Machine learning-fused permafrost region circulating ventilation roadbed cooling system

Abstract

The invention relates to the technical field of permafrost region roadbed cooling, and particularly discloses a permafrost region circulating ventilation roadbed cooling system fused with machine learning. The volume moisture content, the ground temperature and the deformation of the lower part of the roadbed are respectively monitored through a moisture sensor and a distributed optical fiber sensor; the machine learning algorithm module comprises a Prohet time sequence model, and is used for collecting, analyzing and predicting future temperature and ground temperature changes through the Prohet time sequence model; and the control module comprises a fuzzy control strategy and a PID control algorithm. By fusing environmental data and roadbed state data of an artificial intelligence algorithm, opening and closing of the ventilation pipe are automatically controlled, and the problem of thaw collapse deformation of a roadbed in a permafrost region is eliminated and relieved.

Description

Technical Field

[0001] The present invention relates to the technical field of roadbed cooling in permafrost areas, and in particular to a permafrost area circulating ventilation roadbed cooling system integrating machine learning. Background Art

[0002] Since the construction of the Qinghai-Tibet Railway, the permafrost has been warming up at a rate of up to 0.05°C / a due to the combined effects of engineering disturbance and global warming. As its melting range continues to expand, it has become a major issue affecting the line's traffic efficiency and service life. Although the current engineering measures, such as riprap roadbed, heat rods, and ventilated roadbed, have alleviated the problem of permafrost thawing and settlement to a certain extent, the deformation and cracks of the roadbed caused by the melting of permafrost still exist. How to properly deal with the problem of melting permafrost under the roadbed is of great significance to the regional economic development and national defense security of the plateau area.

[0003] Ventilated pipe roadbed is one of the main forms of roadbed in plateau areas. This form of roadbed uses the forced convection caused by the large temperature difference, high altitude and high wind speed in the plateau area to promote the transfer of cold to the lower part of the roadbed in winter, thereby playing the role of actively cooling the roadbed. However, under the background of global warming, the permafrost under the ventilated roadbed is still in an extremely unstable state. The melting of permafrost will inevitably pose a serious threat to the safe operation of the line. The use of effective engineering and technical means to enhance the cooling effect of the ventilated roadbed is a feasible way to cope with the smooth operation of the line under the conditions of climate warming. At present, the use of active suction and ventilation pipe roadbeds with chimneys has been proven to reduce the temperature of the underlying permafrost, but the construction of these methods must be carried out at the same time as the main body of the roadbed. Once damaged in the later stage, it is difficult to repair, and its opening and closing requires manual operation. It is impossible to achieve intelligent and scientific control strategies to fully utilize the unique climatic conditions of the plateau to cool the permafrost under the roadbed. In addition, considering the existing thawing settlement problem of existing lines, the design of the ventilated roadbed structure needs to take into account both existing lines and those that are still in the planning stage. Obviously, the traditional ventilation roadbed structure and opening and closing methods are difficult to meet the actual needs of engineering construction and maintenance. Based on this, the present invention establishes the optimal size of the ventilation roadbed through numerical simulation, and then integrates the machine learning algorithm to accurately control the opening and closing angles of the ventilation pipe to achieve the best cooling effect, constructs a scientific ventilation pipe roadbed structure cooling structure and control strategy, and improves the service life and performance of the roadbed project in permafrost areas. Summary of the invention

[0004] The purpose of the present invention is to provide a permafrost region circulating ventilation roadbed cooling system integrating machine learning to solve the above-mentioned technical problems.

[0005] The purpose of the present invention can be achieved through the following technical solutions: A permafrost region circulating ventilation roadbed cooling system integrating machine learning, comprising: Data acquisition module: Measuring environmental data through meteorological sensors, and monitoring the volume moisture content, ground temperature and deformation of the lower part of the roadbed through moisture sensors and distributed optical fiber sensors. The environmental data include ambient temperature, humidity, wind speed, air pressure, wind direction, long-wave radiation and short-wave radiation; Machine learning algorithm module: including the Prohet time series model, which collects, analyzes and predicts future changes in air temperature and ground temperature through the Prohet time series model; Control module: includes fuzzy control strategy and PID control algorithm. The fuzzy control strategy controls the opening and closing speed of the butterfly disc based on the temperature data monitored on site. The PID control algorithm is used to ensure the opening and closing degree of the butterfly disc. The actuator speed is continuously corrected according to the rotation position sensor data to make the opening and closing degree of the butterfly disc reach the set value.

[0006] As a further solution of the present invention: In the machine learning algorithm module, the specific steps of collecting, analyzing and predicting future air temperature and ground temperature changes by using the Prohet time series model are as follows: The interpolation method is used to preprocess the missing data values, and the wind direction data is expressed in the form of angles and converted into a two-dimensional vector. The specific formula is as follows: ; ; Among them, wind_direction represents the wind direction, x wind_dir represents the wind direction in the x direction, y wind_dir represents the component of wind direction in the y direction; All data are normalized to between 0 and 1. The specific formula is as follows: ; in, is the original data; is the normalized data; is the maximum value in the temperature series; is the minimum value in the temperature series; Use add_regressor to add regression variables. The target output variables are the air temperature and ground temperature for the next 10 days, expressed as: ; ; Wherein, y1(t) is the output air temperature; y2(t) is the output ground temperature; g(t) is the non-periodic change value of air temperature / ground temperature over time; s(t) is the periodic change value of air temperature / ground temperature over time; x1(t) is humidity; x2(t) is wind speed; x3(t) is wind direction; x4(t) is net radiation; x5(t) is volume moisture content; x6(t) is deformation; is a random error; They represent the linear regression coefficients of the corresponding physical variables.

[0007] As a further solution of the present invention: in the machine learning algorithm module, when the daytime and nighttime temperatures are higher than the ground temperature within 2 cm of the ground surface for 10 consecutive days, the butterfly valve is closed through the remote terminal; when the daytime air temperature is higher than the ground temperature, but the nighttime ground temperature is lower than the air temperature, the butterfly valve is closed during the day and opened at night.

[0008] As a further solution of the present invention: when the butterfly valve encounters a cold air event, the butterfly valve is triggered to open automatically. The cold air event means that the temperature drops by 4°C and the humidity increases by more than 10% within 2 hours.

[0009] As a further solution of the present invention: in the control module, the specific steps of the fuzzy control strategy to control the opening and closing speed of the butterfly plate based on the temperature data monitored on site are as follows: Input variables, including temperature, current opening and closing degree and target opening ,The fuzzy set of temperature includes low temperature, suitable temperature and high temperature; The membership function is expressed as: ; In the formula, is the membership function, T is the temperature; The speed adjustment factor f(T) is obtained through logical reasoning, and it is defuzzified into an accurate value to adjust the operating speed of the actuator. , the specific formula is as follows: .

[0010] is the rated speed of the actuator.

[0011] As a further solution of the present invention: in the control module, the specific method of controlling the opening and closing degree of the butterfly plate by the PID control algorithm includes: PID control input butterfly plate current opening x current and target opening x target , the output value is expressed as: ; ; Where μ(t) is the real-time control signal of the actuator; e(t) represents the current error; K p is the proportional gain, which is used to adjust the actuator response speed; K i is the integral gain, used to eliminate steady-state error, K d It is the differential gain, which is used to reduce the actuator overshoot.

[0012] Beneficial effects of the present invention: The present invention realizes automatic control of the opening and closing of ventilation pipes by integrating environmental data and roadbed status data of artificial intelligence algorithms, thereby eliminating and alleviating the problem of roadbed deformation and damage in permafrost areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The present invention will be further described below in conjunction with the accompanying drawings.

[0014] Figure 1 This is a schematic diagram of the grid division of the enlarged port with a diameter 1.0 times the original pipe diameter in the embodiment of the present invention; Figure 2 This is a schematic diagram of the grid division of the enlarged port with a diameter 1.5 times the original pipe diameter in the embodiment of the present invention; Figure 3 This is a schematic diagram of the grid division of the enlarged port 2.0 times the original pipe diameter in the embodiment of the present invention; Figure 4 is a contour map of the expanded opening roadbed temperature when the roadbed has been in service for 30 years under different pipe diameters in an embodiment of the present invention; Figure 5 Schematic diagram of the structure of the circulating ventilation roadbed in an embodiment of the present invention; Figure 6 The present invention Figure 5 An enlarged schematic diagram of the local position of the middle enlarged opening; Figure 7 is a schematic diagram of the structure of the actuator in an embodiment of the present invention; Figure 8 It is a control principle diagram of the circulating ventilation roadbed in an embodiment of the present invention.

[0015] In the figure: 1. Wind speed and direction sensor; 2. Radiation meter; 3. Solar panel; 4. Fixing column; 5. Data acquisition instrument and battery pack; 6. Remote control module; 7. Valve stem; 8. Valve body; 9. Sealing ring; 10. Enlarged mouth; 11. Ventilation pipe; 12. Embankment; 13. Moisture sensor; 14. Distributed optical fiber sensor; 15. Optical fiber demodulator; 16. Low temperature resistant line; 17. Butterfly plate; 18. Rotation position sensor; 19. Bolt; 20. Anti-oxidation coating; 21. Heating wire; 22. Protective cover; 23. SMD thermocouple. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] In this example, the existing Qinghai-Tibet Railway ventilation pipe subgrade is used as the engineering blueprint, and the diameters of both sides of the ventilation pipe are enlarged in equal proportions using the principle of "narrow pipe effect", with the enlargement ratios being 1, 1.5 and 2.0 times of the original pipe diameters respectively. According to the temperature and wind speed change conditions on site, the temperature of the permafrost below the subgrade under different pipe diameters is calculated using FLUENT when the subgrade is in service for the 30th year. The results are as follows: Figure 1-Figure 4 As shown in the figure, it is obvious that as the pipe diameter increases, the temperature of the lower part of the roadbed at the same point gradually decreases, and when the pipe diameter is expanded by 2.0 times, the asymmetric effect of the permafrost temperature under the roadbed is also effectively alleviated. Therefore, the following designs are all carried out on the basis of 2.0 times the original pipe diameter.

[0018] In an embodiment of the present invention, a permafrost region circulating ventilation roadbed cooling system integrating machine learning is specifically provided, comprising: Data acquisition module: It mainly uses meteorological sensors to measure ambient temperature, humidity, wind speed, air pressure and wind direction, long-wave radiation, and short-wave radiation, and uses hydro probe moisture sensor 13 and distributed optical fiber sensor 14 to monitor the volume moisture content, ground temperature, and deformation of the lower part of the roadbed. Environmental data is collected by five-parameter meteorological sensors, and the data monitored and moisture data are recorded by CR1000 data acquisition instrument. The ground temperature and deformation data obtained by the distributed optical fiber sensor 14 are recorded by the optical fiber data acquisition instrument. During the data collection process, the same data collection start time needs to be set, and the collection frequency of all data is 10min / time. The data acquisition device is equipped with a remote wireless data transmission function, which supports remote viewing and export of data. All equipment is powered by solar panels 3 in combination with batteries.

[0019] In the specific data collection process, refer to Figure 5In order to prevent the solar panel 3 from blocking the radiation sensor and causing distortion of the measurement data, the solar panel 3 is fixed to the lower part of the wind speed and wind direction sensor 1 and the radiation meter 2, and the wind speed and wind direction sensor 1 and the radiation meter 2 are fixed to the upper part of the fixed column 4. All connections are rigidly connected by screws. The data acquisition instrument and the battery pack 5 are fixed in the middle of the column to facilitate later maintenance. The ventilation pipe 11 is buried in the middle of the embankment 12, and one side of the expanded opening 10 is connected to the pipe body, and the other is connected to the valve body 8 through a rubber sealing ring 9. The valve body 8 and the valve stem 7 are connected by bolts 19. There is a reserved thread inside the valve stem 7, which is connected to the actuator by a thread. The actuator is fixed to the inside of the protective cover 22 by bolts, and the surface is sprayed with an anti-oxidation coating 20, and the outside is wrapped with a graphene heating wire 21 to ensure that the actuator operates normally under low temperature conditions. A patch thermocouple 23 is installed near the valve stem 7, which is connected to the CR1000 data acquisition instrument to reflect the temperature state of the actuator at all times. Concrete feet are embedded inside the roadbed, and the actuator housing is connected to the feet with expansion screws. Hydro probe moisture sensor 13 and distributed optical fiber sensor 14 are buried in the lower part of the roadbed, and the distributed optical fiber is connected to the optical fiber demodulator 15 to realize the collection and storage of deformation and ground temperature data. The actuator and the battery, and the battery and the solar panel 3 are connected by a low-temperature resistant line 16. The butterfly plate 17 has the same inner diameter as the expansion port 10, and there is a sealing strip between the two. A rotation position sensor 18 is installed at the bottom of the butterfly plate 17 to monitor the rotation angle of the butterfly plate 17. All data collection, analysis and actuator opening are realized by remote control module 6. The protection level of the actuator housing is IP65, ensuring that the actuator is not affected in extreme working environments. At the same time, the actuator needs to be sealed, and the outer side is protected by a protective cover 22 to prevent plateau sand from entering the equipment and ensure long-term stable operation of the circuit. The actuator housing is made of anti-oxidation housing material and coated with an anti-oxidation coating 20, which effectively extends the service life of the equipment. The actuator includes a heating module, which mainly realizes the heating function by a resistance wire. When the patch thermocouple 23 fixed on the actuator senses that the ambient temperature is lower than 0°C, the actuator is heated through the remote control module 6 to avoid component failure due to the extremely low temperature environment.

[0020] The actuator opens and closes at the set angle by receiving the control signal. The control module determines whether to open and close the ventilation pipe 11 according to the environmental sensor data and the Prohet algorithm, generates an opening and closing instruction and sends it to the driver system. The instruction includes the opening and closing angle. The fuzzy control strategy and PID control algorithm are used to parse the instruction and accurately control the opening and closing angle. The drive current is adjusted according to the instruction parameters to control the speed of the actuator. If the instruction is "fully open" or "fully closed", the actuator drives the butterfly plate 17 to the corresponding limit position; if it is "partially open", the motor controls the valve to the set angle. During the execution process, the position sensor feeds back the current position of the butterfly plate 17 to the control circuit in real time to ensure that the butterfly plate reaches the preset position. When an overload occurs, such as freezing between the butterfly plate 17 and the expansion port 10, the control module will automatically stop the actuator and send an alarm signal, and will automatically trigger the safety protection mechanism to cut off the current to avoid damage to the actuator. When the SMD thermocouple 23 detects that the actuator temperature is too high, it will slow down the operating frequency by reducing the speed or delaying to prevent overheating. The drive system feeds back the execution status, including the switch status, current angle, and whether any abnormality occurs, to the drive system through the communication module.

[0021] Machine learning algorithm module: The machine learning algorithm module mainly includes the Prohet time series model, among which the Prohet time series model is mainly used to collect, analyze and predict future changes in air temperature and ground temperature of meteorological data, while the combination of fuzzy control strategy and PID control algorithm is mainly used to achieve precise opening and closing control of the actuator.

[0022] The following will introduce the implementation process of the above machine learning algorithms respectively: The Prohet time series model is used to predict the average daytime and nighttime air temperature and ground temperature in the next 10 days. When the daytime and nighttime air temperatures are higher than the ground temperature within 5 cm of the ground surface for 10 consecutive days, the butterfly plate 17 can be closed through the remote terminal. When the daytime air temperature is higher than the ground temperature, but the nighttime ground temperature is lower than the air temperature, the butterfly plate 17 is closed during the day, and the butterfly plate 17 is opened at night. In addition, when a cold air event occurs, the opening of the butterfly plate 17 is automatically triggered. The automatic opening of the butterfly plate 17 requires that the changes in air temperature and humidity reach the "cold air event" standard. The criterion is that if the temperature suddenly drops by 4°C and the humidity suddenly increases by more than 10% within 2 hours, it is considered to be a cold air event. When the humidity and temperature changes return to the normal range, the butterfly plate 17 is automatically closed, and the "cold air event" is considered to be over.

[0023] First, the interpolation method is used to preprocess the missing data values, express the wind direction data in the form of angles, and convert it into a two-dimensional vector. The specific formula is as follows: (1) (2) wind_direction indicates wind direction, where wind_direction=0° indicates north wind, wind_direction=90° indicates east wind, wind_direction=180° indicates south wind, wind_direction=270° indicates west wind, wind_direction=45° indicates northeast wind, wind_direction=225° indicates southwest wind, wind_direction=135° indicates southeast wind, wind_direction=315° indicates northwest wind, x wind_dir represents the wind direction in the x direction, y wind_dir Represents the component of wind direction in the y direction Next, all the data are normalized to between 0 and 1. The specific formula is as follows: (3) in, is the original data; is the normalized data; is the maximum value in the sequence; is the minimum value in the sequence.

[0024] In order to include all environmental data in the Prohet time series model, it is necessary to further expand the Prohet model in combination with the regression model. Use add_regressor to add regression variables. The target output variables are the air temperature and ground temperature in the next 10 days, which can be expressed as: (4) (5) Wherein, y1(t) is the output air temperature; y2(t) is the output ground temperature; g(t) is the non-periodic change value of air temperature / ground temperature over time; s(t) is the periodic change value of air temperature / ground temperature over time; x1(t) is humidity; x2(t) is wind speed; x3(t) is wind direction; x4(t) is net radiation; x5(t) is volume moisture content; x6(t) is deformation; is a random error; They represent the linear regression coefficients of the corresponding physical variables.

[0025] Control module: The drive system adopts a combination of fuzzy control strategy and PID control algorithm to cope with complex opening and closing operation requirements. The fuzzy control strategy mainly controls the opening and closing speed of the butterfly plate based on the temperature data monitored on site. When the temperature is below 0°C, the actuator speed is reduced to prevent it from freezing. When the temperature is above 0°C, the actuator running speed is also appropriately increased to increase the impact speed of the butterfly plate. The PID control algorithm is used to ensure the opening and closing degree of the butterfly plate. The speed and torque of the actuator are continuously corrected according to the data of the rotation position sensor 18 so that the opening and closing degree of the butterfly plate reaches the set value. The specific implementation steps are as follows: Input variables include air temperature, current opening degree and target opening The fuzzy set of temperature includes low temperature, suitable temperature and high temperature. The membership function can be expressed as: (6) In the formula, is the membership function, and T is the temperature.

[0026] The speed adjustment factor f(T) is obtained through logical reasoning, and it is defuzzified into an accurate value to adjust the operating speed of the actuator. (7) is the rated speed of the actuator Fuzzy control can adjust the initial output upper limit of PID, thereby limiting the actuator speed. When the temperature is low, fuzzy control will reduce , to prevent physical damage to the butterfly plate caused by rapid rotation during freezing, and when the temperature is high, increase , in order to increase the operating speed of the actuator.

[0027] On this basis, the PID control algorithm is used to further accurately control the operating rate of the actuator, that is, to accurately control the movement of the butterfly plate to the set position. PID control input butterfly plate current opening x current and target opening x target , the output value can be expressed as: (8) (9) In the formula, is the real-time control signal of the actuator; e(t) represents the current error; K p is the proportional gain, which is used to adjust the actuator response speed; K i is the integral gain, used to eliminate steady-state error, K d It is the differential gain, which is used to reduce the actuator overshoot.

[0028] The actuator speed is continuously adjusted to the set value by initializing parameters, calculating errors, calculating control outputs, and updating actuator status.

[0029] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A permafrost region circulating ventilation roadbed cooling system integrating machine learning, characterized in that: include: Data acquisition module: Measuring environmental data through meteorological sensors, and monitoring the volume moisture content, ground temperature and deformation of the lower part of the roadbed through moisture sensors and distributed optical fiber sensors. The environmental data include ambient temperature, humidity, wind speed, air pressure, wind direction, long-wave radiation and short-wave radiation; Machine learning algorithm module: including the Prohet time series model, which collects, analyzes and predicts future changes in air temperature and ground temperature through the Prohet time series model; Control module: includes fuzzy control strategy and PID control algorithm. The fuzzy control strategy controls the opening and closing speed of the butterfly disc based on the temperature data monitored on site. The PID control algorithm is used to ensure the opening and closing degree of the butterfly disc. The actuator speed is continuously corrected according to the rotation position sensor data to make the opening and closing degree of the butterfly disc reach the set value.

2. The permafrost region circulating ventilation roadbed cooling system integrating machine learning according to claim 1 is characterized in that: In the machine learning algorithm module, the specific steps of collecting, analyzing and predicting future air temperature and ground temperature changes through the Prohet time series model are as follows: The interpolation method is used to preprocess the missing data values, and the wind direction data is expressed in the form of angles and converted into a two-dimensional vector. The specific formula is as follows: ； ； Among them, wind_direction represents the wind direction, x wind_dir represents the wind direction in the x direction, y wind_dir represents the component of wind direction in the y direction; All data are normalized to between 0 and 1. The specific formula is as follows: ； in, is the original data; is the normalized data; is the maximum value in the sequence; is the minimum value in the sequence; Use add_regressor to add regression variables. The target output variables are the air temperature and ground temperature for the next 10 days, expressed as: ； ； Wherein, y1(t) is the output air temperature; y2(t) is the output ground temperature; g(t) is the non-periodic change value of air temperature / ground temperature over time; s(t) is the periodic change value of air temperature / ground temperature over time; x1(t) is humidity; x2(t) is wind speed; x3(t) is wind direction; x4(t) is net radiation; x5(t) is volume moisture content; x6(t) is deformation; is a random error; They represent the linear regression coefficients of the corresponding physical variables.

3. The permafrost region circulating ventilation roadbed cooling system integrating machine learning according to claim 2 is characterized in that: In the machine learning algorithm module, when the daytime and nighttime temperatures are higher than the ground temperature within 2 cm of the ground surface for 10 consecutive days, the butterfly valve is closed through the remote terminal; when the daytime air temperature is higher than the ground temperature, but the nighttime ground temperature is lower than the air temperature, the butterfly valve is closed during the day and opened at night.

4. The permafrost region circulating ventilation roadbed cooling system integrating machine learning according to claim 2 is characterized in that: When the butterfly valve encounters a cold air event, it triggers the automatic opening of the butterfly valve. The cold air event means that the temperature drops by 4°C and the humidity increases by more than 10% within 2 hours.

5. The permafrost region circulating ventilation roadbed cooling system integrating machine learning according to claim 1 is characterized in that: In the control module, the specific steps of the fuzzy control strategy to control the opening and closing speed of the butterfly plate based on the temperature data monitored on site are as follows: Input variables, including temperature, current opening and closing degree and target opening ,The fuzzy set of temperature includes low temperature, suitable temperature and high temperature; The membership function is expressed as: ； In the formula, is the membership function, T is the temperature; The speed adjustment factor f(T) is obtained through logical reasoning, and it is defuzzified into an accurate value to adjust the operating speed of the actuator. , the specific formula is as follows: ； is the rated speed of the actuator When the temperature is low, the actuator rotates at a lower speed to prevent ice and other phenomena from causing structural damage to the actuator.

6. The permafrost region circulating ventilation roadbed cooling system integrating machine learning according to claim 5 is characterized in that: In the control module, the specific method of controlling the opening and closing degree of the butterfly plate by the PID control algorithm includes: PID control input butterfly plate current opening x current and target opening x target , the output value is expressed as: ； ； Where μ(t) is the real-time control signal of the actuator; e(t) represents the current error; K p is the proportional gain, which is used to adjust the actuator response speed; K i is the integral gain, used to eliminate steady-state error, K d It is the differential gain, which is used to reduce the actuator overshoot.

Images