Water flow state analysis method for ship lock channel based on big data analysis

Abstract

The present invention relates to the technical field of water flow data analysis, and in particular to a water flow state analysis method for a ship lock and waterway based on big data analysis, comprising: constructing a water flow state model, then dividing historical hydrological data into multiple data groups according to the collection time, merging and analyzing the data groups, determining key influencing parameters, determining a single number set based on the key influencing parameters, obtaining a data scenario by analyzing the characteristics of the external environment in the single number set, and establishing a weight change function of the key influencing parameters in each data scenario, finally collecting the hydrological data and environmental data of the ship lock and waterway in real time, matching the information with the data scenario, and calculating the real-time weight coefficients of the key influencing parameters in the corresponding weight change function, so that the system can dynamically adjust the water flow state model according to real-time environmental changes, thereby enhancing the adaptability and flexibility of the system to the complex and changeable ship lock and waterway water flow environment.

Description

Technical Field

[0001] The present invention relates to the technical field of water flow data analysis, and in particular to a water flow state analysis method for a ship lock and waterway based on big data analysis. Background Art

[0002] A ship lock refers to the space enclosed by the upper and lower lockheads and the lock walls on both sides. The lock walls are equipped with mooring posts, floating mooring rings, etc. for ships to moor when they are moored in the lock chamber. Ships passing through the lock rise and fall with the rise and fall of the water level in the lock chamber, allowing ships to overcome the concentrated water level difference in the waterway. It is a box-shaped navigation structure.

[0003] The prior art CN118607400A discloses a method for analyzing water flows in a navigation channel and a waterway during the filling and discharging operation of a ship lock, comprising: using a one-dimensional, two-dimensional, and three-dimensional joint mathematical model to analyze water flows in a navigation channel and a waterway during the filling and discharging operation of a ship lock; through the correlation between the one-dimensional mathematical model, the two-dimensional mathematical model, and the three-dimensional mathematical model, a large amount of computing time can be saved, and the water flow conditions in key areas of concern can be accurately analyzed, thereby greatly improving the accuracy of water flow analysis; setting virtual monitoring points corresponding one-to-one to actual monitoring points, combining the water flow calculation results obtained by numerical simulation with the water flow data measured at the actual monitoring points, and replacing the water flow data measured at the virtual monitoring points with the water flow data measured at the actual monitoring points, thereby making the established neural network model more in line with the actual situation.

[0004] However, the existing lock and channel flow state analysis technology fails to fully utilize the advantages of big data and has obvious deficiencies in data integration, feature extraction, and model construction. This includes the fact that during data processing, the weight coefficients corresponding to different influencing parameters under different environments will dynamically change. If the dynamic changes in parameter weights are not taken into consideration, the accuracy of water flow state analysis will be reduced. Summary of the Invention

[0005] The purpose of the present invention is to solve the problems in the background technology and to propose a water flow state analysis method for a ship lock channel based on big data analysis.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions:

[0007] A method for analyzing water flow conditions in a ship lock and waterway based on big data analysis, the method specifically comprising the following steps:

[0008] Step 1: Collect historical hydrological data of the selected lock channel and obtain a water flow state model based on the recurrent neural network model;

[0009] Step 2: Based on the collection time, the historical hydrological data are divided into multiple data groups, and the key influencing parameters of each data group are determined using the differential calculation method;

[0010] Based on key influencing parameters, the data groups are merged to obtain a single data set. The external environment in the single data set is then obtained and the characteristics of the external environment are analyzed to obtain the data scenario of the single data set.

[0011] Step 3: Calculate the node weights of the key influencing parameters in the single-item data set. Based on the node weights and the data in the single-item data set, establish a multivariate linear regression model to obtain the weight change function of the key influencing parameters in each data scenario.

[0012] Step 4: Collect hydrological and environmental data of the lock channel in real time, match the environmental data with the data scenario, and calculate it in the corresponding weight change function to determine the real-time weight coefficient of the corresponding key influencing parameters;

[0013] Step 5: Input the real-time weight coefficients of key parameters and the real-time collected hydrological data into the water flow state model at the same time, and perform data analysis to obtain the water flow state of the lock channel.

[0014] As a further solution of the present invention, after obtaining historical hydrological data, it is necessary to first perform data preprocessing on the historical hydrological data, wherein the data preprocessing includes removing outliers and filling missing values, and then the preprocessed historical hydrological data is subjected to a recurrent neural network model and differential calculation.

[0015] As a further solution of the present invention, the difference calculation method includes:

[0016] S11: Acquire historical hydrological data under various environments, wherein the historical hydrological data includes flow velocity, flow direction, flow rate, and water level;

[0017] According to the collection time, the historical hydrological data collected at the same time are marked as a data group, wherein a data group contains the flow velocity, flow direction, flow rate and water level data collected at the same time;

[0018] Then, the data groups in the historical hydrological data are arranged in chronological order from far to near to obtain a data sequence;

[0019] S12: Set the first data group in the data sequence as the starting data point. Starting from the starting data point, subtract data group (i-1) from data group i to obtain the differential data of data group i, where i∈[1,n], n represents the total number of remaining data groups in the data sequence except the starting data point. Further, set the data group of the starting data point to data group 0. When i takes the value of 1, data group 1 is subtracted from data group 0 to obtain the differential data of data group 1.

[0020] As a further solution of the present invention, a method for determining key influencing parameters includes:

[0021] Obtain the ship operation status of the selected lock channel and match the ship operation status with the data group according to the time relationship, where the ship operation status refers to the speed of the ship passing through the lock;

[0022] In chronological order, the ship's operating status is processed according to the calculation method of differential data to obtain the differential status;

[0023] The differential data and differential state are obtained at the same time, and the differential data and differential state of the same data group are matched one by one according to the data group. Each data in the differential data is correlated with the differential state in turn to obtain the data correlation value. Then, in the same data group, the maximum value of the data correlation value is taken as the key influencing parameter of this data group.

[0024] As a further solution of the present invention, in a data group collected at the same time, based on the different unit collection times, there are multiple data collected by different unit collection times. At the same time, when calculating the differential data, it is necessary to first perform mean processing on the data in the data group according to the parameter type, and then perform difference calculation on the data after mean processing to obtain differential data.

[0025] As a further solution of the present invention, a method for calculating the node weights of key influencing parameters includes:

[0026] SS1: Select any data scenario and mark it as the target analysis environment. Get the single item data set corresponding to the target analysis environment. First use the formula The data in the single-item data set are normalized, where b represents different types of parameters in the single-item data set. Furthermore, b represents flow velocity, flow direction, flow rate, and water level, and j represents the specific value of parameter b. represents the minimum value in parameter b, Indicates the maximum value of parameter b;

[0027] SS2: After the normalization process is completed, the mean and standard deviation of the parameter b in each data group are calculated according to the data group to obtain the parameter mean Pb and parameter standard deviation Bb;

[0028] SS3: Utilizing formulas Calculate the coefficient of variation CVb of parameter b in each data group, and obtain the coefficient of variation CVg of key influencing parameters, where g∈b, and again use the formula The node weight Qm of the key influencing parameter is calculated, where r represents the total label value of parameter b, and m in the node weight Qm represents different data groups.

[0029] As a further solution of the present invention, a method for obtaining a weight change function of key influencing parameters includes:

[0030] The parameter b in the target analysis environment is set as the independent variable, and the node weight Qm is set as the dependent variable. Then, the linear regression model is used to transfer the independent variable and the dependent variable to the linear regression model respectively, and data training is performed to establish a multiple linear regression model. ,in, represents the intercept, represents the error term, 、 、 as well as is the regression coefficient, D1, D2, D3 and D4 are the data after normalization of parameter b.

[0031] As a further solution of the present invention, a method for obtaining a real-time weight coefficient includes:

[0032] The hydrological data and environmental data in the lock channel are collected in real time. First, based on the real-time collected environmental data, the real-time collected environmental data is searched in the scenario data of the single data set, and the matched scenario data is set as the real-time target data;

[0033] The weight change function corresponding to the real-time target data is obtained, and the real-time collected environmental data is input into the weight change function to obtain the real-time weight coefficient of the key influencing parameters.

[0034] Compared with the existing technology, the advantages of the present invention are:

[0035] The present invention collects historical hydrological data of the selected ship lock and waterway, and constructs a water flow state model based on a recurrent neural network model. The historical hydrological data is then divided into multiple data groups according to the collection time, and the data groups are merged and analyzed to determine the key influencing parameters. A single number set is determined based on the key influencing parameters. By analyzing the characteristics of the external environment in the single number set, a data scenario is obtained, and a weight change function of the key influencing parameters in each data scenario is established. Finally, the hydrological data and environmental data of the ship lock and waterway are collected in real time, and information is matched with the data scenario. The real-time weight coefficients of the key influencing parameters are calculated in the corresponding weight change function, so that the system can dynamically adjust the water flow state model according to real-time environmental changes, thereby enhancing the system's adaptability and flexibility to the complex and changeable ship lock and waterway water flow environment, and further improving the accuracy of the analysis results. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a schematic diagram of the method flow structure of the present invention. DETAILED DESCRIPTION

[0037] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0038] Reference Figure 1 A method for analyzing water flow conditions in a ship lock channel based on big data analysis specifically includes the following steps:

[0039] Step 1: Select a ship lock channel, and based on the selected ship lock channel, obtain historical hydrological data of the ship lock channel in various environments, then preprocess the historical hydrological data, and use an artificial intelligence model to train the preprocessed historical hydrological data to obtain a water flow state model. The artificial intelligence model in this embodiment is set to a recurrent neural network model, and the recurrent neural network belongs to the existing technology and will not be described in detail here. Furthermore, data preprocessing includes removing outliers, filling missing values, etc. to ensure data quality;

[0040] Step 2: Obtain the historical hydrological data after data preprocessing again, and perform data analysis on the historical hydrological data to determine the data scenario. The specific method for determining the data scenario includes:

[0041] S1: Acquire historical hydrological data under various environments, including flow velocity, flow direction, flow rate, and water level, and identify key influencing parameters in each set of historical hydrological data. Specific methods for identifying key influencing parameters include:

[0042] S11: Marking the historical hydrological data collected at the same time as a data group according to the collection time, wherein a data group contains the flow velocity, flow direction, flow rate, and water level data collected at the same time;

[0043] Then, the data groups in the historical hydrological data are arranged in chronological order from far to near to obtain a data sequence;

[0044] S12: Set the first data group in the data sequence as the starting data point. Starting from the starting data point, subtract data group (i-1) from data group i to obtain the differential data of data group i, where i∈[1,n], n represents the total number of data groups in the data sequence except the starting data point. Further, set the data group of the starting data point as data group 0. When i is 1, the differential data of data group 1 is obtained by subtracting data group 0 from data group 1.

[0045] It should be further explained that based on the unit collection time, in a data group collected at the same time, there are multiple data collected by different unit collection times. For example, if a time is set to 30 minutes and the unit collection time is set to 5 minutes, there are 6 hydrological data collected based on the unit time in a data group. At the same time, when calculating the differential data, it is necessary to first perform mean processing on the data in the data group according to the parameter type, and then perform difference calculation on the data after mean processing to obtain the differential data;

[0046] S13: Obtaining the ship operation status of the selected lock channel, and corresponding the ship operation status to the data group according to the time relationship. In this embodiment, the ship operation status refers to the ship's speed through the lock;

[0047] Then, the ship operation status is processed according to the method in step S12 above in chronological order to obtain a differential state;

[0048] Acquire differential data and differential states simultaneously, and make one-to-one correspondence between the differential data and the differential states of the same data group according to the data group. Then, perform correlation analysis on each data in the differential data and the differential state in turn to obtain data correlation values. For example, first perform correlation analysis on the flow velocity in the data group and the differential state to obtain the data correlation value of the flow velocity. Then, perform correlation analysis on the water level in the data group and the differential state to obtain the data correlation value of the water level. This process is repeated to obtain data correlation values ​​of each data in the data group.

[0049] Then, in the same data group, the maximum value of the data correlation value is taken as the key influencing parameter of this data group;

[0050] Furthermore, the correlation analysis method in this embodiment adopts a polynomial regression analysis method for processing. The specific processing process belongs to the existing technology and will not be described in detail here.

[0051] S2: classify the historical hydrological data according to the key influencing parameters, so that the data groups with the same key influencing parameters are merged to obtain multiple single item data sets, wherein the key influencing parameters of the data groups in a single item data set are the same;

[0052] S3: Randomly select a single number set, take this single number set as an example, identify the external environment corresponding to each data group in this single number set, then use the external environment of the data group in this single number set as input data, use the feature extraction algorithm to calculate the environmental features of this single number set, and set this environmental feature as the data scenario of this single number set;

[0053] Furthermore, the feature extraction algorithm in this embodiment selects an autoencoder based on neural network feature extraction, and the use of an autoencoder for feature extraction belongs to the prior art and will not be described in detail here;

[0054] Step 3: Randomly select a data scenario from multiple data scenarios and mark it as the target analysis environment. Taking the target analysis environment as an example, determine the weight change function in the target analysis environment. The specific method for determining the weight change function of the target analysis environment includes:

[0055] SS1: Get the single item number set corresponding to the target analysis environment, first use the formula The data in the single-item data set are normalized, where b represents different types of parameters in the single-item data set. Furthermore, b represents flow velocity, flow direction, flow rate, and water level, and j represents the specific value of parameter b. represents the minimum value in parameter b, Indicates the maximum value of parameter b;

[0056] SS2: After the normalization process is completed, the mean and standard deviation of the parameter b in each data group are calculated according to the data group to obtain the parameter mean Pb and parameter standard deviation Bb. Furthermore, when calculating the parameter mean Pb and parameter standard deviation Bb, the normalized data are used for calculation. This can eliminate the dimension effect between different variables and enable comparison of multiple parameters on the same scale.

[0057] SS3: Utilizing formulas Calculate the coefficient of variation CVb of parameter b in each data group, and obtain the coefficient of variation CVg of key influencing parameters, where g∈b, and again use the formula The node weight Qm of the key influencing parameter is calculated, where r represents the total label value of parameter b and m in the node weight Qm represents different data groups;

[0058] SS4: Set the parameter b in the target analysis environment as the independent variable and the node weight Qm as the dependent variable. Then, use the linear regression model to transfer the independent variable and the dependent variable to the linear regression model respectively, and perform data training to establish a multiple linear regression model. ,in, represents the intercept, represents the error term, 、 、 as well as is the regression coefficient, D1, D2, D3 and D4 are the data after normalization of parameter b;

[0059] Furthermore, the multivariate linear regression model of the target analysis environment is set as the weight change function of the key influencing parameters. At the same time, the remaining data scenarios are sequentially used as target analysis environments and processed according to the above method to obtain the weight change function of the key influencing parameters in each data scenario.

[0060] Step 4: Real-time collection of hydrological and environmental data in the lock channel. First, based on the real-time collected environmental data, the real-time collected environmental data is searched in the scenario data of the single item data set, and the matched scenario data is set as the real-time target data.

[0061] Obtain the weight change function corresponding to the real-time target data, and input the real-time collected environmental data into the weight change function to obtain the real-time weight coefficient of the key influencing parameters;

[0062] Step 5: Obtain the water flow state model of the lock channel, input the real-time weight coefficients of key parameters and the real-time collected hydrological data into the water flow state model at the same time, and perform data analysis to obtain the water flow state of the lock channel.

[0063] The above description is only a preferred specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with the technical field, within the technical scope disclosed by the present invention, who makes equivalent replacements or changes based on the technical solution and inventive concept of the present invention, should be covered by the scope of protection of the present invention.

Claims

1. A method for analyzing water flow conditions in a ship lock and waterway based on big data analysis, characterized in that: The method specifically comprises the following steps: Step 1: Collect historical hydrological data of the selected lock channel and obtain a water flow state model based on the recurrent neural network model; Step 2: Based on the collection time, the historical hydrological data are divided into multiple data groups, and the key influencing parameters of each data group are determined using the differential calculation method; Among them, the methods for determining key influencing parameters include: Obtain the ship operation status of the selected lock channel and match the ship operation status with the data group according to the time relationship, where the ship operation status refers to the speed of the ship passing through the lock; In chronological order, the ship's operating status is processed according to the calculation method of differential data to obtain the differential status; Acquire differential data and differential states simultaneously, match differential data and differential states of the same data group one by one according to the data group, perform correlation analysis on each data in the differential data and the differential state in turn to obtain data correlation values, and then take the maximum value of the data correlation value in the same data group as the key influencing parameter of this data group; Based on key influencing parameters, the data groups are merged to obtain a single data set. The external environment in the single data set is then obtained and the characteristics of the external environment are analyzed to obtain the data scenario of the single data set. Step 3: Calculate the node weights of the key influencing parameters in the single-item data set. Based on the node weights and the data in the single-item data set, establish a multivariate linear regression model to obtain the weight change function of the key influencing parameters in each data scenario. Step 4: Collect hydrological and environmental data of the lock channel in real time, match the environmental data with the data scenario, and calculate it in the corresponding weight change function to determine the real-time weight coefficient of the corresponding key influencing parameters; The calculation method of the node weight of key influencing parameters includes: SS1: Select any data scenario and mark it as the target analysis environment. Get the single item data set corresponding to the target analysis environment. First use the formula The data in the single number set are normalized, where b represents different types of parameters in the single number set, b represents flow velocity, flow direction, flow rate and water level, and j represents the specific value of parameter b. represents the minimum value in parameter b, Indicates the maximum value of parameter b; SS2: After the normalization process is completed, the mean and standard deviation of the parameter b in each data group are calculated according to the data group to obtain the parameter mean Pb and parameter standard deviation Bb; SS3: Utilizing formulas Calculate the coefficient of variation CVb of parameter b in each data group, and obtain the coefficient of variation CVg of key influencing parameters, where g∈b, and again use the formula The node weight Qm of the key influencing parameter is calculated, where r represents the total label value of parameter b and m in the node weight Qm represents different data groups; Step 5: Input the real-time weight coefficients of key parameters and the real-time collected hydrological data into the water flow state model at the same time, and perform data analysis to obtain the water flow state of the lock channel.

2. The water flow state analysis method for a ship lock and waterway based on big data analysis according to claim 1 is characterized in that: After obtaining the historical hydrological data, it is necessary to preprocess the historical hydrological data first. The data preprocessing includes removing outliers and filling missing values. The preprocessed historical hydrological data is then subjected to a recurrent neural network model and differential calculation.

3. The method for analyzing water flow conditions in a ship lock and waterway based on big data analysis according to claim 1 is characterized in that: Differential calculation methods include: S11: Acquire historical hydrological data under various environments, wherein the historical hydrological data includes flow velocity, flow direction, flow rate, and water level; According to the collection time, the historical hydrological data collected at the same time are marked as a data group, wherein a data group contains the flow velocity, flow direction, flow rate and water level data collected at the same time; Then, the data groups in the historical hydrological data are arranged in chronological order from far to near to obtain a data sequence; S12: Set the first data group in the data sequence as the starting data point. Starting from the starting data point, subtract data group (i-1) from data group i to obtain the differential data of data group i, where i∈[1,n], n represents the total number of remaining data groups in the data sequence except the starting data point. Set the data group of the starting data point to data group 0. When i takes the value of 1, data group 1 is subtracted from data group 0 to obtain the differential data of data group 1.

4. The method for analyzing water flow conditions in a ship lock and waterway based on big data analysis according to claim 3 is characterized in that: In the data group collected at the same time, based on the different unit collection times, there are multiple data collected by different unit collection times. At the same time, when calculating the differential data, it is necessary to first perform mean processing on the data in the data group according to the parameter type, and then perform difference calculation on the data after mean processing to obtain differential data.

5. The method for analyzing water flow conditions in a ship lock and waterway based on big data analysis according to claim 1 is characterized in that: The method for obtaining the weight change function of the key influencing parameters includes: The parameter b in the target analysis environment is set as the independent variable, and the node weight Qm is set as the dependent variable. Then, the linear regression model is used to transfer the independent variable and the dependent variable to the linear regression model respectively, and data training is performed to establish a multiple linear regression model. ,in, represents the intercept, represents the error term, 、 、 as well as is the regression coefficient, D1, D2, D3 and D4 are the data after normalization of parameter b.

6. The method for analyzing water flow conditions in a ship lock and waterway based on big data analysis according to claim 1, characterized in that: Methods for obtaining real-time weight coefficients include: The hydrological data and environmental data in the lock channel are collected in real time. First, based on the real-time collected environmental data, the real-time collected environmental data is searched in the scenario data of the single data set, and the matched scenario data is set as the real-time target data; The weight change function corresponding to the real-time target data is obtained, and the real-time collected environmental data is input into the weight change function to obtain the real-time weight coefficient of the key influencing parameters.

Images