A soft-sensing method for typhoon-proof suspension system driven by mechanism and data hybrid

Abstract

The present invention relates to a soft measurement method for a typhoon-proof suspension system driven by a mechanism and data hybrid. The method first identifies the structural characteristics of the typhoon-proof suspension system, and determines the auxiliary variables according to the monitored dominant variables; secondly, based on the selected dominant variables and auxiliary variable types, the sensor type and monitoring position are determined, the monitoring data are collected and data processing is performed based on the multi-source information fusion technology, and a monitoring data history library is established; thirdly, a mechanism model of the typhoon-proof suspension system is established, a data-driven model is established based on a machine learning algorithm, model parameter identification is carried out in combination with the monitoring data, and a soft measurement model of the typhoon-proof suspension system is proposed by integrating the mechanism model and the data-driven model; finally, the soft measurement data is compared with the monitoring data, and the established soft measurement model is corrected online and offline. The present invention realizes the soft measurement of the difficult and unmeasurable data of the typhoon-proof suspension system by constructing a mechanism and data hybrid driven model, and improves the accuracy of the soft measurement data.

Description

Technical Field

[0001] The invention belongs to the field of marine oil and gas development, and in particular relates to a soft measurement method for a typhoon-proof suspension system driven by a hybrid mechanism and data. Background Art

[0002] Oil and gas exploration and development have moved to the ocean, and are constantly moving towards deep waters with harsh environmental conditions and frequent typhoons. Drilling risers are key equipment for offshore oil and gas exploration and development. When encountering strong typhoons, there is a need to evacuate the risers by hanging them from the platform to reduce the time for recovering and lowering the risers. During the navigation and evacuation process of the platform-suspended riser system, the riser system vibrates violently under the excitation of the movement of the drilling platform and the harsh marine environmental load, which is prone to collision with the moon pool and yield fracture accidents. In order to improve the survivability of the riser system under typhoons, a typhoon-proof suspension system is proposed, which includes a hydraulic system, an actuator and a control system.

[0003] The structure of the typhoon-proof suspension system is complex and special, and the system performance and response have the problem of "incomplete measurement": the throttle valve performance parameters and pipeline loss of the hydraulic system are difficult to measure, the friction resistance of the hydraulic cylinder of the actuator is difficult to evaluate, and the friction resistance and stress of the stabilizer position are difficult to monitor or even impossible to measure. However, the performance parameters and accurate real-time response monitoring of the hydraulic system and the actuator are crucial for the control system to achieve efficient control and ensure the safety of the typhoon-proof suspension system and the watertight pipe system. It is urgent to propose a monitoring method for the typhoon-proof suspension system. At present, researchers have tried to solve the problem of "incomplete measurement" based on soft measurement technology. Soft measurement technology is mainly realized through mechanism models, data-driven models or a combination of the two. Due to the complex and special structure of the typhoon-proof suspension system, which also exhibits the characteristics of constraint complexity and nonlinearity of the hydraulic system, it is difficult to achieve accurate online measurement of the performance parameters of the typhoon-proof suspension system and the real-time response status of unmeasurable and difficult-to-measure positions based on a single model. Summary of the invention

[0004] In view of the above problems, the present invention provides a soft measurement method for typhoon protection suspension system driven by mechanism and data hybrid, which realizes online soft measurement of typhoon protection suspension system parameters by fusing mechanism model and data driven model.

[0005] The technical solution adopted by the present invention is: a soft measurement method for typhoon protection suspension system driven by mechanism and data hybrid, and the implementation steps of the method are as follows:

[0006] Step 1: Identify the structural characteristics of the typhoon-proof suspension system, including the components and hydraulic circuits of the hydraulic system, the components and matching relationships of the actuators, and the parameter types of all components; determine the dominant variables to be monitored based on the structural characteristics of the typhoon-proof suspension system, and further determine the auxiliary variables; establish the final auxiliary variables based on correlation analysis;

[0007] Step 2: Determine the type of sensor and sensor layout location based on the selected dominant variable and auxiliary variable types; collect real-time monitoring information from the sensor, perform data processing based on multi-source information fusion technology, and obtain a monitoring data history library;

[0008] Step 3: Establish a mechanism model of the typhoon-proof suspension system based on hydraulic fluid mechanics and pipe column structural mechanics, including a hydraulic system mechanism model and an actuator mechanism model; establish a data-driven model based on a machine learning algorithm; integrate monitoring data and the mechanism model to carry out model parameter identification and data-driven model training; integrate the established mechanism model and data-driven model to propose a soft-sensing model of the typhoon-proof suspension system driven by a combination of mechanism and data;

[0009] Step 4: Use the soft measurement model of the typhoon protection suspension system to obtain real-time soft measurement data; regularly compare the soft measurement data with the monitoring data, and use the updated monitoring data history library to realize online and offline correction of the soft measurement model of the typhoon protection suspension system.

[0010] Furthermore, in the first step, the hydraulic system includes a high-pressure accumulator, a cartridge throttle valve, a one-way valve, a low-pressure accumulator, an oil tank, a variable pump, and a relief valve; the actuator includes a hydraulic cylinder, a suspension pup joint, a stabilizer and an articulated joint; the dominant variables are the pressure drop of the cartridge throttle valve, the lateral force at the top of the hydraulic cylinder, the friction resistance of the hydraulic cylinder, the stress of the suspension pup joint, and the friction resistance at the position of the stabilizer; the auxiliary variables are the oil volume of the high-pressure accumulator, the pressure difference between the inlet and outlet of the cartridge throttle valve, the pressure of the rodless chamber and the rod chamber of the hydraulic cylinder, the load on the top of the hydraulic cylinder and the extension of the piston rod, the stress at the bottom of the suspension pup joint, the oil temperature at some positions of the hydraulic system, and the movement posture of the drilling platform and the actuator; there is an interactive substitution relationship between the dominant variables and the auxiliary variables to solve the complex correlation between the variables.

[0011] Furthermore, in the second step, the sensors are divided into pressure sensors, load sensors, temperature sensors, attitude sensors, displacement sensors, and strain gauges according to the types of variables; the strain gauge group is used to monitor the bottom stress of the suspension pup joint, and the strain gauges are evenly arranged around the bottom side of the suspension pup joint, with four or eight groups of them, which are used to solve the axial stress and bending stress at the monitoring position of the suspension pup joint; based on multi-source information processing technology, feature recognition, signal identification, signal processing estimation and multi-source synthesis processing are performed on the original signal to improve the authenticity and quality of the monitoring data; the platform motion posture and the extension of the hydraulic cylinder piston rod are used to obtain the spatial motion posture of the actuator, including the hydraulic cylinder inclination angle and the piston rod extension, and the motion position of the stabilizer.

[0012] Furthermore, in the third step, the mechanism model of the hydraulic system includes a plug-in throttle valve model, a pipeline pressure loss model, an expansion and compression model of hydraulic oil and pipelines, and a hydraulic cylinder friction model:

[0013] The plug-in throttle valve model is

[0014]

[0015] In the formula, C d is the flow coefficient, A T is the flow cross-sectional area of ​​the cartridge throttle valve port, Δp is the pressure difference before and after the cartridge throttle valve, ρ is the density of the hydraulic oil, λ is the opening of the cartridge throttle valve, and A is the maximum flow cross-sectional area of ​​the cartridge throttle valve;

[0016] The pipeline pressure loss model is:

[0017]

[0018] In the formula, v is the average flow velocity of the hydraulic oil in the pipe, ∑ξ is the comprehensive resistance coefficient;

[0019] The expansion and compression model of hydraulic oil and pipeline is:

[0020]

[0021] In the formula, β g is the comprehensive expansion coefficient, V g0 is the initial volume, ΔV g is the volume increment, Δp g is the pressure increment, where the expansion and compression of the hydraulic oil and pipeline are considered together;

[0022] The friction model of the hydraulic cylinder is:

[0023] f=F p -F n

[0024] In the formula, F p 、F n are respectively the hydraulic oil thrust and the piston rod top force. Due to different hydraulic circuits, F p The expression of is different; further use the solved suspension short section monitoring position axial stress F la The friction force f of the centralizer can be calculated by the force acting on the piston rod top 1a for

[0025] f la =F n -F la

[0026] According to the mechanism model of the actuator, the local finite element model and the global finite element model of the actuator structure mechanics analysis are established as follows:

[0027]

[0028] Wherein, equation ① and ② are local finite element and global finite element models respectively. K is the stiffness matrix, M is the mass matrix, C is the damping matrix, F is the load vector, δ is the displacement vector; the global finite element analysis results provide input parameters for the local finite element analysis;

[0029] According to the monitoring data, the intermediate parameters of the hydraulic system mechanism model are obtained, and a data-driven model of the hydraulic system monitoring data and intermediate parameters is formed based on the reinforcement learning algorithm. The hydraulic system soft measurement model of the typhoon-proof suspension system is formed by integrating the mechanism model and the data-driven model, which is used to monitor the pressure drop of the plug-in throttle valve and the pressure loss parameters of the pipeline. Based on the same method, a hydraulic cylinder friction soft measurement model is established; the mechanical characteristics simulation analysis of the actuator is carried out to obtain the mechanical response parameters, including the top lateral force of the suspension short section and the stress at the lateral constraint position of the suspension short section. The mechanical simulation input parameters, the characteristic parameters of the actuator and the mechanical response parameters are trained based on the machine learning algorithm to obtain the data-driven model of the simulation data. The data-driven model and the monitoring data are processed in a weighted fusion manner to obtain the actuator soft measurement model of the typhoon-proof suspension system. The response parameters of the unmeasurable position are inferred from the response parameters of the measurable position of the actuator; when the data-driven model of the actuator is trained, the cross-sectional stress in the output parameters can be split into axial stress and bending stress to facilitate the calculation of the maximum cross-sectional stress value and the occurrence position.

[0030] Furthermore, in the fourth step, a threshold value of the error between the monitored data and the soft measurement data is set. When the threshold value is exceeded, the soft measurement model of the hydraulic system is corrected online based on the updated monitoring data. Since the actuator correction process is relatively complicated, a combination of offline correction and online correction is adopted. The offline correction performs parameter identification based on the monitoring data, updates the actuator mechanism model and the actuator soft measurement model, and the online correction method is to establish a data-driven model based on the error between the monitoring data and the soft measurement data, and combine it with the original soft measurement model to form a modified soft measurement model.

[0031] Furthermore, in the third and fourth steps, according to the functional characteristics and application of the typhoon protection suspension system, the process of establishing the soft measurement model of the typhoon protection suspension system is further divided into four stages: the no-load test stage is used to unify the readings of different types of sensors; the constant load test stage is used to verify the soft measurement method; the variable load test / application in passive mode and the variable load test / application in active mode are used to establish and correct the soft measurement model online.

[0032] Due to the adoption of the above technical scheme, the present invention has the following advantages: 1. By combining the mechanism model with the data-driven model, the modeling difficulty of the soft measurement model based on the mechanism model is reduced, and the interpretability of the soft measurement model based on the data-driven model is improved, and finally high-precision soft measurement of difficult-to-monitor or unmonitored positions is achieved, providing a data source for the controller and life cycle management of the typhoon protection suspension system, and improving the fatigue analysis accuracy of the overall structure; 2. Through the soft measurement model correction method combining online and offline, the deep correction of the mechanism model and the data-driven model is achieved, and the measurement accuracy of the soft measurement model is further improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 Flowchart of the soft-sensing method for typhoon-proof suspension system driven by a hybrid mechanism and data.

[0034] Figure 2 This is the structural diagram of the soft-sensing model of the typhoon-proof suspension system driven by a hybrid of mechanism and data.

[0035] Figure 3 Establishment and testing process of soft sensor model for typhoon protection suspension system.

[0036] Figure 4 Schematic diagram of the typhoon protection suspension system principle and monitoring layout.

[0037] Figure 5 Schematic diagram of the strain gauge arrangement on the bottom side of the suspension nipple.

[0038] In the figure, 1-1. Suspension pup joint, 1-2. Hydraulic cylinder, 1-3. Stabilizer, 1-4. Articulated joint, 1-5. Riser system, 2-1. Check valve, 2-2. Low-pressure accumulator, 2-3. Variable pump, 2-4. Overflow valve, 2-5. Oil tank, 2-6. Filter, 2-7. Plug-in throttle valve, 2-8. Check valve, 2-9. High-pressure accumulator, 3-1. Pressure sensor, 3-2. Strain gauge group, 3-3. Strain gauge group, 3-4. Strain gauge group, 3-5. Temperature sensor, 3-6. Pressure sensor, 3-7. Temperature sensor, 3-8. Displacement sensor, 3-9. Pressure sensor, 3-10. Attitude sensor, 3-11. Pressure sensor, 3-12. Load sensor, 3-13. Displacement sensor, 4-1. Drilling platform, 5-1. Suspension pup joint section, 5-2. Strain gauge. DETAILED DESCRIPTION

[0039] 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. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of the present invention.

[0040] like Figure 1 As shown, the present invention proposes a soft-sensing method for typhoon-proof suspension system driven by a hybrid mechanism and data, and the implementation method is as follows:

[0041] (1) Determine the auxiliary variables of the typhoon protection suspension system

[0042] like Figure 4 As shown, the lower end of the suspension pup section 1-1 is connected to the articulated joint 1-4 and to the watertight pipe system 1-5, the middle part of the suspension pup section 1-1 is connected to the stabilizer 1-3 and in contact with the drilling platform 4-1, the top of the suspension pup section 1-1 is connected to the hydraulic cylinder 1-2, and the hydraulic cylinder 1-2 is connected to the drilling platform 4-1; the low-pressure accumulator 2-2 is connected to the rod chamber of the hydraulic cylinder 1-2 via the one-way valve 2-1, and the oil tank 2-5 provides hydraulic oil to the low-pressure accumulator 2-2 via the variable pump 2-3; the high-pressure accumulator 2-9 is connected to the rodless chamber of the hydraulic cylinder 1-2 via the one-way valve 2-8 and the cartridge throttle valve 2-7, the high-pressure accumulator 2-9 is connected to the oil tank via the oil filter 2-6, the rodless chamber and the rod chamber of the hydraulic cylinder 1-2 are connected via the one-way valve, and the two groups of one-way valves 2-8 and the cartridge throttle valve 2-7 form two hydraulic circuits in which the hydraulic oil flows to the left and flows to the right. According to the characteristics of the typhoon protection suspension system, the dominant variables are determined to be the pressure drop of the cartridge throttle valve 2-7, the lateral force at the top of the hydraulic cylinder 1-2, the friction resistance of the hydraulic cylinder 1-2, the stress of the suspension short section 1-1, and the friction resistance at the position of the stabilizer 1-3; the auxiliary variables are the oil volume of the high-pressure accumulator 2-9, the front and rear pressure difference of the cartridge throttle valve 2-7, the pressure of the rodless cavity and the rod cavity of the hydraulic cylinder 1-2, the top load and piston rod extension of the hydraulic cylinder 1-2, the bottom stress of the suspension short section 1-1, the oil temperature of some positions of the hydraulic system, and the movement posture of the drilling platform 4-1 and the actuator.

[0043] (2) Establish a monitoring database for typhoon protection suspension systems

[0044] According to the types of the selected dominant variables and auxiliary variables, determine the type of sensor to be selected and the location of the sensor layout, such as Figure 4 and Figure 5As shown. The pressure sensor 3-1 is arranged in the pipeline between the one-way valve 2-1 and the hydraulic cylinder 1-2, the pressure sensors 3-6 and 3-5 are arranged between the one-way valve 2-8 and the plug-in throttle valve 2-7, and the temperature sensor 3-7 is arranged between the high-pressure accumulator 2-9 and the one-way valve 2-8. The displacement sensor 3-8 is installed on the high-pressure accumulator 2-9. The pressure sensor 3-9 and the pressure sensor 3-11 are arranged near the hydraulic cylinder 1-2 and the plug-in throttle valve 2-7 respectively. The load sensor 3-12 and the displacement sensor 3-13 are installed at the contact position between the hydraulic cylinder 1-2 and the suspension short section 1-1. The temperature sensor 3-5 is arranged inside the oil tank 2-5. The attitude sensor 3-10 is arranged at the center of gravity of the drilling platform 4-1. Finally, the strain gauge group 3-2 and the strain gauge group 3-3 are arranged separately at the bottom of the suspension short section 1-1, and the strain gauge group 3-4 is arranged at the top of the watertight pipe system 1-5. The monitoring data is obtained by arranging sensors, in which the strain gauge group is used to monitor the bottom stress of the suspension short section 1-1. The strain gauge group is evenly arranged around the bottom side of the suspension short section 1-1 to solve the axial stress and bending stress at the monitoring position of the suspension short section 1-1. The equation group is in the form of

[0045]

[0046] Where M σ is the maximum bending moment of the suspension short section 5-1, R is the radius of the suspension short section 5-1, σ a is the axial stress, σ1-σ4 are the monitored stresses, and I is the section moment of inertia.

[0047] Based on multi-source information fusion technology, feature recognition, signal identification, signal processing estimation and multi-source synthesis processing are performed on the collected sensor data to improve the authenticity and quality of the monitoring data; the spatial motion posture of the actuator is obtained by using the platform motion posture and the extension of the piston rod of the hydraulic cylinder 1-2, and finally a monitoring data history library is established.

[0048] (3) Establishing a soft-sensing model for typhoon-proof suspension system

[0049] like Figure 2 As shown, the hydraulic system mechanism model is established, and the mechanism model of the plug-in throttle valve 2-7 is

[0050]

[0051] In the formula, C d is the flow coefficient, A T is the flow cross-sectional area of ​​the valve port of the cartridge throttle valve 2-7, Δp is the pressure difference before and after the cartridge throttle valve 2-7, ρ is the hydraulic oil density, λ is the opening of the cartridge throttle valve 2-7, and A is the maximum flow cross-sectional area of ​​the cartridge throttle valve 2-7.

[0052] The pipeline pressure loss model is:

[0053]

[0054] Where v is the average flow velocity of the hydraulic oil in the pipe, and ∑ξ is the comprehensive resistance coefficient.

[0055] The expansion and compression model of hydraulic oil and pipeline is:

[0056]

[0057] In the formula, β g is the comprehensive expansion coefficient, V g0 is the initial volume, ΔV g is the volume increment, Δp g is the pressure increment, and the expansion and compression of the hydraulic oil and pipeline are taken into account here.

[0058] The friction model of hydraulic cylinder 1-2 is:

[0059] f=F p -F n

[0060] In the formula, F p 、F n are respectively the hydraulic oil thrust and the piston rod top force. Due to different hydraulic circuits, F p The expression of is different; further use the solved suspension short section 1-1 to monitor the axial stress F at the position la The friction force f of the centralizer 1-3 is calculated by the force acting on the piston rod top 1a :

[0061] f la =F n -F la

[0062] Further, the intermediate parameters of the hydraulic system mechanism model are obtained according to the monitoring data, and a data-driven model of the hydraulic system monitoring data and the intermediate parameters is formed based on the reinforcement learning algorithm. The hydraulic system soft measurement model of the anti-typhoon suspension system is formed by integrating the mechanism model and the data-driven model. It is used to monitor the pressure drop of the plug-in throttle valve 2-7 and the pressure loss parameters of the pipeline, and a friction soft measurement model of the hydraulic cylinder 1-2 is formed based on the same method.

[0063] A finite element analysis model of the actuator and the riser system 1-5 consisting of the suspension nipple 1-1, the centralizer 1-2, the articulated joint 1-3 is established:

[0064]

[0065] Wherein, equation ① and ② are local finite element and global finite element models respectively. K is the stiffness matrix, M is the mass matrix, C is the damping matrix, F is the load vector, δ is the displacement vector; the global finite element analysis results provide input parameters for the local finite element analysis.

[0066] Carry out simulation analysis of the mechanical characteristics of the actuator to obtain the mechanical response parameters, including the top lateral force of the suspension short section 1-1 and the stress at the lateral constraint position of the suspension short section 1-1. Based on the machine learning algorithm, the mechanical simulation input parameters, the characteristic parameters of the actuator and the mechanical response parameters are trained to obtain the data-driven model of the simulation data. The data-driven model and the monitoring data are processed in a weighted fusion manner to obtain the soft measurement model of the actuator of the typhoon protection suspension system.

[0067] (4) Correction of the soft-sensing model of the typhoon protection suspension system

[0068] The error threshold between the monitored data and the soft measurement data is set to 5%. When the threshold is exceeded, the soft measurement model of the hydraulic system is corrected online based on the updated monitoring data. A combination of offline correction and online correction is adopted. The offline correction performs parameter identification based on the monitoring data and updates the mechanism model and the soft measurement model of the actuator. The online correction method is to establish a data-driven model based on the error between the monitoring data and the soft measurement data, and combine it with the original soft measurement model to form a modified soft measurement model. The soft measurement model of the actuator is corrected based on the monitoring data of multiple groups of sensors at the bottom of the suspension pup joint 1-1.

[0069] like Figure 3 As shown in the figure, according to the functional characteristics and application of the typhoon protection suspension system, the process of establishing the soft measurement model of the typhoon protection suspension system is further divided into four stages: the no-load test stage is used to unify the readings of different types of sensors; the constant load test stage is used to verify the soft measurement method; the variable load test / application in passive mode and the variable load test / application in active mode are used to establish the soft measurement model online.

[0070] Obviously, those skilled in the art should understand that the steps in the flowcharts of the present invention and the above embodiments are displayed in sequence according to the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise clearly stated herein, there is no strict order restriction for the execution of these steps, and they can be executed in other orders.

[0071] The above are only specific steps of the present invention and do not constitute any limitation to the protection scope of the present invention; it can be extended to the relevant fields of marine oil and gas development technology, and any technical solutions formed by equivalent transformation or equivalent replacement shall fall within the protection scope of the present invention.

[0072] The present invention does not elaborate on some of the well-known technologies that belong to those skilled in the art.

Claims

1. A soft-sensing method for typhoon-proof suspension system driven by mechanism and data hybrid, characterized in that: The following steps are included: Step 1: Identify the structural characteristics of the typhoon-proof suspension system, including the components and hydraulic circuits of the hydraulic system, the components and matching relationships of the actuators, and the parameter types of all components; determine the dominant variables to be monitored based on the structural characteristics of the typhoon-proof suspension system, and further determine the auxiliary variables; establish the final auxiliary variables based on correlation analysis; Step 2: Determine the type of sensor and sensor layout location based on the selected dominant variable and auxiliary variable types; collect real-time monitoring information from sensors, process data based on multi-source information fusion technology, and establish a monitoring data history database; Step 3: Establish a mechanism model of the typhoon-proof suspension system based on hydraulic fluid mechanics and pipe column structural mechanics, including a hydraulic system mechanism model and an actuator mechanism model; establish a data-driven model based on a machine learning algorithm; integrate monitoring data and the mechanism model to carry out model parameter identification and data-driven model training; integrate the established mechanism model and data-driven model to propose a soft-sensing model of the typhoon-proof suspension system driven by a combination of mechanism and data; Step 4: Use the soft measurement model of the typhoon protection suspension system to obtain real-time soft measurement data; The soft measurement data is compared with the monitoring data at regular intervals, and the updated monitoring data history library is used to realize the online and offline correction of the soft measurement model of the typhoon protection suspension system.

2. According to claim 1, the soft-sensing method for typhoon-proof suspension system driven by a hybrid mechanism and data is characterized in that: In the first step, the hydraulic system includes a high-pressure accumulator, a cartridge throttle valve, a one-way valve, a low-pressure accumulator, an oil tank, a variable pump, and a relief valve. The actuator includes a hydraulic cylinder, a suspension pup joint, a stabilizer and an articulated joint; the dominant variables are the pressure drop of the cartridge throttle valve, the lateral force at the top of the hydraulic cylinder, the friction resistance of the hydraulic cylinder, the stress of the suspension pup joint, and the friction resistance at the position of the stabilizer; the auxiliary variables are the oil volume of the high-pressure accumulator, the pressure difference between the inlet and outlet of the cartridge throttle valve, the pressure of the rodless chamber and the rod chamber of the hydraulic cylinder, the load on the top of the hydraulic cylinder and the extension of the piston rod, the stress at the bottom of the suspension pup joint, the oil temperature of some positions of the hydraulic system, and the movement posture of the drilling platform and the actuator; there is an interactive substitution relationship between the dominant variables and the auxiliary variables to solve the complex correlation between the variables.

3. The soft-sensing method for typhoon-proof suspension system driven by mechanism and data hybrid according to claim 1 is characterized by: In the second step, the sensors are divided into pressure sensors, load sensors, temperature sensors, attitude sensors, displacement sensors, and strain gauges according to the types of variables; the strain gauge group is used to monitor the bottom stress of the suspension pup joint, and the strain gauges are evenly arranged around the bottom side of the suspension pup joint, with four or eight groups of them, which are used to solve the axial stress and bending stress at the monitoring position of the suspension pup joint; based on multi-source information processing technology, feature recognition, signal identification, signal processing estimation and multi-source synthesis processing are performed on the original signal to improve the authenticity and quality of the monitoring data; the spatial motion posture of the actuator is obtained by using the platform motion posture and the extension of the hydraulic cylinder piston rod, including the inclination angle of the hydraulic cylinder and the extension of the piston rod, and the motion position of the stabilizer.

4. The soft-sensing method for typhoon-proof suspension system driven by mechanism and data hybrid according to claim 1 is characterized by: In the third step, the mechanism model of the hydraulic system includes a plug-in throttle valve model, a pipeline pressure loss model, an expansion and compression model of hydraulic oil and pipelines, and a hydraulic cylinder friction model: The plug-in throttle valve model is In the formula, C d is the flow coefficient, A T is the flow cross-sectional area of ​​the cartridge throttle valve port, Δp is the pressure difference before and after the cartridge throttle valve, ρ is the hydraulic oil density, λ is the opening of the cartridge throttle valve, and A is the maximum flow cross-sectional area of ​​the cartridge throttle valve; The pipeline pressure loss model is: In the formula, v is the average flow velocity of the hydraulic oil in the pipe, ∑ξ is the comprehensive resistance coefficient; The expansion and compression model of hydraulic oil and pipeline is: In the formula, β g is the comprehensive expansion coefficient, V g0 is the initial volume, ΔV g is the volume increment, Δp g is the pressure increment, where the expansion and compression of the hydraulic oil and pipeline are considered together; The friction model of the hydraulic cylinder is: f=F p -F n In the formula, F p 、F n are respectively the hydraulic oil thrust and the piston rod top force. Due to different hydraulic circuits, F p The expression of is different; further use the solved suspension short section monitoring position axial stress F la The friction force f of the centralizer can be calculated by the force acting on the piston rod top 1a for f la =F n -F la According to the mechanism model of the actuator, the local finite element model and the global finite element model of the actuator structure mechanics analysis are established as follows: Wherein, equation ① and ② are local finite element and global finite element models respectively. K is the stiffness matrix, M is the mass matrix, C is the damping matrix, F is the load vector, δ is the displacement vector; the global finite element analysis results provide input parameters for the local finite element analysis; The intermediate parameters of the hydraulic system mechanism model are obtained according to the monitoring data, and a data-driven model of the hydraulic system monitoring data and intermediate parameters is formed based on the reinforcement learning algorithm. The hydraulic system soft measurement model of the typhoon-proof suspension system is formed by integrating the mechanism model and the data-driven model, which is used to monitor the pressure drop of the plug-in throttle valve and the pressure loss parameters of the pipeline. A hydraulic cylinder friction soft measurement model is established based on the same method. The mechanical characteristics simulation analysis of the actuator is carried out to obtain the mechanical response parameters, including the top lateral force of the suspension short section and the stress at the lateral constraint position of the suspension short section. The mechanical simulation input parameters, the characteristic parameters of the actuator and the mechanical response parameters are trained based on the machine learning algorithm to obtain the data-driven model of the simulation data. The data-driven model and the monitoring data are processed by weighted fusion to establish a soft measurement model of the actuator of the typhoon-proof suspension system. The response parameters of the unmeasurable position are inferred from the response parameters of the measurable position of the actuator. When the data-driven model of the actuator is trained, the cross-sectional stress in the output parameters can be split into axial stress and bending stress to facilitate the calculation of the maximum cross-sectional stress value and the occurrence position.

5. The soft-sensing method for typhoon-proof suspension system driven by mechanism and data hybrid according to claim 1 is characterized by: In the fourth step, a threshold value of the error between the monitored data and the soft measurement data is set. When the threshold value is exceeded, the soft measurement model of the hydraulic system is corrected online based on the updated monitoring data. Since the actuator correction process is relatively complicated, a combination of offline correction and online correction is adopted. The offline correction performs parameter identification based on the monitoring data, updates the actuator mechanism model and the actuator soft measurement model, and the online correction method is to establish a data-driven model based on the error between the monitoring data and the soft measurement data, and combine it with the original soft measurement model to form a modified soft measurement model.

6. The soft-sensing method for typhoon-proof suspension system driven by mechanism and data hybrid according to claim 1 is characterized by: In the third and fourth steps, according to the functional characteristics and application of the typhoon protection suspension system, the process of establishing the soft measurement model of the typhoon protection suspension system is further divided into four stages: the no-load test stage is used to unify the readings of different types of sensors; the constant load test stage is used to verify the soft measurement method; the variable load test / application in passive mode and the variable load test / application in active mode are used to establish and correct the soft measurement model online.

Images