A Temperature Control Big Data Mining Method for Ultra-high Arch Dams

Abstract

The present invention relates to a method for data mining of temperature control of extra-high arch dams, comprising the following steps: Step 1, collect the process construction data information of the extra-high arch dam and construct a data information warehouse for the construction process of the extra-high arch dam; Step 2, determine the data mining objective according to the actual analysis requirements; Step 3, obtain the original data of the temperature control stage to be analyzed; Step 4, perform data preprocessing on the obtained temperature control data; Step 5, establish a data mining model by using the "Apriori" association rule algorithm; Step 6, optimize the input parameters of the model; Step 7, output strong association rules. This method makes full use of the temperature control-related data information of the extra-high concrete dam, analyzes the association relationship between the concrete temperature and the temperature control parameters in each temperature control stage, constructs a strong association relationship between the concrete temperature state and the temperature control parameters, and provides a reference for formulating reasonable and effective temperature control measure plans in actual projects.

Description

Technical Field

[0001] The present invention relates to a data mining method, and particularly to a large data mining method for temperature control of ultra-high arch dams. Background Art

[0002] During the construction of ultra-high concrete arch dams, strictly controlling the concrete temperature state is one of the effective measures to prevent temperature cracks in the dam body. The temperature of the dam body concrete is affected by many factors, such as concrete materials, pouring schemes, temperature control schemes, and environmental meteorology. Among them, the pouring scheme includes the opening time of the pouring bay, the incoming temperature, the pouring temperature, the pouring intensity, the concrete layer thickness, etc., and the temperature control scheme includes the starting time of cooling water circulation, the duration of water circulation, the water temperature of water circulation, the water flow rate, etc. The concrete temperature is the result of the combined action of all the above factors. In order to accurately grasp the temperature change process of each pouring bay during the construction period, engineers have obtained a large amount of dam body temperature data by using various monitoring means and data acquisition instruments (such as point thermometers, digital thermometers, hand-held temperature recorders, wireless temperature acquisition instruments, distributed optical fiber temperature measurement systems, etc.), and at the same time, have accumulated the data information of the temperature control measures throughout the process of the pouring bay.

[0003] Data mining methods can explore the potential laws between data from a large amount of data, providing a technical means for analyzing the correlation between the concrete temperature state and temperature control-related data. Among them, association rule mining has been deeply studied and widely applied because it can describe the interdependent relationship between data items. Mining the correlation relationship between the concrete temperature and temperature control-related data from the comprehensive and massive temperature control-related data information can provide reference suggestions for formulating the temperature control measure scheme. Therefore, based on the long-term monitoring data such as the internal temperature state and temperature control parameters of the dam concrete in the real environment, effectively analyzing the mutual influence relationship between the internal temperature of the concrete and the temperature control measures has important engineering practical guiding significance for reasonably formulating and improving the temperature control measure scheme. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to propose a large data mining method for temperature control of ultra-high arch dams, revealing the correlation relationship between temperature control measures and temperature control effects, and having important guiding significance for the adjustment of parameters in the feedback design of temperature control measures during the construction period.

[0005] To solve the above technical problem, the present invention proposes the following technical solutions: A large data mining method for temperature control of ultra-high arch dams, comprising the following steps:

[0006] Step 1, collect the process construction data information of the ultra-high arch dam, and construct a data information warehouse for the construction process of the ultra-high arch dam;

[0007] Step 2, determine the data mining target according to the actual analysis requirements;

[0008] Step 3, obtaining the original data of the temperature control stage to be analyzed;

[0009] Step 4: preprocess the acquired temperature control data;

[0010] Step 5: Use the “Apriori” association rule algorithm to establish a data mining model;

[0011] Step 6: Optimize model input parameters;

[0012] Step 7: Output strong association rules.

[0013] In a preferred embodiment, the establishment steps of step 1 are:

[0014] Step 1-1, select the ultra-high concrete arch dam and collect the temperature control related data information of the project;

[0015] Step 1-2: Based on database technology, a data warehouse is established with the collected temperature control related data information of the ultra-high concrete arch dam.

[0016] In a preferred embodiment, the operation steps of step 3 are as follows:

[0017] Step 3-1: Filter out the original temperature control data that meets the requirements from the collected temperature control related data according to the data mining target;

[0018] Step 3-2: Create an experimental data set.

[0019] In a preferred embodiment, the implementation steps of step 4 are:

[0020] Step 4-1: Recode the discrete construction data according to the requirements of the association rule mining method on the data form;

[0021] Step 4-2: discretize the numerical construction data;

[0022] Step 4-3: Create a transactional data set and count the number of interval elements after each variable is discretized.

[0023] In a preferred embodiment, the implementation steps of step 5 are as follows:

[0024] Step 5-1, use the shape Implicit association rules are used to find the correlation between items in the same event. Assume that I is a set of items, I = {I 1 , I 2 ,…,I n}, Head Assume that the transaction database D = {d 1 , d 2 , …d m}, for a specified subset I of items i (1 ≤ i ≤ n) may contain one or more things d j (1 ≤ j ≤ m), regarding M as the antecedent of the rule and N as the consequent of the rule, establish an association rule support degree Sup data model, where the support degree Sup is the probability that a transaction in the database D contains M or N;

[0025]

[0026] Step 5-2: Establish an association rule confidence Conf mathematical model, where the confidence Conf is the probability that a transaction T in the database D contains N on the premise of containing M;

[0027]

[0028] In a preferred solution, in step 6, the minimum support degree threshold min(Sup) and the minimum confidence degree threshold min(Conf) values are optimized to obtain strong correlation rules.

[0029] A temperature control big data mining method for ultra-high arch dams provided by the present invention makes full use of the measured construction data information of ultra-high concrete arch dams. Based on data mining technology, it reveals the correlation between concrete temperature and temperature control data, which has important engineering practical guiding significance for formulating reasonable and effective temperature control measure plans, can reveal the correlation between each temperature control measure and the temperature control effect, and has important guiding significance for the adjustment of parameters in the feedback design of temperature control measures during the construction period. Brief Description of the Drawings

[0030] Figure 1 It is the operation flow chart of the big data mining method of the present invention. Detailed Embodiment

[0031] According to the measured data of the Baihetan project, deeply analyze the correlation between the temperature control data of low-heat cement concrete dams and the maximum temperature of concrete, and lay a foundation for mastering the construction characteristics of the project and the temperature development process of low-heat cement concrete. As Figure 1 shown, the present embodiment includes the following specific steps:

[0032] Step 1: Collect the process construction data information of the ultra-high arch dam and construct a data information warehouse for the ultra-high arch dam construction process.

[0033] Collect the construction data information related to the concrete temperature state during the construction of the Baihetan Project. The pouring plan includes the opening time of the pouring bin, the incoming bin temperature, the pouring temperature, the pouring intensity, the concrete layer thickness, etc. The temperature control plan includes the starting time of cooling water circulation, the duration of water circulation, the water temperature of water circulation, the water flow rate of water circulation, etc. Use the SQLServer database management system to construct a construction information database for low-heat cement concrete dams, providing basic data support for in-depth analysis of the correlation between the dam body temperature state and various construction parameters.

[0034] Step 2: Determine the data mining objective according to the actual analysis requirements.

[0035] Since strictly controlling the maximum temperature of concrete during the construction of concrete dams is one of the effective measures to prevent temperature cracks in the dam body, the data mining objective of the embodiment is determined to be the correlation between the maximum temperature of concrete and the temperature control measure plan.

[0036] Step 3: Obtain the original data of the temperature control stage to be analyzed.

[0037] Select the original temperature control data that meets the requirements from the collected temperature control-related data according to the data mining objective, as shown in Table 1. The statistical data includes the opening date of the pouring bin, the construction season, the concrete grade, the average air temperature, the type of pouring bin, the layer thickness of the pouring bin, the pouring duration, the average pouring intensity, the incoming bin temperature, the pouring temperature, whether water is circulated before the maximum temperature, whether water is continuously circulated before the maximum temperature, the starting water circulation age, the average water temperature of water circulation, the average water flow rate of water circulation, the duration of water circulation, the maximum temperature, and the age corresponding to the maximum temperature, etc. A total of 686 data information is obtained.

[0038] Step 4: Perform data preprocessing on the obtained temperature control data.

[0039] Step 4-1: Since there are many discrete variables in the pouring information data, such character variables cannot be directly used for modeling and need to be re-coded. There are various re-coding methods, such as converting the character value to an integer value, dummy variable (0-1 variable), One-Hot encoding, etc.

[0040] In this embodiment, there are three character variables: the type of bin, the construction season, and the concrete grade. Among them, for the type of bin, the method of "character to numerical value" is used to re-code this variable, and the corresponding relationship between the variable and the numerical value is shown in Table 2; for the construction season and the concrete grade, the method of "dummy variable" is used to re-code this variable, and the corresponding relationship between the variable and the numerical value is shown in Table 2.

[0041]

[0042]

[0043] Table 2 Correspondence Table for Recoding Character Variables

[0044]

[0045] Step 4-2: Discretize the numerical construction data, and the discretization results are shown in Tables 3 to 17.

[0046] Table 3 Discrete Table of Construction Seasons for Pouring Bins

[0047]

[0048] Table 4 Discrete Table of Concrete Grades for Pouring Bins

[0049]

[0050] Table 5 Discrete Table of Pouring Durations for Pouring Bins

[0051]

[0052] Table 6 Discrete Table of Average Temperatures

[0053]

[0054]

[0055] Table 7 Discrete Table of Pouring Intensities for Pouring Bins

[0056]

[0057] Table 8 Discrete Table of Average Pouring Temperatures for Pouring Bins

[0058]

[0059] Table 9 Discrete Table of Average Pouring Temperatures for Pouring Bins

[0060]

[0061] Table 10 Discrete Table of Temperatures after Pouring Bins are Closed

[0062]

[0063] Table 11 Discrete Table of Pouring Bin Types

[0064]

[0065] Table 12 Discrete Table of Layer Thicknesses for Pouring Bins

[0066]

[0067] Table 13 Discrete Table of Maximum Concrete Temperatures

[0068]

[0069]

[0070] Table 14 Discrete Table of Age Corresponding to the Highest Temperature

[0071]

[0072] Table 15 Discrete Table of Cooling Water Passing Duration

[0073]

[0074] Table 16 Discrete Table of Cooling Water Temperature

[0075]

[0076] Table 17 Discrete Table of Cooling Water Flow Rate

[0077]

[0078]

[0079] Step 4-3: Establish a transactional data set, count the number of interval elements after discretization of each variable, as shown in Tables 3 to 17.

[0080] Step 5: Establish a data mining model using the "Apriori" association rule algorithm.

[0081] Step 5-1: According to the mining objective, respectively determine the number of transaction data d contained in M and N, and based on this, count the number of occurrences of each temperature control parameter that has been discretized in M and N. Use the support Sup data model to find all frequent sets that meet the support. j Step 5-2: Use the confidence Conf mathematical model to determine the strong association rules.

[0082]

[0083] Step 5-2: Use the confidence Conf mathematical model to determine the strong association rules.

[0084]

[0085] Step 6: Optimize the input parameters of the model. In this embodiment, the "trial and error method" is used, and combined with the actual situation, the input parameters of the association rule analysis model are selected: the minimum support is 16.6% and the minimum confidence is 50%.

[0086] Step 7: Use the association rule mining model to obtain the association rules between each pouring scheme parameter and the initial temperature state of the concrete according to the set minimum support and minimum confidence. The model calculation results are shown in Table 18.

[0087] Table 18 Results Table of the Association Rule Model for the Initial Temperature of Concrete

[0088]

[0089] As can be seen from Table 18, in the case where the incoming temperature of C40 concrete in the high-temperature season is between (6.5, 7.5] °C and the average pouring temperature is between (11.0, 12.0] °C, for the whiteboard silo, when the pouring duration is between (26.0, 29.0] h and the average air temperature is between (17.0, 19.0] °C, after the pouring silo is closed, the support degree of the concrete temperature within the range of (12.0, 14.0] °C is 20%, and the confidence level is 100%; for the whiteboard silo, when the pouring duration is between (29.0, 32.0] h and the average air temperature is between (25.0, 27.0] °C, after the pouring silo is closed, the support degree of the concrete temperature within the range of (14.0, 16.0] °C is 25%, and the confidence level is 75%; for the whiteboard silo, when the pouring duration is between (32.0, 35.0] h and the average air temperature is between (23.0, 25.0] °C, after the pouring silo is closed, the support degree of the concrete temperature within the range of (16.0, 18.0] °C is 18.8%, and the confidence level is 60%; for the corridor silo, when the pouring duration is between (29.0, 32.0] h and the average air temperature is between (27.0, 29.0] °C, after the pouring silo is closed, the support degree of the concrete temperature within the range of (16.0, 18.0] °C is 25%, and the confidence level is 66.7%.

[0090] In the case where the incoming temperature of C40 concrete in the low-temperature season is between (8.5, 9.5] °C and the average pouring temperature is between (10.0, 11.0] °C, for the whiteboard silo, when the pouring duration is between (23.0, 26.0] h and the average air temperature is between (13.0, 15.0] °C, after the pouring silo is closed, the support degree of the concrete temperature within the range of (10.0, 12.0] °C is 21.1%, and the confidence level is 75%. In the case where the incoming temperature of C35 concrete in the high-temperature season is between (6.5, 7.5] °C and the average pouring temperature is between (11.0, 12.0] °C, for the whiteboard silo, when the pouring duration is between (32.0, 35.0] h and the pouring intensity is between (144.0, 153.0] m 3 / h, after the pouring silo is closed, the support degree of the concrete temperature within the range of (16.0, 18.0] °C is 26.1%, and the confidence level is 66.7%; for the orifice silo, when the pouring duration is between (29.0, 32.0] h, the average air temperature is between (25.0, 27.0] °C, and the pouring intensity is between (126.0, 135.0] m 3 / h, after the pouring silo is closed, the support degree of the concrete temperature within the range of (16.0, 18.0] °C is 16.7%, and the confidence level is 66.7%.

[0091] When the incoming temperature of C35 concrete in the low-temperature season is between (10.5, 11.5] °C and the average pouring temperature is between (11.0, 12.0] °C, for the white board silo, when the pouring duration is between (26.0, 29.0] h and the average air temperature is between (11.0, 13.0] °C, after the pouring silo is closed, the support degree of the concrete temperature within the range of (12.0, 14.0] °C is 16.7%, and the confidence level is 66.7%; when the incoming temperature of C35 concrete in the low-temperature season is between (11.5, 12.5] °C and the average pouring temperature is between (12.0, 13.0] °C, for the white board silo, when the pouring duration is between (23.0, 26.0] h and the average air temperature is between (13.0, 15.0] °C, after the pouring silo is closed, the support degree of the concrete temperature within the range of (14.0, 16.0] °C is 21.1%, and the confidence level is 75%; when the incoming temperature of C35 concrete in the low-temperature season is between (10.5, 11.5] °C and the average pouring temperature is between (12.0, 13.0] °C, for the corridor silo, when the average air temperature is between (13.0, 15.0] °C, after the pouring silo is closed, the support degree of the concrete temperature within the range of (14.0, 16.0] °C is 71.4%, and the confidence level is 60%.

Claims

1. A method for mining big data on temperature control of ultra-high arch dams, characterized in that, it includes the following steps: Step 1: Collect the process construction data information of the ultra-high arch dam and construct a data information warehouse for the construction process of the ultra-high arch dam; Step 2: Determine the data mining target according to the actual analysis requirements; Step 3: Obtain the original data of the temperature control stage to be analyzed; Step 4: Perform data preprocessing on the obtained temperature control data, and the implementation steps are: Step 4-1: According to the requirements of the association rule mining method for the data form, re-encode the discrete construction data, including three items: bin type, construction season, and concrete grade. Among them, the bin type uses the "character to numerical value" method to re-encode the variable, and the construction season and concrete grade use the "dummy variable" method to re-encode the variable; Step 4-2: Discretize the numerical construction data; Step 4-3: Establish a transactional data set and count the number of interval elements of each variable after discretization; Step 5: Use the "Apriori" association rule algorithm to establish a data mining model; The implementation steps of Step 5 are as follows: Step 5-1: Use the association rule in the form of "M⇒N" to discover the correlation of items in the same event. Assume that I is the set of items, , , , and . Let the transaction database be . For the specified item subset which may contain one or more transactions . Regard as the antecedent of the rule and as the consequent of the rule to establish the support data model, where the support is the probability that the transaction in the database D contains or . ； Step 5-2, Establish the confidence of association rules mathematical model, where the confidence is the probability that the transaction in database D contains on the premise of containing ; ； Step 6: Optimize the model input parameters; Step 7: Output strong association rules.

2. A method for mining big data on temperature control of ultra-high arch dams according to claim 1, characterized in that, the establishment steps of Step 1 are: Step 1-1: Select an ultra-high concrete arch dam and collect the temperature control-related data information of the project; Step 1-2: Based on database technology, establish a data warehouse for the temperature control-related data information of the ultra-high concrete arch dam collected.

3. A method for mining big data on temperature control of ultra-high arch dams according to claim 1, characterized in that, the operation steps of Step 3 are as follows: Step 3-1: Screen out the original temperature control data that meets the requirements from the collected temperature control-related data according to the data mining target; Step 3-2: Establish an experimental data set.

4. A method for mining big data on temperature control of ultra-high arch dams according to claim 1, characterized in that, The step 6 optimizes the values of the minimum support threshold and the minimum confidence threshold to obtain strong correlation rules.

Images