Intelligent pipe cooling circulation control system and method based on ship lock construction

Abstract

The application discloses an intelligent pipe cooling circulation control system and method based on ship lock construction, which comprises the following steps: dividing a ship lock construction structure into A pipe cooling circulation control subareas; covering the A pipe cooling circulation control subareas by B cooling circulation pipelines; receiving multi-dimensional data of each pipe cooling circulation control subarea by each edge key control node, and transmitting the data to a cloud cooling control service; after receiving the data, the cloud cooling control service generates A cooling circulation load demand risk coefficients based on a first deep learning algorithm; if it is judged that at least one of the A cooling circulation load demand risk coefficients is at risk, then the B cooling circulation pipelines are subjected to pipe cooling circulation control based on a second algorithm, and a PID control algorithm is used to adjust the cooling water flow and the cooling water temperature of the pipelines; the application realizes precise pipe cooling circulation control adjustment, improves the ship lock construction quality and reduces the risk of concrete hardening cracking through deep learning algorithm prediction and monitoring and the introduction of an unmanned aerial vehicle for the inspection of the outer surface of the cooling pipe.

Description

Technical Field

[0001] This invention relates to the field of pipe cooling circulation control technology in artificial intelligence, and more specifically, to an intelligent pipe cooling circulation control system and method based on lock construction. Background Technology

[0002] During the construction of ship locks, the construction quality of the concrete structure is of paramount importance. Because ship lock structures typically have large volumes, complex shapes, and high load-bearing requirements, the concrete undergoes temperature changes after pouring due to factors such as heat of hydration, leading to internal stresses. If these stresses are not effectively controlled, they can cause cracks, seriously affecting the structural safety and durability of the ship lock.

[0003] Traditional methods of concrete temperature control primarily rely on manual monitoring and adjustment of the cooling system. For example, construction workers periodically read temperature data by embedding temperature sensors inside the concrete, and then manually open or close cooling water valves based on experience to regulate the flow rate and control the concrete temperature. However, this method has several drawbacks. Firstly, the frequency of manual monitoring is limited, making it impossible to obtain real-time, continuous information on temperature changes within the concrete. This can easily lead to missing critical moments in temperature fluctuations, resulting in untimely and inaccurate temperature control. Secondly, manually adjusting the cooling system depends on the experience and subjective judgment of the construction workers. The varying skill levels of different personnel make it difficult to guarantee consistent and stable cooling effects. Furthermore, in the complex environment of a ship lock construction site, numerous interfering factors, such as mechanical vibration and personnel activity, can affect the accuracy of manual monitoring and adjustment.

[0004] With the continuous advancement of technology, the application of intelligent control technology in the construction field has gradually attracted attention. In the construction of ship locks, the intelligent pipe cooling circulation control method can effectively solve the problems existing in traditional methods. By introducing advanced sensor technology, automated control technology, and computer algorithms, it is possible to achieve real-time and accurate monitoring and automatic adjustment of concrete temperature, thereby improving the construction quality of ship locks and reducing construction risks. This has significant practical implications and broad application prospects.

[0005] The existing technology has the following problems:

[0006] 1. No zoned temperature control was implemented for the structure during the construction of the ship lock;

[0007] 2. There was no comprehensive and accurate analysis of the internal and external environmental factors for the structural zones of the lock construction;

[0008] 3. The risk was not accurately predicted based on the comprehensive factors of each zone using neural network algorithms;

[0009] 4. When it is found that there is a regulation risk in a certain partition, it is impossible to dynamically and precisely adjust the control parameters of the tube cooling cycle according to the risk level. Summary of the Invention

[0010] The purpose of the present invention is to provide an intelligent tube cooling cycle control system and method based on lock construction to solve the above problems existing in the prior art.

[0011] Specifically, this application is as follows:

[0012] An intelligent tube cooling cycle control method based on lock construction, the method comprising the following steps:

[0013] Divide the lock construction structure into pier areas and lock wall areas, divide key points in the pier areas and lock wall areas, generate A tube cooling cycle control partitions, and preset A edge key control nodes for the A tube cooling cycle control partitions. Among them, each partition represents an independent structure type in the pier and the lock wall, and each partition contains several key points, and the key points are used for data collection;

[0014] Cover the A tube cooling cycle control partitions through B cooling cycle pipelines, and each cooling cycle pipeline is equipped with an electromagnetic regulating valve and a turbine flowmeter, where B ≤ A;

[0015] Each edge key control node receives multi-dimensional data of each tube cooling cycle control partition, preprocesses the multi-dimensional data to generate preprocessed data, and transmits the preprocessed data to the cloud cooling control service; the multi-dimensional data includes key node basic data, key node inspection data on the outer surface of the cooling tube, key node scaling data inside the cooling tube, key node ambient temperature, key node ambient humidity, key node air velocity, key node tube cooling equipment status; the preprocessed data includes first preprocessed data and second preprocessed data, where the first preprocessed data does not include key node inspection data on the outer surface of the cooling tube, and the second preprocessed data at least includes key node inspection data on the outer surface of the cooling tube;

[0016] After the cloud cooling control service receives the A preprocessed data of the A edge key control nodes, it uses the first deep learning algorithm to predict the A preprocessed data to generate A cold cycle load demand risk coefficients; the cold cycle load demand risk coefficients are set to three values according to the risk type: the first risk value b1, the second risk value b2, and the third risk value b3, where b1 < b2 < b3; the tube cooling cycle control partition and several key points under the tube cooling cycle control partition are both set with risk types, and the risk types include no risk type, low risk type, high risk type, and equipment abnormal type. Among them, the risk type of each tube cooling cycle control partition is calculated from the sum of the risk types of several key points controlled under each tube cooling cycle control partition;

[0017] If it is determined that at least one of the cold cycle load demand risk coefficients in the A cold cycle control zone is at risk, the second algorithm is used to control the B cooling cycle pipelines based on the A cold cycle load demand risk coefficients to generate a pipe cooling cycle control decision.

[0018] Based on the set values ​​of the cooling circulation control decision output of the B cooling circulation pipelines, the PID control algorithm is used to adjust the cooling water flow rate and cooling water temperature of each pipeline according to the set values.

[0019] Furthermore, the process of dividing the lock construction structure into a pier area and a wall area, and then dividing the pier area and wall area into key points to generate A pipe cooling circulation control zones, specifically includes:

[0020] Based on GPS technology, the initial flight path of the drone is determined, and a camera is installed on the drone.

[0021] The drone flew along its initial flight path and collected basic information and images of the lock construction structure using its onboard camera.

[0022] Based on basic information and image sets, combined with image recognition algorithms, the lock construction structure is divided into two regions: the lock pier region and the lock wall region.

[0023] Key points in two areas are collected to obtain coordinate and image information of several key points. Clustering analysis is performed on the coordinate information of several key points to generate a first set of key points. Using a cosine similarity algorithm, the feature vectors of image information of historical lock construction structures in the database and the image information feature vectors corresponding to the first set of key points are calculated to generate a second set of key points. The total number of the second set of key points is represented by A pipe cooling cycle control zones. The total number of the sets is greater than 1. The image information of the historical lock construction structures represents key point marker information, which is used for the division of pipe cooling cycle control zones.

[0024] The coordinates corresponding to the image set, the basic information, and the second key point set are input into the navigation path planning model, and the navigation path information of the UAV is output. The navigation path information is used for the UAV to collect data on the A pipe cooling cycle control partitions. The navigation path planning model is built by performing BP neural network training based on the navigation path decision tree. The navigation path decision tree is built by mapping and associating partition identification nodes and partition key points configured through decision processing of historical sample data.

[0025] Furthermore, each critical edge control node receives multidimensional data from each pipe cooling cycle control zone and preprocesses the multidimensional data to generate preprocessed data, including:

[0026] The ambient temperature of each pipe cold circulation control zone is collected by temperature sensor, the ambient humidity of each pipe cold circulation control zone is collected by humidity sensor, the outer surface inspection data of the cooling pipe of each pipe cold circulation control zone is collected by drone, the internal scaling data of the cooling pipe of each pipe cold circulation control zone is collected by pipe wall vibration sensor, and the air velocity of each pipe cold circulation control zone is collected by air velocity sensor.

[0027] The multidimensional data is collected at fixed intervals, and the collected data is transmitted to each critical edge control node.

[0028] Furthermore, each critical edge control node receives multidimensional data from each pipe cooling cycle control zone, preprocesses the multidimensional data, and generates preprocessed data, which also includes:

[0029] The sensor-acquired data is preprocessed sequentially using spatiotemporal alignment, wavelet filtering, feature enhancement algorithm, and confidence weighting to generate the first preprocessed data.

[0030] The image data acquired by the UAV is preprocessed by sequentially using image enhancement algorithm, shift image segmentation algorithm, and keyframe extraction algorithm to generate the second preprocessed data.

[0031] Furthermore, each critical edge control node receives multidimensional data from each pipe cooling cycle control zone, preprocesses the multidimensional data, and generates preprocessed data, which also includes:

[0032] Confidence modeling is performed on the confidence scores of the first and second preprocessed data, assuming that both the first and second preprocessed data follow a normal distribution:

[0033]

[0034] Where, x j θ represents the output values ​​of the first and second preprocessed data. j This represents the mean. The noise mean square error, This indicates that the measured values ​​follow a mean of θ. j The mean squared error is The normal distribution;

[0035] The optimized posterior distribution is inferred using the Markov chain Monte Carlo method, and the confidence weights of each first and second preprocessed data are calculated.

[0036] A preset threshold is set, and the confidence weights of the first and second preprocessed data are compared with the preset threshold. The first and second preprocessed data that meet the threshold requirements are sent to the edge key control node.

[0037] Furthermore, the first deep learning algorithm includes:

[0038] A unique cold cycle load demand risk function is set for each pipe cold cycle control zone. The cold cycle load demand risk function is used to analyze the preprocessed data of several key points under each pipe cold cycle control zone. The preprocessed data of several key points must be input into the cold cycle load demand risk function for analysis. The analysis results are used to determine the risk type of the key points. The key points are classified into risk types through the cold cycle load demand risk function. The risk threshold of the preprocessed data of the key points is calculated through the cold cycle load demand risk function.

[0039] The cloud-based cooling control service receives pre-processed data from several key points under each pipe cooling cycle control zone in real time. It then inputs the pre-processed data into the cooling cycle load demand risk function to determine the risk type of the real-time pre-processed data. Based on the risk type, it calculates the cooling cycle load demand risk coefficient of each key node and sums the cooling cycle load demand risk coefficients of each key node to obtain the cooling cycle load demand risk coefficient of each pipe cooling cycle control zone.

[0040] The method for setting a unique cold cycle load demand risk function for each pipe cold cycle control zone includes: extracting key factors that reflect each risk type from preprocessed data; the method for obtaining the cold cycle load demand risk function is as follows: obtaining historical information of preprocessed data corresponding to the risk type; selecting safe and risky preprocessed data from the historical information of preprocessed data based on the time and environmental characteristics of the preprocessed data; assigning values ​​to the selected preprocessed data according to their size; assigning continuous values ​​to continuous preprocessed data and type-specific values ​​to discrete preprocessed data; and obtaining the cold cycle load demand risk function for continuous and discrete preprocessed data respectively through radial basis function interpolation and decision tree neural network algorithms.

[0041] Furthermore, the risk threshold of the preprocessed data obtained from the cold cycle load demand risk function includes:

[0042] The cold cycle load demand risk function includes a continuous risk function and a discontinuous risk function.

[0043] If the preprocessed data is continuous, the risk threshold is determined by calculating the intersection of safe preprocessed data and risky preprocessed data using the continuous risk function. If the preprocessed data is discrete, the risk threshold is determined by calculating the non-continuous risk function, i.e., the input and output of the decision tree neural network model. The determination of the risk threshold by the input and output of the decision tree neural network model refers to: based on the preprocessed data with multiple adjacent input values ​​in the decision tree neural network model, judging the type of preprocessed data by the output, and obtaining the risk threshold by combining the preprocessed data with adjacent input values ​​and the output.

[0044] The step of incorporating the preprocessed data into the cold cycle load demand risk function includes:

[0045] The domain range of the cold cycle load demand risk function is determined by analyzing the physical characteristics and historical data of the preprocessed data. If the preprocessed data is within the domain range of the cold cycle load demand risk function, the preprocessed data is substituted into the cold cycle load demand risk function, and the risk type of the preprocessed data is determined based on the result.

[0046] Furthermore, the step of using the second algorithm combined with the risk coefficients of the load demands of A cooling cycles to control the cooling cycles of B cooling cycles, and generating cooling cycle control decisions, includes:

[0047] Calculate the comprehensive adjustment coefficient for pipe cooling circulation control zone A, and calculate the power values ​​for pipe cooling water flow regulation and pipe cooling water temperature regulation for pipe cooling circulation control zone A in low-risk and high-risk types respectively. The calculation method is as follows:

[0048]

[0049] Among them, TidAjust i b1 represents the overall adjustment coefficient for the i-th pipe cooling cycle control zone. i b2 represents the first risk value of the i-th pipe cooling cycle control zone. i b3 represents the second risk value of the i-th pipe cooling cycle control zone. i This represents the third risk value of the i-th pipe cooling cycle control zone, where a1, a2, and a3 correspond to b1 respectively. i b2 i b3 i Weighting coefficients; Prun1 i This indicates the power value required to adjust the cooling water flow rate of the corresponding cooling circulation pipe when the i-th pipe cooling circulation control zone is in a low-risk type, controlling the output control power of the cooling water flow rate; Prun2 iThis indicates the power value required to adjust the cooling water flow rate of the corresponding cooling circulation pipe when the i-th pipe cooling circulation control zone is in a high-risk category; Prun3 represents the output control power for controlling the cooling water flow rate. i This indicates the power value required to adjust the cooling water temperature of the corresponding cooling circulation pipe when the i-th pipe cooling circulation control zone is in a low-risk category; Prun4 represents the output control power for controlling the cooling water temperature. i This indicates the power value required to adjust the cooling water temperature of the corresponding cooling circulation pipe when the i-th pipe cooling circulation control zone is in a high-risk type, and the output control power to control the cooling water temperature; P1 represents the initial power value for adjusting the cooling water flow rate; P2 represents the initial power value for adjusting the cooling water temperature.

[0050] Prun1 i Prun2 i Prun3 i Prun4 i All of these represent the set values ​​of the cooling circulation control decision output for the cooling circulation pipeline corresponding to the low-risk and high-risk types in the i-th pipe cooling circulation control zone.

[0051] Furthermore, the step of adjusting the cooling water flow rate and cooling water temperature of each pipe using a PID control algorithm based on the set value output by the pipe cooling circulation control decision of the B cooling circulation pipes includes:

[0052] Based on the identified cooling circulation pipelines that need adjustment, an input matrix sequence consisting of the set values ​​of the cooling circulation pipelines that need adjustment is established. The cooling circulation pipelines that need adjustment are determined according to whether the corresponding cooling circulation control zone is in a low-risk or high-risk type.

[0053] Based on the integral PID control algorithm, the control command for the output signal is obtained by reading the data of the input matrix sequence composed of setpoints;

[0054] According to the control instructions of the output signal, the control execution component adjusts the pipeline cooling water flow rate and pipeline cooling water temperature of the i-th pipeline cooling circulation control zone to realize intelligent dynamic feedback control of pipeline cooling circulation for lock construction.

[0055] The dynamic feedback control includes:

[0056] A visualization and interactive platform was established to display the change curves of multi-dimensional environmental data for all pipe cooling circulation control zones during the construction of the ship lock, as well as the parameter change curves of the intelligent pipe cooling circulation dynamic feedback control.

[0057] Intelligent tube cooling circulation control system based on ship lock construction. The system is applied to any of the methods of the intelligent tube cooling circulation control method based on ship lock construction, and includes:

[0058] A regional division module, which is used to divide the ship lock construction structure into pier areas and lock wall areas, conduct key point division on the pier areas and lock wall areas, generate A tube cooling circulation control zones, and preset A edge key control nodes for the A tube cooling circulation control zones. Among them, each zone represents an independent structural type in the pier and the lock wall, and each zone contains several key points, and the key points are used for data collection;

[0059] A covering module, which is used to cover the A tube cooling circulation control zones through B cooling circulation pipelines. Each cooling circulation pipeline is equipped with an electromagnetic regulating valve and a turbine flowmeter, where B ≤ A;

[0060] A preprocessing module, which is used for each edge key control node to receive multi-dimensional data of each tube cooling circulation control zone, preprocess the multi-dimensional data to generate preprocessed data, and transmit the preprocessed data to the cloud cooling control service; the multi-dimensional data includes key node basic data, inspection data on the outer surface of the cooling pipes at key nodes, scaling data inside the cooling pipes at key nodes, ambient temperature at key nodes, ambient humidity at key nodes, air velocity at key nodes, and the state of tube cooling equipment at key nodes; the preprocessed data includes first preprocessed data and second preprocessed data, where the first preprocessed data does not include inspection data on the outer surface of the cooling pipes at key nodes, and the second preprocessed data at least includes inspection data on the outer surface of the cooling pipes at key nodes;

[0061] A cloud computing module, which is used for the cloud cooling control service to receive the A preprocessed data of the A edge key control nodes, and based on the first deep learning algorithm, predict the A preprocessed data to generate A cold circulation load demand risk coefficients; the cold circulation load demand risk coefficients are set to four values according to the risk type: the first risk value b1, the second risk value b2, and the third risk value b3, where b1 < b2 < b3; the risk type is set for the tube cooling circulation control zone and several key points under the tube cooling circulation control zone. The risk type includes no risk type, low risk type, high risk type, and equipment abnormal type. Among them, the risk type of each tube cooling circulation control zone is calculated from the sum of the risk types of several key points controlled under each tube cooling circulation control zone;

[0062] A decision generation module. If it is judged that there is a risk in the cold circulation load demand risk coefficient of at least 1 tube cooling circulation control zone among the A cold circulation load demand risk coefficients, based on the second algorithm and combined with the A cold circulation load demand risk coefficients, conduct tube cooling circulation control on the B cooling circulation pipelines to generate a tube cooling circulation control decision;

[0063] The execution module, based on the set values ​​output by the cooling cycle control decision of the B cooling cycle pipelines, uses a PID control algorithm to adjust the cooling water flow rate and cooling water temperature of each pipeline according to the set values.

[0064] Compared with the prior art, the present invention achieves the following beneficial effects:

[0065] This invention provides a method for dividing the construction structure of a ship lock into A pipe cooling circulation control zones; covering these A zones with B cooling circulation pipelines; receiving multi-dimensional data from each pipe cooling circulation control zone at key edge control nodes and transmitting it to a cloud-based cooling control service; after receiving the data, the cloud-based cooling control service generates A cooling circulation load demand risk coefficients based on a first deep learning algorithm; if at least one of the A cooling circulation load demand risk coefficients is deemed to be at risk, then the B cooling circulation pipelines are controlled using a second algorithm, employing a PID control algorithm to adjust the cooling water flow rate and temperature; this invention uses deep learning algorithms for prediction and monitoring, introduces drones for inspecting the outer surface of the cooling pipes, and combines this with the analysis of internal and external data from multiple areas of the ship lock construction to achieve precise pipe cooling circulation control and adjustment, thereby improving the quality of ship lock construction and reducing the risk of concrete hardening and cracking. Attached Figure Description

[0066] Figure 1 This is a flowchart illustrating the intelligent pipe cooling circulation control method based on lock construction provided in an embodiment of the present invention.

[0067] Figure 2 This is an architecture diagram of an intelligent pipe cooling circulation control system based on lock construction provided in an embodiment of the present invention. Detailed Implementation

[0068] The present invention will now be described in detail with reference to the accompanying drawings.

[0069] Example 1

[0070] This invention provides an intelligent pipe cooling circulation control method based on lock construction, such as... Figure 1 The method includes:

[0071] S1. Divide the lock construction structure into a pier area and a wall area. Divide the pier area and wall area into key points to generate A pipe cold circulation control zones. Preset A edge key control nodes in the A pipe cold circulation control zones. Each zone represents an independent structural type in the pier and wall. Each zone contains several key points, which are used for data acquisition.

[0072] S2. Cover the A tube cooling cycle control zones through B cooling cycle pipelines. Each cooling cycle pipeline is equipped with an electromagnetic regulating valve and a turbine flowmeter, where B ≤ A;

[0073] S3. Each edge key control node receives the multi-dimensional data of each tube cooling cycle control zone, preprocesses the multi-dimensional data to generate preprocessed data, and transmits the preprocessed data to the cloud cooling control service; the multi-dimensional data includes key node basic data, key node outer surface inspection data of the cooling tube, key node internal scaling data of the cooling tube, key node ambient temperature, key node ambient humidity, key node air velocity, key node tube cooling equipment status; the preprocessed data includes first preprocessed data and second preprocessed data, where the first preprocessed data does not include the key node outer surface inspection data of the cooling tube, and the second preprocessed data at least includes the key node outer surface inspection data of the cooling tube;

[0074] S4. After receiving the A preprocessed data from the A edge key control nodes, the cloud cooling control service uses the first deep learning algorithm to predict the A preprocessed data to generate A cold cycle load demand risk coefficients; the cold cycle load demand risk coefficients are set to three values according to the risk type: the first risk value b1, the second risk value b2, the third risk value b3, where b1 < b2 < b3; the risk type is set for the tube cooling cycle control zone and several key points under the tube cooling cycle control zone, and the risk type includes no risk type, low risk type, high risk type and equipment abnormal type, where the risk type of each tube cooling cycle control zone is calculated from the sum of the risk types of several key points controlled under each tube cooling cycle control zone;

[0075] S5. If it is judged that there is a risk in the cold cycle load demand risk coefficient of at least one tube cooling cycle control zone among the A cold cycle load demand risk coefficients, based on the second algorithm and combined with the A cold cycle load demand risk coefficients, perform tube cooling cycle control on the B cooling cycle pipelines to generate a tube cooling cycle control decision;

[0076] S6. According to the set value output by the tube cooling cycle control decision of the B cooling cycle pipelines, use the PID control algorithm to adjust the cooling water flow and cooling water temperature of each pipeline for the set value.

[0077] Specifically, the lock construction structure is divided into A pipe cooling circulation control zones; B cooling circulation pipelines cover A pipe cooling circulation control zones; each edge key control node receives multi-dimensional data from each pipe cooling circulation control zone and transmits it to the cloud cooling control service; after receiving the data, the cloud cooling control service generates A cooling circulation load demand risk coefficients based on a first deep learning algorithm; if it is determined that at least one of the A cooling circulation load demand risk coefficients is at risk, the B cooling circulation pipelines are controlled based on a second algorithm, using a PID control algorithm to adjust the pipeline cooling water flow rate and pipeline cooling water temperature; this invention uses deep learning algorithm prediction and monitoring, introduces drones to inspect the outer surface of the cooling pipes, and combines the analysis of internal and external data from multiple areas of the lock construction to achieve precise pipe cooling circulation control and adjustment, improve the lock construction quality, and reduce the risk of concrete hardening and cracking.

[0078] In the above embodiments, specifically, the step of dividing the lock construction structure into a pier area and a wall area, and dividing the pier area and wall area into key points to generate A pipe cooling circulation control zones, specifically includes:

[0079] Based on GPS technology, the initial flight path of the drone is determined, and a camera is installed on the drone.

[0080] The drone flew along its initial flight path and collected basic information and images of the lock construction structure using its onboard camera.

[0081] Based on basic information and image sets, combined with image recognition algorithms, the lock construction structure is divided into two regions: the lock pier region and the lock wall region.

[0082] Key points in two areas are collected to obtain coordinate and image information of several key points. Clustering analysis is performed on the coordinate information of several key points to generate a first set of key points. Using a cosine similarity algorithm, the feature vectors of image information of historical lock construction structures in the database and the image information feature vectors corresponding to the first set of key points are calculated to generate a second set of key points. The total number of the second set of key points is represented by A pipe cooling cycle control zones. The total number of the sets is greater than 1. The image information of the historical lock construction structures represents key point marker information, which is used for the division of pipe cooling cycle control zones.

[0083] The coordinates corresponding to the image set, the basic information, and the second key point set are input into the navigation path planning model, and the navigation path information of the UAV is output. The navigation path information is used for the UAV to collect data on the A pipe cooling cycle control partitions. The navigation path planning model is built by performing BP neural network training based on the navigation path decision tree. The navigation path decision tree is built by mapping and associating partition identification nodes and partition key points configured through decision processing of historical sample data.

[0084] It should be noted that the deployment of drones for image acquisition of the lock pier and lock wall areas of the lock construction structure involves using drone cameras to acquire omnidirectional, multi-angle images of these areas. The order of the acquired images is adjusted based on changes in the navigation path, generating image sets corresponding to the two areas. Preferably, the image acquisition distance can be set and adjusted to acquire both global and zone-specific images separately, ensuring the accuracy of image acquisition and the comprehensiveness of image coverage.

[0085] Based on the image sets corresponding to the two regions and their three-dimensional spatial locations, the three-dimensional coordinate axes are determined, and a three-dimensional coordinate system is constructed. Structural identification is performed on the gate pier and gate wall regions of the target lock construction structure. Multiple positioning points characterizing the structural conditions of the gate pier and gate wall regions are identified and extracted, using algorithms such as wavelet transform edge detection, YOLOv7 positioning, and RetinaNet positioning. The specific density of the positioning point distribution depends on the structural complexity of different locations. Based on the three-dimensional coordinate system, the determined positioning points are mapped to their corresponding positions, and the coordinates of key points in the three-dimensional coordinate system are determined, forming the key point distribution coordinate set. Based on the key point distribution coordinate set, the main framework of the target lock construction structure can be determined, thereby determining the gate pier and gate wall regions, as well as the coordinate and image information of several key points within these two regions.

[0086] Specifically, a navigation path planning model is constructed to plan the navigation path for data collection by the UAV. An exemplary, feasible modeling approach is as follows: A big data survey and statistical analysis is conducted to determine project data for multiple lock construction projects, including a set of sample images and basic sample information. Based on this basic information, positioning point analysis is performed to determine the distribution coordinates of key points in the samples, and control parameters corresponding to each sample are extracted. Since the project data represents already constructed projects, the aforementioned parameter data can be directly obtained. The set of sample images, the basic sample information, and the distribution coordinates of key points in the samples are used as identification information to determine key points in different zones. The corresponding sample control parameters are used as zone identification nodes. The zone identification nodes and the zone key points are mapped and connected to generate a navigation path decision tree. Based on the navigation path decision tree, the navigation path planning model is generated by training a BP neural network. The coordinates corresponding to the image set, the basic information, and the second key point set are input into the navigation path planning model, and the navigation path information of the UAV is output. The navigation path information is used for subsequent data collection by the UAV on the A-pipe cold circulation control zone. By adopting the navigation path planning model, the speed of external factor analysis of the lock construction structure can be improved. Combined with the data collected by internal sensors, the accuracy of the analysis of the cold circulation load demand risk of the entire lock construction structure can be improved. This is also the highlight that distinguishes this invention from the prior art.

[0087] In the above embodiments, specifically, each edge key control node receives multidimensional data from each pipe cooling cycle control zone, and preprocesses the multidimensional data to generate preprocessed data, including:

[0088] The ambient temperature of each pipe cold circulation control zone is collected by temperature sensor, the ambient humidity of each pipe cold circulation control zone is collected by humidity sensor, the outer surface inspection data of the cooling pipe of each pipe cold circulation control zone is collected by drone, the internal scaling data of the cooling pipe of each pipe cold circulation control zone is collected by pipe wall vibration sensor, and the air velocity of each pipe cold circulation control zone is collected by air velocity sensor.

[0089] The multidimensional data is collected at fixed intervals, and the collected data is transmitted to each critical edge control node.

[0090] In the above embodiments, specifically, each edge key control node receives multi-dimensional data from each pipe cooling cycle control zone, preprocesses the multi-dimensional data, and generates preprocessed data, further including:

[0091] The sensor-acquired data is preprocessed sequentially using spatiotemporal alignment, wavelet filtering, feature enhancement algorithm, and confidence weighting to generate the first preprocessed data.

[0092] The image data acquired by the UAV is preprocessed by sequentially using image enhancement algorithm, shift image segmentation algorithm, and keyframe extraction algorithm to generate the second preprocessed data.

[0093] In the above embodiments, specifically, each edge key control node receives multi-dimensional data from each pipe cooling cycle control zone, preprocesses the multi-dimensional data, and generates preprocessed data, further including:

[0094] Confidence modeling is performed on the confidence scores of the first and second preprocessed data, assuming that both the first and second preprocessed data follow a normal distribution:

[0095]

[0096] Where, x j θ represents the output values ​​of the first and second preprocessed data. j This represents the mean. The noise mean square error, This indicates that the measured values ​​follow a mean of θ. j The mean squared error is The normal distribution;

[0097] The optimized posterior distribution is inferred using the Markov chain Monte Carlo method, and the confidence weights of each first and second preprocessed data are calculated.

[0098] A preset threshold is set, and the confidence weights of the first and second preprocessed data are compared with the preset threshold. The first and second preprocessed data that meet the threshold requirements are sent to the edge key control node.

[0099] It should be noted that the optimization posterior distribution is inferred using the Markov Chain Monte Carlo (MCMC) method, and the confidence weights for each of the first and second preprocessed data points are calculated as follows:

[0100] Step 1: Define the model and posterior distribution

[0101] First, we define a probabilistic model that describes the relationship between data and parameters. Suppose we have a parameter θ that describes the data generation process. Given data D, we infer the posterior distribution P(θ|D) of parameter θ. According to Bayes' theorem, the posterior distribution can be expressed as:

[0102] P(θ∣D)∝P(D∣θ)P(θ),

[0103] Where P(D|θ) is the likelihood function and P(θ) is the prior distribution;

[0104] Step 2: Select a suitable MCMC algorithm

[0105] Choose a suitable MCMC algorithm to generate samples of parameter θ. Common MCMC algorithms include Metropolis-Hastings algorithm, Gibbs sampling and Hamiltonian Monte Carlo (HMC), etc. In this embodiment, Metropolis-Hastings algorithm is preferred. These algorithms construct a Markov chain such that the stationary distribution of the chain is equal to the posterior distribution of the target.

[0106] Step 3: Calculate the confidence weights

[0107] After the MCMC algorithm converges, a series of parameter samples are obtained. These samples can be used to estimate the expected value, variance and other statistics of the posterior distribution. For each first preprocessed data and second preprocessed data, their confidence weights are calculated. The confidence weights are the density values ​​of the posterior distribution.

[0108] In the above embodiments, specifically, the first deep learning algorithm includes:

[0109] A unique cold cycle load demand risk function is set for each pipe cold cycle control zone. The cold cycle load demand risk function is used to analyze the preprocessed data of several key points under each pipe cold cycle control zone. The preprocessed data of several key points must be input into the cold cycle load demand risk function for analysis. The analysis results are used to determine the risk type of the key points. The key points are classified into risk types through the cold cycle load demand risk function. The risk threshold of the preprocessed data of the key points is calculated through the cold cycle load demand risk function.

[0110] The cloud-based cooling control service receives pre-processed data from several key points under each pipe cooling cycle control zone in real time. It then inputs the pre-processed data into the cooling cycle load demand risk function to determine the risk type of the real-time pre-processed data. Based on the risk type, it calculates the cooling cycle load demand risk coefficient of each key node and sums the cooling cycle load demand risk coefficients of each key node to obtain the cooling cycle load demand risk coefficient of each pipe cooling cycle control zone.

[0111] The method for setting a unique cold cycle load demand risk function for each pipe cold cycle control zone includes: extracting key factors that reflect each risk type from preprocessed data; the method for obtaining the cold cycle load demand risk function is as follows: obtaining historical information of preprocessed data corresponding to the risk type; selecting safe and risky preprocessed data from the historical information of preprocessed data based on the time and environmental characteristics of the preprocessed data; assigning values ​​to the selected preprocessed data according to their size; assigning continuous values ​​to continuous preprocessed data and type-specific values ​​to discrete preprocessed data; and obtaining the cold cycle load demand risk function for continuous and discrete preprocessed data respectively through radial basis function interpolation and decision tree neural network algorithms.

[0112] In the above embodiments, specifically, the risk threshold of the preprocessed data of key points obtained through the cold cycle load demand risk function includes:

[0113] The cold cycle load demand risk function includes a continuous risk function and a discontinuous risk function.

[0114] If the preprocessed data is continuous, the risk threshold is determined by calculating the intersection of safe preprocessed data and risky preprocessed data using the continuous risk function. If the preprocessed data is discrete, the risk threshold is determined by calculating the non-continuous risk function, i.e., the input and output of the decision tree neural network model. The determination of the risk threshold by the input and output of the decision tree neural network model refers to: based on the preprocessed data with multiple adjacent input values ​​in the decision tree neural network model, judging the type of preprocessed data by the output, and obtaining the risk threshold by combining the preprocessed data with adjacent input values ​​and the output.

[0115] The step of incorporating the preprocessed data into the cold cycle load demand risk function includes:

[0116] The domain range of the cold cycle load demand risk function is determined by analyzing the physical characteristics and historical data of the preprocessed data. If the preprocessed data is within the domain range of the cold cycle load demand risk function, the preprocessed data is substituted into the cold cycle load demand risk function, and the risk type of the preprocessed data is determined based on the result.

[0117] In the above embodiments, specifically, the step of controlling the cooling cycle of B cooling cycle pipelines based on the second algorithm combined with the risk coefficient of A cooling cycle load demand, and generating the cooling cycle control decision, includes:

[0118] Calculate the comprehensive adjustment coefficient for pipe cooling circulation control zone A, and calculate the power values ​​for pipe cooling water flow regulation and pipe cooling water temperature regulation for pipe cooling circulation control zone A in low-risk and high-risk types respectively. The calculation method is as follows:

[0119]

[0120] Among them, TidAjust i b1 represents the overall adjustment coefficient for the i-th pipe cooling cycle control zone. i b2 represents the first risk value of the i-th pipe cooling cycle control zone. i b3 represents the second risk value of the i-th pipe cooling cycle control zone. i This represents the third risk value of the i-th pipe cooling cycle control zone, where a1, a2, and a3 correspond to b1 respectively. i b2 i b3 i Weighting coefficients; Prun1 i This indicates the power value required to adjust the cooling water flow rate of the corresponding cooling circulation pipe when the i-th pipe cooling circulation control zone is in a low-risk type, controlling the output control power of the cooling water flow rate; Prun2 i This indicates the power value required to adjust the cooling water flow rate of the corresponding cooling circulation pipe when the i-th pipe cooling circulation control zone is in a high-risk category; Prun3 represents the output control power for controlling the cooling water flow rate. i This indicates the power value required to adjust the cooling water temperature of the corresponding cooling circulation pipe when the i-th pipe cooling circulation control zone is in a low-risk category; Prun4 represents the output control power for controlling the cooling water temperature. i This indicates the power value required to adjust the cooling water temperature of the corresponding cooling circulation pipe when the i-th pipe cooling circulation control zone is in a high-risk type, and the output control power to control the cooling water temperature; P1 represents the initial power value for adjusting the cooling water flow rate; P2 represents the initial power value for adjusting the cooling water temperature.

[0121] Prun1 i Prun2 i Prun3 i Prun4 i All of these represent the set values ​​of the cooling circulation control decision output for the cooling circulation pipeline corresponding to the low-risk and high-risk types in the i-th pipe cooling circulation control zone.

[0122] It should be noted that the power values ​​for regulating the cooling water flow rate and temperature are only adjusted if the pipe cooling circulation control zone is in a low-risk or high-risk condition. First, calculate the comprehensive adjustment coefficient for each pipe cooling circulation control zone. Then, based on this coefficient, calculate the corresponding power values ​​for regulating the cooling water flow rate and temperature for both low-risk and high-risk conditions. If a pipe cooling circulation control zone is in a low-risk condition, adjust using the power values ​​corresponding to that condition. If a pipe cooling circulation control zone is in a high-risk condition, adjust using the power values ​​corresponding to that condition. This allows for precise adjustment of the power values ​​for regulating the cooling water flow rate and temperature. The power value of the section, based on the second algorithm combined with the risk coefficient of the A-type cold cycle load demand, is used to control the cooling cycle of B-type cooling cycle pipes. The technical feature of generating the cooling cycle control decision is used to solve the problem that when a certain zone is found to have adjustment risk, the cooling cycle control parameters cannot be dynamically and accurately adjusted according to the risk level. This technical feature is one of the technical highlights of the present invention. It should be further explained that the first risk value b1, the second risk value b2, and the third risk value b3 are all derived from the first deep learning algorithm based on the actual parameters of each cooling cycle control zone. If a certain cooling cycle control zone is found to have adjustment risk, the power value of the pipe cooling water flow rate adjustment and the power value of the pipe cooling water temperature adjustment are precisely adjusted in real time, which improves the accuracy of adjusting the cooling cycle load demand of each cooling cycle control zone. This is also a technical highlight of the present invention.

[0123] In the above embodiments, specifically, the step of adjusting the cooling water flow rate and cooling water temperature of each pipe using a PID control algorithm based on the set value output by the pipe cooling circulation control decision of the B cooling circulation pipes includes:

[0124] Based on the identified cooling circulation pipelines that need adjustment, an input matrix sequence consisting of the set values ​​of the cooling circulation pipelines that need adjustment is established. The cooling circulation pipelines that need adjustment are determined according to whether the corresponding cooling circulation control zone is in a low-risk or high-risk type.

[0125] Based on the integral PID control algorithm, the control command for the output signal is obtained by reading the data of the input matrix sequence composed of setpoints;

[0126] According to the control instructions of the output signal, the control execution component adjusts the pipeline cooling water flow rate and pipeline cooling water temperature of the i-th pipeline cooling circulation control zone to realize intelligent dynamic feedback control of pipeline cooling circulation for lock construction.

[0127] The dynamic feedback control includes:

[0128] A visualization and interactive platform was established to display the change curves of multi-dimensional environmental data for all pipe cooling circulation control zones during the construction of the ship lock, as well as the parameter change curves of the intelligent pipe cooling circulation dynamic feedback control.

[0129] Example 2

[0130] A smart pipe cooling circulation control system based on lock construction, wherein the system is applied to any one of the smart pipe cooling circulation control methods based on lock construction, including:

[0131] The region division module is used to divide the lock construction structure into the gate pier region and the gate wall region, divide the gate pier region and the gate wall region into key points, generate A pipe cold circulation control partitions, and pre-set A edge key control nodes in the A pipe cold circulation control partitions. Each partition represents an independent structural type in the gate pier and the gate wall, and each partition contains several key points, which are used for data acquisition.

[0132] A covering module is used to cover the A pipe cooling circulation control zone through B cooling circulation pipes. Each cooling circulation pipe is equipped with an electromagnetic regulating valve and a turbine flow meter, wherein B≤A;

[0133] The preprocessing module is used by each edge key control node to receive multidimensional data from each pipe cooling circulation control zone, preprocess the multidimensional data to generate preprocessed data, and transmit the preprocessed data to the cloud cooling control service. The multidimensional data includes key node basic data, key node cooling pipe outer surface inspection data, key node cooling pipe internal scaling data, key node ambient temperature, key node ambient humidity, key node airflow rate, and key node pipe cooling equipment status. The preprocessed data includes first preprocessed data and second preprocessed data, wherein the first preprocessed data does not include key node cooling pipe outer surface inspection data, and the second preprocessed data includes at least key node cooling pipe outer surface inspection data.

[0134] A cloud computing module is configured to receive the A preprocessed data from the A edge critical control nodes for cloud cooling control services, and then generate A cold cycle load demand risk coefficients by predicting the A preprocessed data based on a first deep learning algorithm. The cold cycle load demand risk coefficients are set to four values according to the risk type: a first risk value b1, a second risk value b2, and a third risk value b3, where b1 < b2 < b3. The risk types are set for the tube cooling cycle control zones and several key points under the tube cooling cycle control zones. The risk types include a risk-free type, a low-risk type, a high-risk type, and an equipment anomaly type. Among them, the risk type of each tube cooling cycle control zone is calculated from the sum of the risk types of several key points controlled under each tube cooling cycle control zone.

[0135] A decision-making generation module, if it is determined that there is a risk in the cold cycle load demand risk coefficient of at least one tube cooling cycle control zone among the A cold cycle load demand risk coefficients, based on a second algorithm and combined with the A cold cycle load demand risk coefficients, performs tube cooling cycle control on the B cooling cycle pipelines to generate a tube cooling cycle control decision.

[0136] An execution module adjusts the cooling water flow rate and the cooling water temperature of each pipeline according to the set value output by the tube cooling cycle control decision of the B cooling cycle pipelines, using a PID control algorithm for the set value.

[0137] It should be understood that the above embodiments are one or more embodiments of the present invention. Based on the present invention, there are many other embodiments and their variations. When ordinary technicians in this industry do not make pioneering innovations, the variations and modifications made through the present invention fall within the protection scope of the present invention.

Claims

1. A smart pipe cooling circulation control method based on lock construction, characterized in that, The method includes the following steps: Divide the lock construction structure into pier areas and lock wall areas, divide key points in the pier areas and lock wall areas, generate A chilled water circulation control zones, and preset A edge key control nodes for the A chilled water circulation control zones. Each zone represents an independent structural type in the pier and the lock wall, and each zone contains several key points for data collection; Cover the A chilled water circulation control zones with B cooling circulation pipelines, and each cooling circulation pipeline is equipped with an electromagnetic regulating valve and a turbine flowmeter, where B ≤ A; Each edge key control node receives multi-dimensional data of each chilled water circulation control zone, preprocesses the multi-dimensional data to generate preprocessed data, and transmits the preprocessed data to the cloud cooling control service; the multi-dimensional data includes basic data of key nodes, inspection data of the outer surface of the cooling pipes at key nodes, scaling data inside the cooling pipes at key nodes, ambient temperature at key nodes, ambient humidity at key nodes, air velocity at key nodes, and the state of the chilled water equipment at key nodes; the preprocessed data includes first preprocessed data and second preprocessed data, where the first preprocessed data does not include the inspection data of the outer surface of the cooling pipes at key nodes, and the second preprocessed data at least includes the inspection data of the outer surface of the cooling pipes at key nodes; After the cloud cooling control service receives the A preprocessed data from the A edge key control nodes, it uses the first deep learning algorithm to predict the A preprocessed data to generate A chilled water circulation load demand risk coefficients; the chilled water circulation load demand risk coefficients are set to three values according to the risk type: the first risk value b1, the second risk value b2, and the third risk value b3, where b1 < b2 < b3; the risk type is set for the chilled water circulation control zone and several key points under the chilled water circulation control zone. The risk type includes a risk-free type, a low-risk type, a high-risk type, and an equipment anomaly type. Among them, the risk type of each chilled water circulation control zone is calculated from the sum of the risk types of several key points controlled under each chilled water circulation control zone; If it is determined that there is a risk in the chilled water circulation load demand risk coefficient of at least 1 chilled water circulation control zone among the A chilled water circulation load demand risk coefficients, perform chilled water circulation control on the B cooling circulation pipelines based on the second algorithm combined with the A chilled water circulation load demand risk coefficients to generate a chilled water circulation control decision; According to the set values output by the chilled water circulation control decision of the B cooling circulation pipelines, use the PID control algorithm to adjust the cooling water flow rate and the cooling water temperature of each pipeline for the set values.

2. The intelligent pipe cooling circulation control method based on lock construction according to claim 1, characterized in that, The step of dividing the lock construction structure into pier areas and lock wall areas, dividing key points in the pier areas and lock wall areas, and generating A chilled water circulation control zones specifically includes: Based on GPS technology, determine the initial flight path of the unmanned aerial vehicle (UAV), and configure a camera on the UAV; The UAV flies according to the initial flight path and collects the basic information and image set of the lock construction structure through the carried camera; Based on basic information and image sets, combined with image recognition algorithms, the lock construction structure is divided into two regions: the lock pier region and the lock wall region. Key points in two areas are collected to obtain coordinate and image information of several key points. Clustering analysis is performed on the coordinate information of several key points to generate a first set of key points. Using a cosine similarity algorithm, the feature vectors of image information of historical lock construction structures in the database and the image information feature vectors corresponding to the first set of key points are calculated to generate a second set of key points. The total number of the second set of key points is represented by A pipe cooling cycle control zones. The total number of the sets is greater than 1. The image information of the historical lock construction structures represents key point marker information, which is used for the division of pipe cooling cycle control zones. The coordinates corresponding to the image set, the basic information, and the second key point set are input into the navigation path planning model, and the navigation path information of the UAV is output. The navigation path information is used for the UAV to collect data on the A pipe cooling cycle control partitions. The navigation path planning model is built by performing BP neural network training based on the navigation path decision tree. The navigation path decision tree is built by mapping and associating partition identification nodes and partition key points configured through decision processing of historical sample data.

3. The intelligent pipe cooling circulation control method based on lock construction according to claim 1, characterized in that, Each critical edge control node receives multidimensional data from each pipe cooling cycle control zone, and preprocesses the multidimensional data to generate preprocessed data, including: The ambient temperature of each pipe cold circulation control zone is collected by temperature sensor, the ambient humidity of each pipe cold circulation control zone is collected by humidity sensor, the outer surface inspection data of the cooling pipe of each pipe cold circulation control zone is collected by drone, the internal scaling data of the cooling pipe of each pipe cold circulation control zone is collected by pipe wall vibration sensor, and the air velocity of each pipe cold circulation control zone is collected by air velocity sensor. The multidimensional data is collected at fixed intervals, and the collected data is transmitted to each critical edge control node.

4. The intelligent pipe cooling circulation control method based on lock construction according to claim 1, characterized in that, Each critical edge control node receives multidimensional data from each pipe cooling cycle control zone, preprocesses the multidimensional data, and generates preprocessed data, which also includes: The sensor-acquired data is preprocessed sequentially using spatiotemporal alignment, wavelet filtering, feature enhancement algorithm, and confidence weighting to generate the first preprocessed data. The image data acquired by the UAV is preprocessed by sequentially using image enhancement algorithm, shift image segmentation algorithm, and keyframe extraction algorithm to generate the second preprocessed data.

5. The intelligent pipe cooling circulation control method based on lock construction according to claim 4, characterized in that, Each critical edge control node receives multidimensional data from each pipe cooling cycle control zone, preprocesses the multidimensional data, and generates preprocessed data, which also includes: Confidence modeling is performed on the confidence scores of the first and second preprocessed data, assuming that both the first and second preprocessed data follow a normal distribution: Where, x j θ represents the output values ​​of the first and second preprocessed data. j This represents the mean. The noise mean square error, This indicates that the measured values ​​follow a mean of θ. j The mean squared error is The normal distribution; The optimized posterior distribution is inferred using the Markov chain Monte Carlo method, and the confidence weights of each first and second preprocessed data are calculated. The confidence weights of the first and second preprocessed data are compared with preset thresholds, and the first and second preprocessed data that meet the threshold requirements are sent to the edge critical control node.

6. The intelligent pipe cooling circulation control method based on lock construction according to claim 1, characterized in that, The first deep learning algorithm includes: A unique cold cycle load demand risk function is set for each pipe cold cycle control zone. The cold cycle load demand risk function is used to analyze the preprocessed data of several key points under each pipe cold cycle control zone. The preprocessed data of several key points must be input into the cold cycle load demand risk function for analysis. The analysis results are used to determine the risk type of the key points. The key points are classified into risk types through the cold cycle load demand risk function. The risk threshold of the preprocessed data of the key points is calculated through the cold cycle load demand risk function. The cloud-based cooling control service receives pre-processed data from several key points under each pipe cooling cycle control zone in real time. It then inputs the pre-processed data into the cooling cycle load demand risk function to determine the risk type of the real-time pre-processed data. Based on the risk type, it calculates the cooling cycle load demand risk coefficient of each key node and sums the cooling cycle load demand risk coefficients of each key node to obtain the cooling cycle load demand risk coefficient of each pipe cooling cycle control zone. The method for setting a unique cold cycle load demand risk function for each pipe cold cycle control zone includes: extracting key factors that reflect each risk type from preprocessed data; the method for obtaining the cold cycle load demand risk function is as follows: obtaining historical information of preprocessed data corresponding to the risk type; selecting safe and risky preprocessed data from the historical information of preprocessed data based on the time characteristics, environmental characteristics, and construction characteristics of the preprocessed data; assigning values ​​to the selected preprocessed data according to their size; assigning continuous values ​​to continuous preprocessed data and type-specific values ​​to discrete preprocessed data; and obtaining the cold cycle load demand risk function for continuous and discrete preprocessed data using radial basis function interpolation and decision tree neural network algorithms, respectively.

7. The intelligent pipe cooling circulation control method based on lock construction according to claim 6, characterized in that, The risk thresholds for the preprocessed data of key points obtained through the cold cycle load demand risk function include: The cold cycle load demand risk function includes a continuous risk function and a discontinuous risk function. If the preprocessed data is continuous, the risk threshold is determined by calculating the intersection of safe preprocessed data and risky preprocessed data using the continuous risk function. If the preprocessed data is discrete, the risk threshold is determined by calculating the non-continuous risk function, i.e., the input and output of the decision tree neural network model. The determination of the risk threshold by the input and output of the decision tree neural network model refers to: based on the preprocessed data with multiple adjacent input values ​​in the decision tree neural network model, judging the type of preprocessed data by the output, and obtaining the risk threshold by combining the preprocessed data with adjacent input values ​​and the output. The step of incorporating the preprocessed data into the cold cycle load demand risk function includes: The domain range of the cold cycle load demand risk function is determined by analyzing the physical characteristics and historical data of the preprocessed data. If the preprocessed data is within the domain range of the cold cycle load demand risk function, the preprocessed data is substituted into the cold cycle load demand risk function, and the risk type of the preprocessed data is determined based on the result.

8. The intelligent pipe cooling circulation control method based on lock construction according to claim 1, characterized in that, The step of using the second algorithm combined with the risk coefficient of the load demand of A cold cycles to control the cooling cycle of B cooling cycles, and generating cooling cycle control decisions, includes: Calculate the comprehensive adjustment coefficient for pipe cooling circulation control zone A, and calculate the power values ​​for pipe cooling water flow regulation and pipe cooling water temperature regulation for pipe cooling circulation control zone A in low-risk and high-risk types respectively. The calculation method is as follows: Among them, TidAjust i b1 represents the overall adjustment coefficient for the i-th pipe cooling cycle control zone. i b2 represents the first risk value of the i-th pipe cooling cycle control zone. i b3 represents the second risk value of the i-th pipe cooling cycle control zone. o This represents the third risk value of the i-th pipe cooling cycle control zone, where a1, a2, and a3 correspond to b1 respectively. i b2 i b3 i Weighting coefficients; Prun1 i This indicates the power value required to adjust the cooling water flow rate of the corresponding cooling circulation pipe when the i-th pipe cooling circulation control zone is in a low-risk type, controlling the output control power of the cooling water flow rate; Prun2 i This indicates the power value required to adjust the cooling water flow rate of the corresponding cooling circulation pipe when the i-th pipe cooling circulation control zone is in a high-risk category; Prun3 represents the output control power for controlling the cooling water flow rate. i This indicates the power value required to adjust the cooling water temperature of the corresponding cooling circulation pipe when the i-th pipe cooling circulation control zone is in a low-risk category; Prun4 represents the output control power for controlling the cooling water temperature. i This indicates the power value required to adjust the cooling water temperature of the corresponding cooling circulation pipe when the i-th pipe cooling circulation control zone is in a high-risk type, and the output control power to control the cooling water temperature; P1 represents the initial power value for adjusting the cooling water flow rate; P2 represents the initial power value for adjusting the cooling water temperature. Prun1 i Prun2 i Prun3 i Prun4 i All of these represent the set values ​​of the cooling circulation control decision output for the cooling circulation pipeline corresponding to the low-risk and high-risk types in the i-th pipe cooling circulation control zone.

9. The intelligent pipe cooling circulation control method based on lock construction according to claim 1, characterized in that, The step of adjusting the cooling water flow rate and cooling water temperature of each pipe based on the set value output by the pipe cooling circulation control decision of B cooling circulation pipes, using a PID control algorithm, includes: Based on the identified cooling circulation pipelines that need adjustment, an input matrix sequence consisting of the set values ​​of the cooling circulation pipelines that need adjustment is established. The cooling circulation pipelines that need adjustment are determined according to whether the corresponding cooling circulation control zone is in a low-risk or high-risk type. Based on the integral PID control algorithm, the control command for the output signal is obtained by reading the data of the input matrix sequence composed of setpoints; According to the control instructions of the output signal, the control execution component adjusts the pipeline cooling water flow rate and pipeline cooling water temperature of the i-th pipeline cooling circulation control zone to realize intelligent dynamic feedback control of pipeline cooling circulation for lock construction. The dynamic feedback control includes: A visualization and interactive platform was established to display the change curves of multi-dimensional environmental data for all pipe cooling circulation control zones during the construction of the ship lock, as well as the parameter change curves of the intelligent pipe cooling circulation dynamic feedback control.

10. An intelligent pipe cooling circulation control system based on lock construction, characterized in that, The system is applied to the method according to any one of claims 1-9, comprising: The region division module is used to divide the lock construction structure into the gate pier region and the gate wall region, divide the gate pier region and the gate wall region into key points, generate A pipe cold circulation control partitions, and pre-set A edge key control nodes in the A pipe cold circulation control partitions. Each partition represents an independent structural type in the gate pier and the gate wall, and each partition contains several key points, which are used for data acquisition. Covering module, used to cover the A tube cooling cycle control zones through B cooling cycle pipelines, each cooling cycle pipeline equipped with an electromagnetic regulating valve and a turbine flowmeter, where B ≤ A; Pretreatment module, used for each edge key control node to receive multi-dimensional data of each tube cooling cycle control zone, preprocess the multi-dimensional data to generate preprocessed data, and transmit the preprocessed data to the cloud cooling control service; the multi-dimensional data includes key node basic data, key node inspection data on the outer surface of the cooling tube, key node scaling data inside the cooling tube, key node ambient temperature, key node ambient humidity, key node air velocity, key node tube cooling equipment status; the preprocessed data includes first preprocessed data and second preprocessed data, where the first preprocessed data does not include key node inspection data on the outer surface of the cooling tube, and the second preprocessed data at least includes key node inspection data on the outer surface of the cooling tube; Cloud computing module, used for the cloud cooling control service to receive the A preprocessed data of the A edge key control nodes, and based on the first deep learning algorithm, predict the A preprocessed data to generate A cold cycle load demand risk coefficients; the cold cycle load demand risk coefficient is set to four values according to the risk type: the first risk value b1, the second risk value b2, the third risk value b3, where b1 < b2 < b3; the risk type is set for the tube cooling cycle control zone and several key points under the tube cooling cycle control zone, and the risk type includes no risk type, low risk type, high risk type and equipment abnormal type, where the risk type of each tube cooling cycle control zone is calculated from the sum of the risk types of several key points controlled under each tube cooling cycle control zone; Decision generation module, if it is judged that there is a risk in the cold cycle load demand risk coefficient of at least 1 tube cooling cycle control zone among the A cold cycle load demand risk coefficients, based on the second algorithm and combined with the A cold cycle load demand risk coefficients, perform tube cooling cycle control on the B cooling cycle pipelines to generate a tube cooling cycle control decision; Execution module, according to the set value output by the tube cooling cycle control decision of the B cooling cycle pipelines, uses the PID control algorithm to adjust the cooling water flow and cooling water temperature of each pipeline for the set value.

Images