared with the flow set value to obtain the flow deviation, and the trend of the flow deviation is corrected based on the pressure feedback value.

[0043] The proportional-integral-derivative controller generates a valve opening adjustment amount based on the corrected flow deviation, and then drives the regulating valve to change the valve opening after limiting the valve opening adjustment amount.

[0044] The actual spray flow rate is generated based on the changed valve opening, and the actual spray flow rate is associated with the spray area number for use in calculating the film formation continuity index value.

[0045] Optionally, step S5 specifically includes:

[0046] Align the actual spray flow rate with the spray area number, determine the post-spray detection delay according to the preset flow delay mapping table, and convert the post-spray detection delay into the post-spray detection distance according to the vehicle speed. When the post-spray detection distance is reached, trigger the appearance acquisition device to collect the post-spray appearance features, which include the post-spray image and reflectivity.

[0047] The appearance features after spraying are cut out according to the boundary of the sprayed area to obtain the appearance segment of the sprayed area. The appearance segment of the sprayed area is divided into film-forming regions to obtain film-forming pixels and non-film-forming pixels, and the film coverage and exposed area ratio are calculated accordingly.

[0048] In the appearance segment of the spraying area, the interruption segment of the film-forming pixel is detected along the vehicle's direction of travel, and the number of interruption segments is counted as the number of broken bands. The film coverage, exposed area ratio and number of broken bands are fused according to preset weights to generate a film continuity index value.

[0049] The current film formation deviation value is obtained by comparing the film formation continuity index value with the film formation continuity threshold. When the film formation continuity index value is lower than the film formation continuity threshold, the difference between the film formation continuity threshold and the film formation continuity index value is used as the current film formation deviation value and output in association with the spraying area number. When the film formation continuity index value is not lower than the film formation continuity threshold, the current film formation deviation value is set to zero.

[0050] Optionally, step S6 specifically includes:

[0051] The current film formation deviation value is mapped one by one with the historical film formation deviation value according to the spray area number to obtain the current deviation vector of the spray area;

[0052] Using the current deviation vector of the spraying area as input, the historical film formation deviation value and the current film formation deviation value are weighted and fused using a preset update coefficient to obtain the updated historical film formation deviation value. When the current film formation deviation value is zero, the historical film formation deviation value is updated by attenuation according to a preset attenuation coefficient.

[0053] For the updated historical film formation deviation value, calculate the deviation difference between adjacent spray areas along the spray width direction, compare the deviation difference with the preset deviation jump limit threshold, and perform truncation correction on the deviation difference that exceeds the preset deviation jump limit threshold to obtain the smoothed historical film formation deviation value, and perform preset deviation value range limitation on the smoothed historical film formation deviation value.

[0054] The historical film formation deviation value is smoothed and used as the historical film formation deviation value in the measurement value of the spraying area of ​​the next construction section. The historical film formation deviation value is then used as the scalar dimension of the film formation demand characteristic quantity of the spraying area and input into the Kolmogorov-Arnold network.

[0055] The beneficial effects of this invention are:

[0056] 1. This proposal suggests an improved method for generating settings for spraying waterproof bonding layers in tunnels. By dividing the spray width into spray zones corresponding to each execution channel, and collecting moisture content data from multiple points on the substrate within each zone to form a moisture content sequence, the setting calculation no longer relies solely on single-point or average moisture content. The method further calculates the wet point ratio and wet-dry difference from the moisture content distribution, and combines this with temperature and humidity correction mapping to obtain the film-forming water-affected value. This is used to determine the spray volume jump limit value, as well as the upper and lower limits of the spray volume, ensuring a closed loop between "operating condition identification—setting output—engineering constraints" during the setting generation stage. Unlike traditional rule-based grading or simple regression, the Kolmogorov–Arnold network configures a univariate mapping subnetwork for each dimension of input features and uses piecewise function neurons to complete nonlinear mapping. Then, it obtains the initial spray flow rate by weighted fusion in the combination layer. Under the constraints of jump limit and upper and lower limit, it truncates and corrects the spray flow rate difference between adjacent spray areas and limits the boundary, so that the set value has a continuous constraint in the lateral space and is kept within the range where the material can form a film. This helps to reduce the set value abrupt changes and local spray flow rate mismatch caused by wet-dry transition or local wet spots.

[0057] 2. This proposal puts forward a novel closed-loop update mechanism for film formation deviation based on post-spray appearance results. It uses post-spray images and reflectivity to construct a calculable film formation continuity index, and uses this result as input for the historical film formation deviation generated for the next construction section. The scheme determines the post-spray detection delay through a flow-time delay mapping table and calculates the detection distance based on vehicle speed, achieving temporal / spatial alignment between spray input and post-spray appearance. Then, appearance segments are cropped according to the spray area boundary, film formation areas are segmented based on reflectivity thresholds, film coverage and exposed substrate area ratio are calculated, and the number of broken bands is obtained by detecting film formation interruptions along the travel direction. These are then fused to form a film formation continuity index value, which is then output as the current film formation deviation value. Subsequently, historical film formation deviation values ​​are updated using a weighted fusion and zero-deviation attenuation strategy, and adjacent spray area deviation jump truncation and range limitation are performed along the spray width direction to achieve controllable deviation memory and lateral smoothing. Compared to existing methods that rely solely on post-construction inspections or manual parameter adjustments, this mechanism allows the post-spray film formation performance to be fed back into the set model as a structured deviation. This supports iterative correction of the settings as operating conditions change or material / environmental conditions persist, thereby reducing the probability of similar film formation problems recurring in subsequent construction phases.

[0058] 3. This proposal puts forward an adaptive control method for the spraying setting of tunnel waterproofing bonding layer based on Kolmogorov-Arnold network and PID, which integrates "zoning perception - nonlinear setting generation - multi-channel closed-loop execution - post-spraying evaluation - deviation update" into a closed-loop control link oriented towards construction scenarios. At the execution layer, each spraying execution channel uses the target spraying flow rate as the setpoint. The PID generates deviation based on flow feedback and corrects the deviation trend based on pressure feedback. Combined with integral limiting and valve opening limiting to drive the regulating valve, the stability of setting tracking is maintained under disturbances such as fluctuations in supply pressure, ensuring that the zoning setting output by Kolmogorov-Arnold network can be reliably implemented as the actual spraying flow rate. Overall, this method addresses the core issue of "achieving adaptive generation of spray flow settings based on spray zone / channel under conditions of non-uniform water content distribution and changes in temperature, humidity, and vehicle speed, and enabling closed-loop updates based on post-spray film formation results." Through collaborative design of distribution feature modeling, setting-end engineering constraints, post-spray deviation reinjection, and channel-level PID execution, it improves the correlation and adjustability between spray settings and construction results. Attached Figure Description

[0059] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0060] Figure 1 This is a flowchart of the adaptive control method for setting the spraying of tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID proposed in this invention;

[0061] Figure 2 This is a flowchart of the spraying zone measurement matrix generation process for the adaptive control method for spraying tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID proposed in this invention.

[0062] Figure 3 This is a flowchart illustrating the generation of the film-forming demand feature matrix for the spraying zone in the adaptive control method for spraying tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID proposed in this invention.

[0063] Figure 4 This is a flowchart illustrating the calculation of the target spray flow rate in the spraying area for the adaptive control method for spraying tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID proposed in this invention.

[0064] Figure 5This is a flowchart of the closed-loop control of the spraying execution channel flow for the adaptive control method for spraying tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID proposed in this invention.

[0065] Figure 6 This is a flowchart of the post-spray appearance feature evaluation and current film formation deviation value generation of the adaptive control method for spraying tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID proposed in this invention.

[0066] Figure 7 This is a flowchart showing the historical film formation deviation update and smoothing process of the adaptive control method for spraying tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID proposed in this invention.

[0067] Figure 8 Two figures showing the comparative effects of the adaptive control method for spraying waterproof bonding layer in tunnels based on Kolmogorov-Arnold network and PID proposed in this invention.

[0068] Figure 9 This is a diagram of the Kolmogorov-Arnold network based on the Kolmogorov-Arnold network and PID proposed in this invention. Detailed Implementation

[0069] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0070] refer to Figures 1 to 9 An adaptive control method for spraying waterproof adhesive layer in tunnels based on Kolmogorov-Arnold network and PID control is characterized by the following steps:

[0071] S1. Divide the tunnel base surface into spraying zones according to the spraying width, and obtain the base surface moisture content, temperature and humidity, vehicle speed and historical film formation deviation values ​​at multiple points in the spraying zone to form the spraying zone measurement values.

[0072] S2. Based on the multi-point base surface moisture content measured in the spraying area, the moisture content distribution is obtained by converging the spraying area. The moisture content distribution characterizes the wet and dry areas and reflects the wet-dry transition. The wet point ratio and wet-dry difference are calculated from the moisture content distribution. Then, the film formation water influence value is calculated in combination with temperature and humidity. Based on the film formation water influence value, the spray volume jump limit value, the upper limit value and the lower limit value of the spray volume are determined. The film formation demand characteristic quantity of the spraying area is generated by integrating the moisture content distribution, wet point ratio, wet-dry difference, film formation water influence value, vehicle speed and historical film formation deviation value.

[0073] S3. Input the film formation demand characteristics of the spraying area, including the water content distribution and the influence of water on film formation, into the Kolmogorov-Arnold network. The network includes a univariate mapping subnetwork and a combination layer that correspond to each of the film formation demand characteristics of the spraying area. The univariate mapping subnetwork maps each item to obtain the film formation contribution value. The combination layer weights and fuses the film formation contribution value. Under the constraint of the spray volume jump limit value, the spray volume difference between adjacent spraying areas is corrected. Under the constraint of the upper limit value and the lower limit value of the spray volume, the correction result is limited to form the target spray flow rate value of the spraying area.

[0074] S4. Using the target spray flow rate of the spraying area as the set value, the proportional-integral-derivative controller adjusts the valve opening based on the flow feedback value and the pressure feedback value to obtain the actual spray flow rate;

[0075] S5. Based on the actual spraying flow rate, collect the post-spray appearance characteristics of the sprayed area, calculate the film formation continuity index value and compare it with the film formation continuity threshold to form the current film formation deviation value.

[0076] S6. Update the historical film formation deviation value according to the spraying area based on the current film formation deviation value. The historical film formation deviation value is used as the historical film formation deviation value in the measurement value of the spraying area of ​​the next construction section.

[0077] In this embodiment, step S1 specifically includes:

[0078] The spraying device is installed on the construction vehicle. The spraying device includes multiple spraying execution channels. Each spraying execution channel is equipped with a regulating valve and corresponds to a spraying area. A base surface moisture content sensor array is set at the front of the construction vehicle to collect the base surface moisture content at multiple points. The construction vehicle is equipped with a temperature and humidity sensor to collect temperature and humidity, including tunnel gas temperature and tunnel gas relative humidity. The construction vehicle is equipped with a speed acquisition unit to collect vehicle speed. The controller receives the base surface moisture content, temperature and humidity, vehicle speed and historical film formation deviation values ​​at multiple points and generates a spraying area measurement matrix. The spraying area measurement matrix serves as the input basis for the water content distribution and film formation requirement characteristics of the spraying area in subsequent steps.

[0079] When establishing a spraying zone, the spraying width of the spraying device is recorded as... The number of spraying execution channels is recorded as The width of the tunnel foundation is denoted as The controller uses the width direction of the tunnel base surface as the coordinate axis of the spray width coordinate system. Taking the left boundary of the tunnel foundation as the origin of the coordinate system, based on... , and Calculate the width of the spraying area and along The direction sets the boundary of the spraying area, and the number of spraying areas is recorded as follows. And satisfy and Consistently, the spraying areas are numbered from left to right. ,in Increasing from one to The controller establishes a correspondence between the spraying execution channels and the spraying area numbers according to the installation sequence, so that the spraying control quantity of each spraying area can be executed independently by the corresponding spraying execution channel, and the order of the spraying area numbers is consistent with the order of the parallel input rows of the subsequent Kolmogorov-Arnold network.

[0080] When collecting moisture content data at multiple base points, the number of preset moisture measurement points within each spraying area is recorded as follows: The preset measurement point spacing is recorded as The preset detection distance before spraying is recorded as... The base surface moisture content sensor array is fixedly arranged in each spraying zone. The locations of the water content measuring points are aligned with the boundary of the spray area in the spray width direction and in the vehicle travel direction. The spacing is arranged such that the distance between the foremost water-bearing measuring point and the nozzle is [missing information]. This ensures that the distance between the remaining water content measuring points and the nozzle is less than [the required distance]. To ensure that the preset detection distance is maintained before spraying. The controller synchronously triggers the base surface moisture content sensor array at various moisture content measuring points during vehicle movement to collect the moisture content at multiple base surface measuring points, and numbers the spraying area accordingly. The The water content at each water-bearing measuring point is recorded as follows: ,in Increasing from one to When the location of the water content measuring point falls on the boundary of the spraying area, the controller uses a preset boundary assignment rule to assign the location of the water content measuring point to the spraying area with the smaller spraying area number, so as to ensure that each spraying area obtains the same number of measuring point values ​​of water content measuring points and forms a fixed dimension input.

[0081] When generating a moisture content sequence and simultaneously collecting environmental data, the controller sorts the moisture content of multiple datum points belonging to the same spraying area according to the location of the moisture content measuring points in the direction of vehicle travel, resulting in the spraying area number. Water-bearing sequence , water-bearing sequence As the initial input for water content distribution in subsequent steps, the controller collects... Temperature and humidity were collected at the same time, and the tunnel gas temperature was recorded as... The relative humidity of the tunnel gas is denoted as The controller synchronously collects the vehicle speed and records the vehicle speed as... The controller will , and The data is assigned to all spraying areas under the given spraying width according to the time of collection, so that the spraying area measurement values ​​of each spraying area include the temperature, humidity and vehicle speed at the same time.

[0082] When reading historical film-forming deviation values ​​and constructing the spray area measurement matrix, the controller reads the historical film-forming deviation value corresponding to the spray area number from the update results of the previous construction section, and assigns the spray area number as... Historical film formation deviation values ​​are denoted as The controller will determine the water content sequence. Temperature and humidity and Vehicle speed Historical film formation deviation value According to the spraying area number Combined into measurements of the sprayed area The measurement values ​​of all sprayed areas are then stacked in order of sprayed area number to form a sprayed area measurement matrix. Among them, the spraying area measurement matrix Each row corresponds to a spraying zone number, and the first row... The data are water-bearing sequences. The ordered water content, and the subsequent data are as follows: , , and This provides a consistent data organization method for subsequently setting the input structure of the univariate mapping subnetwork item by item according to the dimension.

[0083] In this embodiment, step S2 specifically includes:

[0084] Step S2 uses the spray area measurement matrix As input, the spray area measurement matrix According to the spraying area number Arranged row by row, the number of sprayed areas is recorded as follows: The measured value of each sprayed area is recorded as follows: This includes the moisture content of multiple base surfaces within the spraying area. to Tunnel gas temperature relative humidity of tunnel gas Vehicle speed Historical film formation deviation value ,in This represents the number of moisture measurement points within the spraying area.

[0085] During alignment and convergence, the controller reads In to As the spraying area number The controller assigns the set of water content measurement points to a spraying zone with a smaller zone number according to a preset boundary assignment rule if a water content measurement point is located at the boundary of a spraying zone. This ensures that the set of water content measurement points obtained by each spraying zone contains a fixed number of water content measurement points. The water content measurement points were determined, and subsequent feature dimensions were kept consistent across the entire spraying area.

[0086] When a water content distribution is formed, the controller sorts the set of water content measurement points according to the vehicle's direction of travel, and the sorting result is recorded as the water content sequence. The controller will determine the water content sequence. As a water content distribution and to maintain the water content distribution includes The moisture content, the controller from The maximum water content is selected as the maximum water content. The minimum water content is selected as the minimum water content. ,Will Sum of the water content and divide by Obtain the mean water content Maximum water content Minimum water content Mean water content As a statistical measure of water content distribution, it is included in subsequent calculations;

[0087] When calculating the wet point ratio, the controller uses the mean moisture content. As a benchmark, each point in the water content distribution is individually assessed to determine whether it is greater than [the required value]. The number of measuring points that meet the criteria is recorded as the number of wet points. The controller will count the number of wet points. Number of water content measuring points The ratio of the two values ​​is used as the wet point ratio. and wet point ratio value With spraying area number Linking them for future integration;

[0088] When calculating the wet-to-dry difference and generating the water content change characteristic, the controller uses the maximum water content value. With minimum water content The difference between wet and dry is obtained. To ensure that the wet-to-dry difference can represent the magnitude of water content abrupt changes in a dimensionless form that can be directly incorporated into the fusion, the controller uses the wet-to-dry difference... Divide by mean water content The normalized wet-dry difference was obtained, and the mean moisture content was calculated. When the water content is less than the preset lower limit of the mean water content, the preset lower limit of the mean water content is used to replace the mean water content. The controller adjusts the wet point ratio value according to the preset weight fusion rules. The data is then fused with the normalized wet-to-dry difference, and the preset weighted fusion rule includes a first weight coefficient. With the second weighting coefficient The controller will set the wet point ratio value. The weighted result is added to the weighted result of the normalized wet-dry difference to obtain the water content abrupt change characteristic. ;

[0089] When calculating the effect of water on film formation, the controller will use the tunnel gas temperature in the temperature and humidity settings. relative humidity of tunnel gas The input temperature and humidity correction mapping is implemented using a two-dimensional lookup table. The preset temperature range sequence and the preset humidity range sequence form a lookup table grid, and each grid cell stores the temperature and humidity correction coefficients. The controller according to Determine the temperature range according to Locate the humidity range and read the temperature and humidity correction coefficients of the grid cells. or When located within the grade range, the controller selects and The temperature and humidity correction coefficients for four adjacent grid cells are obtained by first performing linear interpolation along the temperature direction under a fixed humidity level to obtain two intermediate correction coefficients, and then performing linear interpolation along the humidity direction on the two intermediate correction coefficients to obtain the temperature and humidity correction coefficients. The controller calculates the influence of water on film formation based on the following formula. And the influence of water on film formation. Limited to the lower limit of the preset value range With respect to the upper limit of the preset value range Internally, the temperature and humidity correction mapping uses a water-affected function, which is specifically as follows:

[0090] ;

[0091] in, Number the spraying area The effect of water on film formation This is the lower limit of the preset value range. To preset the upper limit of the value range, Number the spraying area Temperature and humidity correction factor, As the first weighting coefficient, This is the second weighting coefficient. Number the spraying area The wet point ratio value, Number the spraying area The wet-dry difference, Number the spraying area The average moisture content, To perform the minimum value operation, This is for calculating the maximum value.

[0092] When generating the spray volume constraint parameters, the controller will consider the influence of water on film formation. The impact level is determined by comparing it with a preset impact level threshold, and the impact level is denoted as [insert threshold here]. The preset impact level threshold includes the first impact level threshold. Compared with the second level of impact threshold The controller is based on and , The comparison results will The controller will determine the impact level as either the first, second, or third level. The preset parameter table serves as an index for accessing the preset parameter table, which stores the injection volume jump limit value for each impact level. upper limit of spray volume Lower limit of spray volume Controller reads and corresponding , , This causes the injection volume to jump to the limit value. upper limit of spray volume Lower limit of spray volume With spraying area number correspond;

[0093] When generating the feature matrix of film formation requirements in the spraying area, the controller uses the moisture content components of the water content distribution as the preceding dimension in sequence, and the wet point ratio value... wet-dry difference Film formation affected by water value Vehicle speed Historical film formation deviation value Added in a fixed order as scalar dimensions, forming the characteristic quantities of film formation requirements in the spraying area. Film formation demand characteristics in the spraying area The number of dimensions is Add five, and fix the order of each dimension, so that each dimension corresponds item by item with the univariate mapping subnetwork of the Kolmogorov-Arnold network, and the controllers are stacked in order of spray zone number. Obtain the characteristic matrix of film formation requirements in the spraying area. It outputs a feature matrix of film formation requirements in the spraying area. Line-aligned injection volume jump limit value sequence upper limit sequence of injection volume Injection volume lower limit sequence This is used for combining layers to implement spray volume jump limits and boundary constraints.

[0094] In this embodiment, step S3 specifically includes:

[0095] In this embodiment, step S3 receives the film formation requirement feature matrix of the spraying area output in step S2. Injection volume jump limit value sequence upper limit sequence of injection volume With the lower limit value sequence of injection volume The data is then input into the Kolmogorov-Arnold network to calculate the target spray flow rate for the sprayed area. The number of sprayed areas is denoted as [missing information]. The spraying area is numbered as follows ,in Increasing from one to The number of water content measuring points within the spraying area is recorded as follows: The number of dimensions of the film-forming demand characteristics in the spraying area is denoted as... ,in for Add five;

[0096] The controller stacks all film formation demand feature quantities of the sprayed areas in order of sprayed area number to form a film formation demand feature matrix for the sprayed areas. And input it all at once into the input layer of the Kolmogorov-Arnold network, making the number of rows of the input tensor... The number of columns is , of which The row corresponds to the spraying area number. The film formation requirement characteristics of the spraying area, and the sequence of spray volume jump limit values. upper limit sequence of injection volume Injection volume lower limit sequence The spray zone number is aligned with the input tensor row and input to the composite layer, serving as an internal constraint parameter for jump correction and boundary limitation.

[0097] The Kolmogorov-Arnold network employs a structure of univariate mapping subnetworks and combined layers, with the controller based on the number of columns in the input layer. Configure a univariate mapping subnetwork for each column of the input layer, and denote the input dimension index as... ,in Increasing from one to ,when From one to When the input dimension is the water content component of the water content distribution, when for When the input dimension is the wet point ratio value, when for When the input dimension is the wet-to-dry difference, when for When, the input dimension is the value of film formation affected by water, when for When the input dimension is vehicle speed, for In this case, the input dimension is the historical film formation deviation value;

[0098] Each univariate mapping subnetwork contains input neurons, piecewise function neurons, and output neurons. The controller numbers the spraying areas. Dimension Index The input feature value is denoted as and will Enter the corresponding dimension index The input neuron, the piecewise function neuron, is divided into predefined piecewise nodes. Perform piecewise function mapping, with the preset set of piecewise nodes denoted as . The preset number of segment nodes is denoted as , Depend on The piecewise function neuron consists of nodes arranged in ascending numerical order, and stores the set of node output values ​​at each segment node. , and Correspondingly, piecewise function neurons Perform interval positioning and determine Calculate the adjacent segment node intervals that fall into the range. The relative position ratio within this interval, and the output values ​​at both ends of the interval. Linear interpolation is performed to obtain the piecewise function output value, which is then fed into the output neuron. The output neuron applies a preset output scaling factor to the piecewise function output value. With preset output bias Obtain film-forming contribution value This increases the contribution value of film formation. With input feature values Item-by-item correspondence;

[0099] The controller numbers the same spraying area. Total film-forming contribution value to The weighted fusion neurons converge to the combined layer, and the weighted fusion neurons index each dimension. Configure weight parameters and configure fusion bias The initial spray flow rate value is obtained by performing a weighted summation of the film-forming contribution values ​​and superimposing the fusion bias. To make the initial spray flow rate value With spraying area number correspond;

[0100] The jump correction unit of the combined layer calculates the difference in initial spray flow rate values ​​between adjacent spray zones according to the spray zone numbering order and performs truncation correction. The boundary limitation unit of the combined layer performs boundary limitation on the truncation correction result. For spray zone number one, the jump correction unit uses the initial spray flow rate value as the jump correction spray flow rate value and records it as... Boundary limiting unit will With the upper limit of injection volume and the lower limit of spray volume After comparing and defining the boundary conditions, the target spray flow rate value for the spraying area is obtained, and the spraying area is numbered. In cases where the number is greater than one, the combined layer is numbered according to the spraying area. Jump correction spray flow rate value As a benchmark, based on the difference in initial spray flow rates between adjacent spray zones With the limit value of injection volume jump After completing the truncation correction and applying boundary constraints to the correction results, the spraying zone number is obtained using the following formula. Target spray flow rate value of the spraying area The target spray flow rate value in the spraying area is determined by the target spray function, which is specifically:

[0101] ;

[0102] in, Number the spraying area The target spray flow rate value for the spraying area. Number the spraying area The upper limit of the spray volume, Number the spraying area The lower limit of the spray volume, Number the spraying area The jump correction spray flow rate value, Number the spraying area The limit value for the jump in injection volume, Number the spraying area The initial spray flow rate value, Number the spraying area The initial spray flow rate value, Number the spraying area. To perform the minimum value operation, This is for calculating the maximum value.

[0103] Jump correction unit in calculation First, adjust the initial spray flow rate difference between adjacent spraying zones according to the spray volume jump limit value. Truncate the difference while preserving its direction, then compare the truncated result with... Add them together to get the jump correction spray flow rate value ,in This is the intermediate result before the boundary constraint operation in the formula is executed; the boundary constraint unit is... The input completion boundary is defined to obtain and will As the next spraying zone number Jump correction reference input;

[0104] The controller outputs the target spray flow rate values ​​for all spraying zones in sequence according to the spraying zone number, forming a target spray flow rate sequence. and the target spray flow sequence As the setpoint input for the subsequent proportional-integral-derivative controller, the target spray flow rate value of the spraying area is determined by the film-forming contribution value output by the univariate mapping sub-network and the spray volume jump limit value, spray volume upper limit value, and spray volume lower limit value of the combined layer.

[0105] In this embodiment, step S4 specifically includes:

[0106] Step S4 is used to implement the target spray flow rate values ​​of each spraying zone output by the Kolmogorov-Arnold network into the spray execution channel in a closed loop. The spraying zone number is denoted as... The number of sprayed areas is recorded as The controller has a built-in proportional-integral-derivative (PID) controller, which operates at a fixed control period. run, The time interval between two consecutive control updates;

[0107] The controller reads the spray zone number. The target spray flow rate value for the spraying area is recorded as... ,Will The flow rate setpoint for the spray execution channel is recorded as follows: ,in and If the values ​​are consistent, the controller will assign the spray zone number accordingly. The setpoint is sent to the spraying execution channel corresponding to the spraying area, so that the proportional-integral-derivative controller of each spraying execution channel only receives the flow rate setpoint corresponding to the spraying area number;

[0108] Each spraying channel's feed line is equipped with a flow sensor and a pressure sensor, and the controller operates in each control cycle. Collection of spraying area number The traffic feedback value is recorded as Collect pressure feedback values ​​and record them as The controller records the flow deviation as and will Set to flow rate setting Subtract traffic feedback value The difference, the controller sets a preset reference pressure The pressure deviation is denoted as and will Set as pressure feedback value Subtract the preset reference pressure The difference is determined by the controller based on the pressure deviation. The sign and amplitude are selected from the preset pressure correction table using the pressure correction coefficient. The preset pressure correction table maps the pressure deviation range to the pressure correction coefficient value, and the controller records the corrected flow deviation as... and will Set as pressure correction factor With flow deviation The product of these factors allows the pressure feedback value to participate in correcting the trend of flow deviation changes.

[0109] The proportional-integral-derivative controller receives the corrected flow deviation. It also generates the valve opening adjustment amount, and sets the proportional-integral-derivative controller to the proportional gain. Integral gain With differential gain The proportional-integral-derivative controller sets the proportional term as the proportional gain. With the corrected flow deviation The product of and , the proportional-integral-derivative controller records the integral state as Set the integral state of the current control cycle to the integral state of the previous control cycle and the corrected flow deviation. Multiply by control period The sum, and the integral state Implement integral limiting to confine the integral state to the lower integral limit. With the maximum points Between these points, the proportional-integral-derivative controller sets the derivative term to the corrected flow deviation. The difference between the current control cycle and the previous control cycle is divided by the control cycle. The result, and apply differential gain to the differential term. The proportional-integral-derivative controller combines the proportional term and the integral gain. With integral state The product and the weighted sum of the differential terms are used to obtain the valve opening adjustment amount, which is recorded as . The controller records the current valve opening of the spray execution channel as... ,Will With valve opening adjustment amount The sums form the updated valve opening, and a valve opening limit is applied to the updated valve opening, restricting it to the lower limit of the valve opening. With upper limit of valve opening Between these times, the regulating valve is then driven to operate according to the updated valve opening degree;

[0110] After the valve opening is changed, the flow rate in the feed pipeline enters a new steady state. The controller continuously collects the flow feedback value within the preset steady-state sampling window. The average value of the flow feedback values ​​within the sampling window is used to obtain the actual spray flow rate, which is then recorded as follows: and the actual spraying flow rate With spraying area number The associated output allows the actual spray flow rate to be used as the spray input for calculating subsequent film formation continuity index values.

[0111] In this embodiment, step S5 specifically includes:

[0112] Step S5 is used to calculate the film formation continuity index value based on the post-spray appearance characteristics corresponding to the actual spray flow rate, and output the current film formation deviation value corresponding to the spray area number, which is denoted as [spray area number]. The number of sprayed areas is recorded as The spraying area number output in step S4 The actual spray flow rate is recorded as ;

[0113] The controller will display the actual spray flow rate. With spraying area number Align and Input a preset flow rate delay mapping table to determine the post-spray detection delay. A preset traffic latency mapping table stores multiple traffic tier values ​​and their corresponding latency values, allowing the controller to locate... The interval is defined, and linear interpolation is performed based on the time delay values ​​at both ends of the interval to obtain the value. The controller reads and records the vehicle speed corresponding to the spraying time. Post-spray detection delay Convert the vehicle speed to the post-spray detection distance and record it as follows: The controller accumulates the distance traveled by calculating the vehicle speed from the moment of spraying and records it as such. ,when Not less than The appearance acquisition device is triggered to collect post-spray appearance features, which include post-spray images. and reflectivity Among them, reflectivity The reflected light intensity sample value is output by the appearance acquisition device along the spray width direction and aligned with the spray area number;

[0114] The controller uses the spray zone boundary in the spray width coordinate system to analyze the post-spray image. The spray area appearance segment is obtained by cropping. Number the spraying area The left boundary coordinates are: The right boundary coordinates are: The appearance acquisition device performs coordinate calibration in advance to obtain linear mapping parameters from the spray width coordinates to the image pixel column. and Two calibration lines with known coordinates are set in the spray width direction. Calibration images are acquired and the pixel column numbers corresponding to the two calibration lines are extracted. The correspondence between the two sets of coordinates and pixel column numbers is used to obtain the... and The controller will and Map the left and right crop column numbers and round them down to obtain the cropped column number. The controller Perform film-forming region segmentation to obtain film-forming pixels and non-film-forming pixels: Intensity of reflection... Input a preset reflection threshold mapping table to obtain the segmentation threshold ,Will Convert to grayscale and read the grayscale values ​​pixel by pixel, ensuring the grayscale values ​​are not less than... The pixels that are marked as having formed a film are labeled as such, and the remaining pixels are labeled as not having formed a film, thus forming a binary mask for film formation. Controller statistical film formation binary mask The number of film-forming pixels is denoted as The number of pixels without film formation is recorded as follows. The total number of pixels is recorded as The film coverage is obtained by comparing the number of pixels forming the film with the total number of pixels. The exposed area ratio is obtained by comparing the number of unfilmed pixels with the total number of pixels. ;

[0115] The controller is used in the binary mask for film formation. The controller detects interruptions in the film-forming pixels along the vehicle's travel direction, and scans the spray area's appearance segment line by line along the travel direction. For each line, it calculates the ratio of the number of film-forming pixels in that line to the total number of pixels in that line, and sets the ratio below a preset line film-forming threshold. Continuous line segments are identified as interrupted segments, and segments with a length not less than the preset minimum interruption length are allowed to pass through. The number of interrupted segments is counted, and the count result is used as the number of broken bands and recorded as . The controller will count the number of broken bands. With the preset maximum number of broken bands Compare and truncate, then divide the number of broken bands by 1 / 2. The breakage ratio is obtained, and the breakage score is obtained by subtracting the breakage ratio from the score. The controller then calculates the film coverage. As a coverage score, As the score for exposing the underside, the controller multiplies the coverage score, the underside score, and the band break score by preset weights. , , The three weighted results are added together to obtain the film formation continuity index value. And the film formation continuity index value With spraying area number Associated storage;

[0116] The controller will set the film formation continuity index value. With film formation continuity threshold The current film formation deviation value is obtained by comparison. ,when Below At that time, the controller will and The difference is used as the current film formation deviation value. and number the spraying area Related output, when Not less than At that time, the controller will display the current film formation deviation value. Set to zero and number the spraying area. Related output.

[0117] In this embodiment, step S6 specifically includes:

[0118] Step S6 updates the current film-forming deviation value output in step S5 to the historical film-forming deviation value used in the next construction section. The spraying area number is denoted as [missing information]. The number of sprayed areas is recorded as Spraying area number The current film formation deviation value is denoted as Spraying area number Historical film formation deviation values ​​are denoted as ,

[0119] The controller reads the current film formation deviation value of each spray zone according to the spray zone number. And according to the spraying area number from one to... The sequential stacking forms the current deviation vector of the spray area. The controller synchronously reads the historical film formation deviation values ​​corresponding to the same spraying area number from the storage unit. And stacked in the same order to form a historical deviation vector. This causes the current deviation vector of the spray area to... Deviation vector from history Each one is matched with the same spraying zone number;

[0120] The controller uses the current deviation vector of the spray area. For input, number each spraying area. The controller updates historical film formation deviation values ​​using preset update coefficients. ,in The coefficients used for weighted fusion are limited to a range of zero to one, and satisfy the following conditions: Non-zero spray zone number The controller will display the current film formation deviation value. Multiply by the update factor Obtain the current weighting term and the historical film formation deviation value. Multiply The historical weighted terms are obtained, and the current weighted terms are added to the historical weighted terms to obtain the updated historical film formation deviation value, which is then recorded as follows. , for satisfying Spraying area number with a value of zero The controller uses a preset attenuation coefficient. The historical film formation deviation values ​​are updated by attenuation, where The attenuation coefficient is limited to a range of zero to one, and the controller uses historical film formation deviation values. With attenuation coefficient Multiplying yields the updated historical film formation deviation value. The controller stacks the update history film formation deviation values ​​of all sprayed areas in order of spray area number to form an update history deviation vector. ;

[0121] The controller updates the historical deviation vector. Smoothing is performed between adjacent spray zones along the spray width direction, with the controller having a preset deviation jump limit threshold. And the smoothing historical film formation deviation value of sprayed area number one is recorded as ,Will Set as Number the spraying area If the number is greater than one, the controller calculates the spray zone number. With spraying area number The deviation difference is recorded as ,in Set as minus The difference is used by the controller to determine the deviation difference. Amplitude and deviation jump limit threshold Comparison, when the deviation difference does not exceed At that time, Set as When the deviation difference exceeds At that time, the deviation difference Truncation is equal to the amplitude And maintain Keeping the positive and negative directions unchanged, the difference in deviation after truncation is then compared with... Adding them together gives This allows the historical film formation deviation value difference between adjacent spraying areas to be subject to a deviation jump limit threshold constraint, and the controller presets a lower limit for the deviation value range. Upper limit of the deviation value range Each spraying area was numbered. Smooth historical film formation deviation value Scope of execution is limited, when Less than When Set as ,when Greater than When Set as ;

[0122] The controller will smooth out historical film deposition deviation values. The measured values ​​of the sprayed area in the next construction section are recorded as historical film-forming deviation values, and these historical film-forming deviation values ​​are still recorded as... When generating the characteristic quantities of film formation requirements in the spraying area, the controller will use historical film formation deviation values. The scalar dimension of the film formation requirement characteristics of the spraying area is input into the Kolmogorov-Arnold network, so that the historical film formation deviation value participates in the calculation of the target spraying flow rate value of the spraying area through the corresponding univariate mapping subnetwork.

[0123] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An adaptive control method for spraying waterproof adhesive layer in tunnels based on Kolmogorov-Arnold network and PID, characterized in that, Includes the following steps: S1. Divide the tunnel base surface into spraying zones according to the spraying width, and obtain the base surface moisture content, temperature and humidity, vehicle speed and historical film formation deviation values ​​at multiple points in the spraying zone to form the spraying zone measurement values. S2. Based on the multi-point base surface moisture content measured in the spraying area, the moisture content distribution is obtained by converging the spraying area. The moisture content distribution characterizes the wet and dry areas and reflects the wet-dry transition. The wet point ratio and wet-dry difference are calculated from the moisture content distribution. Then, the film formation water influence value is calculated in combination with temperature and humidity. Based on the film formation water influence value, the spray volume jump limit value, the upper limit value and the lower limit value of the spray volume are determined. The film formation demand characteristic quantity of the spraying area is generated by integrating the moisture content distribution, wet point ratio, wet-dry difference, film formation water influence value, vehicle speed and historical film formation deviation value. S3. Input the film formation demand characteristics of the spraying area, including the water content distribution and the influence of water on film formation, into the Kolmogorov-Arnold network. The network includes a univariate mapping subnetwork and a combination layer that correspond to each of the film formation demand characteristics of the spraying area. The univariate mapping subnetwork maps each item to obtain the film formation contribution value. The combination layer weights and fuses the film formation contribution value. Under the constraint of the spray volume jump limit value, the spray volume difference between adjacent spraying areas is corrected. Under the constraint of the upper limit value and the lower limit value of the spray volume, the correction result is limited to form the target spray flow rate value of the spraying area. S4. Using the target spray flow rate of the spraying area as the set value, the proportional-integral-derivative controller adjusts the valve opening based on the flow feedback value and the pressure feedback value to obtain the actual spray flow rate; S5. Based on the actual spraying flow rate, collect the post-spray appearance characteristics of the sprayed area, calculate the film formation continuity index value and compare it with the film formation continuity threshold to form the current film formation deviation value. S6. Update the historical film formation deviation value according to the spraying area based on the current film formation deviation value. The historical film formation deviation value is used as the historical film formation deviation value in the measurement value of the spraying area of ​​the next construction section.

2. The adaptive control method for setting the spraying of tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID according to claim 1, characterized in that, Step S1 is as follows: Based on the spraying width of the spraying device, the number of spraying execution channels, and the width of the tunnel base, a spraying width coordinate system is established and the spraying area boundary is set along the spraying width direction. A spraying area number corresponding to the spraying area boundary is generated so that each spraying area corresponds to a spraying execution channel. In each spraying area, the positions of water content measuring points are set according to the preset number of water content measuring points and the preset spacing between measuring points. The water content of the base surface at multiple points within the preset detection distance before spraying is collected, and the water content of the base surface at multiple points is assigned to the corresponding spraying area according to the position of the water content measuring points to ensure that each spraying area obtains the same number of measuring point values. The moisture content of multiple base surfaces belonging to the same spraying area is sorted according to the vehicle's direction of travel to form a moisture content sequence. Temperature, humidity and vehicle speed are collected simultaneously. Temperature and humidity include tunnel gas temperature and tunnel gas relative humidity. The temperature and humidity are assigned to each spraying area under the spraying width according to the collection time. Read the historical film formation deviation value corresponding to each spraying zone number, combine the water content sequence, temperature and humidity, vehicle speed and historical film formation deviation value according to the spraying zone number to form the spraying zone measurement value, and arrange the spraying zone measurement value in the order of the spraying zone number to form a spraying zone measurement matrix for parallel processing.

3. The adaptive control method for setting the spraying of tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID according to claim 1, characterized in that, Step S2 is as follows: The multi-point base surface moisture content in the spraying area measurement values ​​is aligned with the spraying area boundary according to the location of the moisture content measurement points. The multi-point base surface moisture content falling into the same spraying area is aggregated into a set of moisture content measurement points for that spraying area. The moisture content measurement points located at the spraying area boundary are allocated to a single spraying area according to the preset boundary assignment rules. The set of water content measurement points is sorted according to the vehicle travel direction to obtain the water content sequence. The water content sequence is used as the water content distribution and the water content distribution includes the number of water content measurement points in the spraying area. The maximum water content, minimum water content, and mean water content are calculated from the water content sequence as statistical quantities. Using the mean water content as a comparison benchmark, each point in the water content sequence is judged to determine whether the water content is greater than the mean water content. The number of measuring points that meet the judgment is counted and compared with the number of water content measuring points in the spraying area to obtain the wet point ratio value. The wet point ratio value is associated with the spraying area number. The wet-dry difference is determined by the difference between the maximum and minimum water content, and the wet point ratio and wet-dry difference are fused according to the preset weight fusion rule to obtain the water content change characteristics, so that the water content change characteristics can simultaneously reflect the change in water film coverage area and the water content change amplitude. The temperature and relative humidity of the tunnel gas in the temperature and humidity are input into the temperature and humidity correction mapping. The correction amount is applied to the water content change characteristics and the value of the influence of water on film formation is output. The value of the influence of water on film formation is limited to a preset range as the calculation input for the spray volume constraint. The influence level is determined by comparing the film-forming water-affected value with the preset influence level threshold. Based on the influence level, the spray volume jump limit value is selected and used to limit the spray volume difference between adjacent spraying areas. Based on the influence level, the film-forming range of the material is corrected to obtain the upper limit value and lower limit value of the spray volume, so that the spray volume jump limit value, the upper limit value and the lower limit value of the spray volume correspond to the spraying area number. The water content components of the water content distribution are sequentially used as the preceding dimensions. The wet point ratio, wet-dry difference, film formation affected by water, vehicle speed, and historical film formation deviation are added in a fixed order as scalar dimensions to generate the film formation demand feature quantity of the spraying area. The number of dimensions of the film formation demand feature quantity of the spraying area is equal to the number of water content measurement points in the spraying area plus five. Each dimension is made to correspond to the univariate mapping subnetwork of the Kolmogorov Arnold network. The feature matrix of film formation demand of the spraying area is formed by arranging the feature matrix of film formation demand of the spraying area in the order of the spraying area number. The spray volume jump limit value, spray volume upper limit value, and spray volume lower limit value are output and aligned with the rows of the feature matrix of film formation demand of the spraying area.

4. The adaptive control method for setting the spraying of tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID according to claim 3, characterized in that, When inputting the tunnel gas temperature and relative humidity from the temperature and humidity data into the temperature and humidity correction mapping, applying a correction to the water content abrupt change characteristics, and outputting the film formation water influence value, the temperature and humidity correction mapping adopts a water influence function, wherein the water influence function is specifically: ； in, Number the spraying area The effect of water on film formation This is the lower limit of the preset value range. To preset the upper limit of the value range, Number the spraying area Temperature and humidity correction factor, As the first weighting coefficient, This is the second weighting coefficient. Number the spraying area The wet point ratio value, Number the spraying area The wet-dry difference, Number the spraying area The average moisture content, To perform the minimum value operation, This is for calculating the maximum value.

5. The adaptive control method for setting the spraying of tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID according to claim 1, characterized in that, Step S3 is as follows: The film formation demand feature of the spraying area is arranged in the order of the spraying area number to form the film formation demand feature matrix of the spraying area. The film formation demand feature matrix of the spraying area is input into the input layer of the Kolmogorov-Arnold network at once, so that the number of rows of the input tensor is the number of spraying areas and each row corresponds to a film formation demand feature of a spraying area. At the same time, the spray volume jump limit value, spray volume upper limit value and spray volume lower limit value are aligned according to the spraying area number and input into the combination layer as constraint parameters. Based on the fact that the dimension of the input layer is the sum of the number of water-containing measurement points in the spraying area and the five scalar feature dimensions, a corresponding univariate mapping subnetwork is set for each dimension of the input layer in the Kolmogorov-Arnold network. In each univariate mapping subnetwork, input neurons, piecewise function neurons and output neurons are set in sequence to fix the mapping path from a single input feature to the film-forming contribution value. Each water content component of the water content distribution is input into the corresponding univariate mapping subnetwork to obtain the water content film formation contribution value component. The wet point ratio value, wet-dry difference value, film formation affected by water value, vehicle speed and historical film formation deviation value are input into the corresponding univariate mapping subnetwork to obtain the corresponding film formation contribution value. The piecewise function neuron performs piecewise function mapping on the input features according to the preset piecewise nodes and outputs the nonlinear film formation contribution value. All film-forming contribution values ​​of the same spraying area are aggregated into the weighted fusion neuron of the composite layer. The film-forming contribution values ​​are weighted and summed using preset weight parameters to obtain the initial spraying flow rate value, so that the initial spraying flow rate value corresponds to the spraying area number. The initial spray flow rate value is input into the jump correction unit of the combined layer. The difference between the initial spray flow rate values ​​of adjacent spray areas is calculated according to the spray area number order. The magnitude of the difference between the initial spray flow rate values ​​is compared with the corresponding spray volume jump limit value and a jump judgment is made. If the spray volume jump limit value is met, the initial spray flow rate value is retained. If the spray volume jump limit value is not met, the difference between the initial spray flow rate values ​​is truncated according to the spray volume jump limit value and the direction of the difference is maintained to generate the jump correction spray flow rate value. The jump-corrected spray flow rate value is input into the boundary limiting unit of the combined layer. The jump-corrected spray flow rate value is compared with the corresponding upper and lower limits of the spray volume and boundary limiting is performed. When the jump-corrected spray flow rate value exceeds the upper limit of the spray volume, the upper limit of the spray volume is output. When the jump-corrected spray flow rate value is lower than the lower limit of the spray volume, the lower limit of the spray volume is output. The target spray flow rate value of the spray area is obtained and output in the order of the spray area number.

6. The adaptive control method for setting the spraying of tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID according to claim 5, characterized in that, When the combined layer performs jump correction and boundary limitation on the initial spray flow rate value based on the spray flow jump limit value, the upper limit value, and the lower limit value to obtain the target spray flow rate value of the spray area, the target spray flow rate value of the spray area is determined by the target spray function, wherein the target spray function is specifically: ； in, Number the spraying area The target spray flow rate value for the spraying area. Number the spraying area The upper limit of the spray volume, Number the spraying area The lower limit of the spray volume, Number the spraying area The jump correction spray flow rate value, Number the spraying area The limit value for the jump in injection volume, Number the spraying area The initial spray flow rate value, Number the spraying area The initial spray flow rate value, Number the spraying area. To perform the minimum value operation, This is for calculating the maximum value.

7. The adaptive control method for setting the spraying of tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID according to claim 1, characterized in that, Step S4 is as follows: The target spray flow rate value for each spraying zone is issued to the spraying execution channel corresponding to the spraying zone according to the spraying zone number, and used as the flow rate set value for the proportional-integral-derivative controller; Flow feedback and pressure feedback values ​​are collected in the material supply pipeline of the spraying execution channel. The flow feedback value is compared with the flow set value to obtain the flow deviation, and the trend of the flow deviation is corrected based on the pressure feedback value. The proportional-integral-derivative controller generates a valve opening adjustment amount based on the corrected flow deviation, and then drives the regulating valve to change the valve opening after limiting the valve opening adjustment amount. The actual spray flow rate is generated based on the changed valve opening, and the actual spray flow rate is associated with the spray area number for use in calculating the film formation continuity index value.

8. The adaptive control method for setting the spraying of tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID according to claim 1, characterized in that, Step S5 is as follows: Align the actual spray flow rate with the spray area number, determine the post-spray detection delay according to the preset flow delay mapping table, and convert the post-spray detection delay into the post-spray detection distance according to the vehicle speed. When the post-spray detection distance is reached, trigger the appearance acquisition device to collect the post-spray appearance features, which include the post-spray image and reflectivity. The appearance features after spraying are cut out according to the boundary of the sprayed area to obtain the appearance segment of the sprayed area. The appearance segment of the sprayed area is divided into film-forming regions to obtain film-forming pixels and non-film-forming pixels, and the film coverage and exposed area ratio are calculated accordingly. In the appearance segment of the spraying area, the interruption segment of the film-forming pixel is detected along the vehicle's direction of travel, and the number of interruption segments is counted as the number of broken bands. The film coverage, exposed area ratio and number of broken bands are fused according to preset weights to generate a film continuity index value. The current film formation deviation value is obtained by comparing the film formation continuity index value with the film formation continuity threshold. When the film formation continuity index value is lower than the film formation continuity threshold, the difference between the film formation continuity threshold and the film formation continuity index value is used as the current film formation deviation value and output in association with the spraying area number. When the film formation continuity index value is not lower than the film formation continuity threshold, the current film formation deviation value is set to zero.

9. The adaptive control method for setting the spraying of tunnel waterproof bonding layer based on Kolmogorov-Arnold network and PID according to claim 1, characterized in that, Step S6 is as follows: The current film formation deviation value is mapped one by one with the historical film formation deviation value according to the spray area number to obtain the current deviation vector of the spray area; Using the current deviation vector of the spraying area as input, the historical film formation deviation value and the current film formation deviation value are weighted and fused using a preset update coefficient to obtain the updated historical film formation deviation value. When the current film formation deviation value is zero, the historical film formation deviation value is updated by attenuation according to a preset attenuation coefficient. For the updated historical film formation deviation value, calculate the deviation difference between adjacent spray areas along the spray width direction, compare the deviation difference with the preset deviation jump limit threshold, and perform truncation correction on the deviation difference that exceeds the preset deviation jump limit threshold to obtain the smoothed historical film formation deviation value, and perform preset deviation value range limitation on the smoothed historical film formation deviation value. The historical film formation deviation value is smoothed and used as the historical film formation deviation value in the measurement value of the spraying area of ​​the next construction section. The historical film formation deviation value is then used as the scalar dimension of the film formation demand characteristic quantity of the spraying area and input into the Kolmogorov-Arnold network.

Images