Automatic air cushion bin liquid level control method and system, electronic equipment and storage medium

Abstract

The invention discloses an air cushion bin liquid level automatic control method and system, electronic equipment and a storage medium, the method is based on data driving modeling, dynamic characteristics contained in slurry shield tunneling air cushion bin liquid level control related data are fully excavated, a constructed air cushion bin liquid level dynamic characteristic model is combined with a BP network model and an ARX model, and the air cushion bin liquid level automatic control method and system are established. A BP neural network is used for fitting a regression coefficient matrix of an ARX model, so that the model has very high global nonlinear description capability and is more fit with the actual working characteristics of the liquid level of the air cushion bin of the shield tunneling machine; moreover, a slurry feeding and discharging flow difference value prediction controller is constructed based on the air cushion bin liquid level dynamic characteristic model, the optimal solution of the slurry feeding and discharging flow difference value can be obtained by carrying out optimization solution on a prediction control objective function after real-time tunneling data are collected, and the optimal solution can enable the air cushion bin liquid level to be stabilized near a reference value. The adjusting range of the slurry feeding and discharging pump is small, the whole circulation system is stable, and the pressure fluctuation of the muddy water bin is small.

Description

Technical Field

[0001] The present invention relates to the technical field of slurry shield tunneling machines, and in particular, to a method and system for automatically controlling the liquid level of an air cushion chamber, an electronic device, and a computer-readable storage medium. Background Art

[0002] A shield tunneling machine is a special construction equipment for tunnel boring projects, which has functions such as cutter head excavation, hydraulic propulsion, segment assembly, muck transportation, and guidance measurement. It is a complex system integrating advanced technologies such as machinery, electricity, hydraulics, and optics. Among them, slurry shield tunneling machines account for a large proportion in the entire shield market. The excavation chamber of a slurry shield tunneling machine is divided into a slurry chamber and an air cushion chamber. The slurry chamber is filled with slurry. The lower part of the air cushion chamber is slurry, and the upper part is compressed air. During normal tunneling of the slurry shield, the slurry liquid level in the air cushion chamber needs to be ensured to be above and below the central axis. At present, the liquid level control of the air cushion chamber mostly adopts manual control, with a dedicated person monitoring the change of the liquid level in the air cushion chamber on the slurry operation panel interface throughout the day and manually adjusting it. This manual control method has high requirements for the operator's level. If the operation is improper, a large range of fluctuations in the liquid level of the air cushion chamber will have an adverse impact on the pressure of the slurry chamber and the automatic pressure maintenance of the air cushion chamber, and more seriously, it will cause ground settlement. Therefore, the automatic control of the liquid level of the air cushion chamber of a slurry shield tunneling machine has important practical significance. The Chinese patent application for invention with the publication number CN114810104A discloses a liquid level adjustment system and method for the air cushion chamber of a slurry-gas balance shield. It monitors the liquid level of the air cushion chamber through an air cushion chamber liquid level monitoring system, feeds back the liquid level change signal to the slurry circulation system of the shield tunneling machine, and the slurry circulation system adjusts the opening and closing of the ball valve or the flow rate of the slurry pump according to the received signal, so as to realize the function of automatically adjusting the stability of the liquid level of the air cushion chamber. However, the solution of this patent application limits the liquid level of the air cushion chamber between the high liquid level and the low liquid level, resulting in a large adjustment range of the slurry inlet and outlet pumps, poor stability of the entire circulation system, and large pressure fluctuations in the slurry chamber. Summary of the Invention

[0003] The present invention provides a method and system for automatically controlling the liquid level of an air cushion chamber, an electronic device, and a computer-readable storage medium, which can make the liquid level of the air cushion chamber stable near the reference value, with a very small adjustment range of the slurry inlet and outlet pumps, stable operation of the entire circulation system, and very small pressure fluctuations in the slurry chamber.

[0004] According to one aspect of the present invention, there is provided a method for automatically controlling the liquid level of an air cushion chamber, including the following steps:

[0005] Obtain the tunneling data during the tunneling process of the slurry shield tunneling machine and perform normalization processing on it, where the tunneling data includes the liquid level of the air cushion chamber, the difference between the slurry inlet and outlet flow rates, the average propulsion speed, and the air cushion chamber pressure value;

[0006] Construct a dynamic characteristic model of the air cushion bin liquid level based on the BP neural network model and the ARX model, and use the tunneling data after normalization to iteratively train the dynamic characteristic model of the air cushion bin liquid level until the iterative termination condition is met, and save the trained dynamic characteristic model of the air cushion bin liquid level;

[0007] Construct a prediction controller for the difference between the slurry inflow and outflow rates based on the trained dynamic characteristic model of the air cushion bin liquid level. Collect real-time tunneling data, perform normalization processing, and then input it into the prediction controller for the difference between the slurry inflow and outflow rates. Optimize and solve the prediction control objective function of the prediction controller for the difference between the slurry inflow and outflow rates to obtain the optimal solution for the difference between the slurry inflow and outflow rates;

[0008] Control the slurry inflow and outflow rates based on the optimal solution of the difference between the slurry inflow and outflow rates, so that the liquid level of the air cushion bin is stabilized at the reference value.

[0009] Furthermore, control the slurry inflow and outflow rates based on the following formula:

[0010]

[0011] where, u * (t) represents the optimal solution of the difference between the slurry inflow and outflow rates, represents the actual value of the slurry inflow rate in the current time, represents the actual value of the slurry outflow rate in the current time, represents the set value of the slurry outflow rate in the current time, ΔF(t) represents the increased value of the frequency setting of the inverter in the slurry outflow pipeline at the current time, k p 、k i 、k d respectively represent the proportional factor, integral factor, and differential factor of the slurry outflow pipeline flow controller, er(t) represents the slurry outflow pipeline flow error at the current time, er(t - 1) represents the slurry outflow pipeline flow error at the previous time, er(t - 2) represents the slurry outflow pipeline flow error at the time before the previous time, e 0 represents the preset threshold.

[0012] Furthermore, the dynamic characteristic model of the air cushion bin liquid level can be expressed as:

[0013]

[0014] where, y(t) represents the model output, that is, the liquid level of the air cushion bin, t represents the current time; p represents the order of the model output; q represents the order of the model input; s represents the order of the measurable disturbance of the model; f represents the delay of the model input; g represents the delay of the measurable disturbance of the model; u represents the difference between the slurry inflow and outflow rates; d represents the measurable disturbance vector of the model, d 1 represents the average propulsion speed, d 2 represents the air cushion bin pressure; T represents the transpose; ε(t) represents the Gaussian white noise signal; a0,t-1 The coefficient matrix representing the model bias, where a 11，0,t-1 is an element in a 0,t-1 ; a i,t-1 The coefficient matrix representing the model output, where a 11,i,t-1 is an element in a i,t-1 ; b j,t-1 The coefficient matrix representing the model input, where b 11,j,t-1 is an element in b j,t-1 ; c k,t-1 The coefficient matrix representing the measurable disturbance of the model, where c 11,k,t-1 , c 12,k,t-1 is an element in c k,t-1 ; a 0,t-1 , a i,t-1 , b j,t-1 , c k,t-1 are obtained by calculating through the BP neural network with the state vector; x(t - 1) represents the state vector, and n w represents the order of the state vector and serves as the input to the BP neural network; represents the n x -th node in the state vector; represents the weight from the l-th node in the hidden layer of the BP network to the n x -th node in the input layer; h l represents the output of the l-th node in the hidden layer of the BP network; b l represents the bias of the l-th node in the hidden layer of the BP network; n l represents the number of nodes in the hidden layer of the BP network; represents the activation function; n o represents the number of nodes in the output layer of the BP network; W m_l represents the weight from the m-th node in the output layer of the BP network to the l-th node in the hidden layer; O m represents the output of the m-th node in the output layer of the BP network; b m represents the bias of the m-th node in the output layer of the BP network; O 1 represents the output of the first node in the output layer of the BP network; O 2 represents the output of the second node in the output layer of the BP network; O 1+p represents the output of the (1 + p)-th node in the output layer of the BP network; O 1+p+1 represents the output of the (1 + p + 1)-th node in the output layer of the BP network; O 1+p+q represents the output of the (1 + p + q)-th node in the output layer of the BP network; O 1+p+q+1 represents the output of the (1 + p + q + 1)-th node in the output layer of the BP network; O 1+p+q+2s represents the output of the (1 + p + q + 2s)-th node in the output layer of the BP network; θ t-1 represents the state-dependent coefficient vector.

[0015] Further, the state - space model of the inlet - outlet pulp flow difference prediction controller is as follows:

[0016]

[0017] Among them, y(t) represents the output variable, x(t) represents the state variable, A t , B t , Φ t , C are all coefficient matrices, a 1,t represents the coefficient matrix when the model output order at time t is 1, a 2,t represents the coefficient matrix when the model output order at time t is 2, represents the coefficient matrix when the model output order at time t is k n - 1, represents the coefficient matrix when the model output order at time t is k n , b 1,t represents the coefficient matrix when the model input order at time t is 1, b 2,t represents the coefficient matrix when the model input order at time t is 2, represents the coefficient matrix when the model input order at time t is k n - 1, represents the coefficient matrix when the model input order at time t is k n , φ(t) represents the sum of the bias of the model at time t and the interferences of each order model, Ξ(t + 1) represents the matrix of white noise at time t + 1, ε(t + 1) represents the white noise at time t + 1, and T represents the transpose of the vector.

[0018] Further, the predictive control objective function is:

[0019]

[0020] Among them, minJ represents the predictive control objective function, represents 's 2 - norm, represents 's 2 - norm, represents the future N y - step forward prediction vector of the air - cushion bin liquid level y obtained based on the state - space model, y r represents the desired vector of the air - cushion bin liquid level, y min represents the lower limit of the air - cushion bin liquid level, y max represents the upper limit of the air - cushion bin liquid level, u min represents the lower limit of the inlet - outlet pulp flow difference, u max represents the upper limit of the inlet - outlet pulp flow difference, Denote the control vector obtained by solving at time t Increment vector of, Δu min Denote the lower limit of the increment of the difference between the slurry inflow and outflow, Δu max Denote the upper limit of the increment of the difference between the slurry inflow and outflow, Is the control vector obtained by solving at time t, R 1 (t) represents the weighting coefficient.

[0021] Furthermore, the weighting coefficient is adaptively adjusted based on the following formula:

[0022]

[0023] Where, β represents the adjustment coefficient, N g Denote the observation window length, R min Denote the minimum value of the weighting coefficient, R max Denote the maximum value of the weighting coefficient, y(t-i 3 ) represents the liquid level of the air cushion bin at time t-i 3 Time, y(t-i 3 -1) represents the liquid level of the air cushion bin at time t-i 3 -1 time.

[0024] Furthermore, in the process of obtaining the tunneling data, on the premise of ensuring the safe state of the liquid level of the air cushion bin, the difference between the slurry inflow and outflow is adjusted to make the liquid level of the air cushion bin change within a large range.

[0025] In addition, the present invention also provides an automatic control system for the liquid level of the air cushion bin, including:

[0026] Tunneling data acquisition module, used to acquire the tunneling data during the tunneling process of the slurry balance shield machine and perform normalization processing on it, where the tunneling data includes the liquid level of the air cushion bin, the difference between the slurry inflow and outflow, the average propulsion speed, and the air cushion bin pressure value;

[0027] Model construction and training module, used to construct a dynamic characteristic model of the air cushion bin liquid level based on the BP network model and the ARX model, and use the normalized tunneling data to iteratively train the dynamic characteristic model of the air cushion bin liquid level until the iterative termination condition is met, and save the trained dynamic characteristic model of the air cushion bin liquid level;

[0028] Slurry inflow and outflow difference prediction control module, used to construct a slurry inflow and outflow difference prediction controller based on the trained dynamic characteristic model of the air cushion bin liquid level, collect real-time tunneling data, perform normalization processing, and input it into the slurry inflow and outflow difference prediction controller, and optimize and solve the prediction control objective function of the slurry inflow and outflow difference prediction controller to obtain the optimal solution of the slurry inflow and outflow difference;

[0029] The inflow and outflow slurry flow automatic control module is used to control the inflow and outflow slurry flow based on the optimal solution of the difference between the inflow and outflow slurry flow, so that the liquid level of the air cushion bin is stabilized at the reference value.

[0030] In addition, the present invention also provides an electronic device, including a processor and a memory. A computer program is stored in the memory, and the processor is used to execute the steps of the method as described above by calling the computer program stored in the memory.

[0031] In addition, the present invention also provides a computer-readable storage medium for storing a computer program for automatically controlling the liquid level of the air cushion bin. When the computer program runs on a computer, it executes the steps of the method as described above.

[0032] The present invention has the following beneficial effects:

[0033] The method for automatically controlling the liquid level of the air cushion bin of the present invention is based on data-driven modeling, fully excavating the dynamic characteristics contained in the data related to the liquid level control of the air cushion bin during the tunneling of the slurry shield. Moreover, the constructed dynamic characteristic model of the air cushion bin liquid level combines the BP network model and the ARX model, and uses the BP neural network to fit the regression coefficient matrix of the autoregressive ARX model with exogenous variables, making the model have a strong global non-linear description ability and being more in line with the actual working characteristics of the liquid level of the air cushion bin of the shield machine. And, a prediction controller for the difference between the inflow and outflow slurry flow is constructed based on the dynamic characteristic model of the air cushion bin liquid level. After collecting real-time tunneling data and optimizing and solving the prediction control objective function, the optimal solution of the difference between the inflow and outflow slurry flow can be obtained. This optimal solution can make the liquid level of the air cushion bin stable near the reference value, the adjustment range of the inflow and outflow slurry pumps is very small, the entire circulation system is stable, and the pressure fluctuation in the slurry chamber is very small.

[0034] In addition, the automatic control system for the liquid level of the air cushion bin of the present invention also has the above advantages.

[0035] In addition to the purposes, features and advantages described above, the present invention has other purposes, features and advantages. The present invention will be further described in detail below with reference to the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The drawings constituting a part of this application are used to provide a further understanding of the present invention. The schematic embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation to the present invention. In the drawings:

[0037] Figure 1 is a schematic flow chart of the method for automatically controlling the liquid level of the air cushion bin in the preferred embodiment of the present application;

[0038] Figure 2 is a schematic network structure diagram of the dynamic characteristic model of the air cushion bin liquid level in the preferred embodiment of the present application;

[0039] Figure 3 is Figure 1 a schematic diagram of the sub - process of step S2 in

[0040] Figure 4 is Figure 1 another schematic diagram of the sub - process of step S2 in

[0041] Figure 5 a schematic diagram of the module structure of the automatic control system for the liquid level of the air - cushion bin in another embodiment of the present application. Specific embodiments

[0042] It should be noted that, without conflict, the embodiments in the present application and the features in the embodiments can be combined with each other. The present application will be described in detail below with reference to the drawings and in combination with the embodiments.

[0043] Refer to Figure 1 , a preferred embodiment of the present application provides an automatic control method for the liquid level of the air - cushion bin, including the following:

[0044] Step S1: Obtain the tunneling data during the tunneling process of the slurry - balanced shield machine and perform normalization processing on it. Among them, the tunneling data includes the liquid level of the air - cushion bin, the difference between the inflow and outflow slurry flow rates, the average propulsion speed, and the air - cushion bin pressure value;

[0045] Step S2: Construct a dynamic characteristic model of the liquid level of the air - cushion bin based on the BP neural network model and the ARX model, and use the normalized tunneling data to iteratively train the dynamic characteristic model of the liquid level of the air - cushion bin until the iterative termination condition is met, and save the trained dynamic characteristic model of the liquid level of the air - cushion bin;

[0046] Step S3: Construct a predictive controller for the difference between the inflow and outflow slurry flow rates based on the trained dynamic characteristic model of the liquid level of the air - cushion bin, collect the real - time tunneling data, perform normalization processing, and input it into the predictive controller for the difference between the inflow and outflow slurry flow rates, and optimize and solve the predictive control objective function of the predictive controller for the difference between the inflow and outflow slurry flow rates to obtain the optimal solution of the difference between the inflow and outflow slurry flow rates;

[0047] Step S4: Control the inflow and outflow slurry flow rates based on the optimal solution of the difference between the inflow and outflow slurry flow rates to keep the liquid level of the air - cushion bin stable at the reference value.

[0048] It can be understood that for the automatic control method of the air-cushion chamber liquid level in this embodiment, the tunneling data during the tunneling process of the slurry shield is first collected, and then a dynamic characteristic model of the air-cushion chamber liquid level is constructed based on the BP network model and the ARX model. The tunneling data is used to iteratively train the dynamic characteristic model of the air-cushion chamber liquid level. Then, a prediction controller for the difference in the inflow and outflow slurry flow rates is constructed based on the trained dynamic characteristic model of the air-cushion chamber liquid level. After collecting the real-time tunneling data, the prediction control objective function of the prediction controller for the difference in the inflow and outflow slurry flow rates can be optimized to obtain the optimal solution for the difference in the inflow and outflow slurry flow rates. Finally, based on the optimal solution for the difference in the inflow and outflow slurry flow rates, the inflow and outflow slurry flow rates are controlled to keep the air-cushion chamber liquid level stable near the reference value. The present invention is based on data-driven modeling, fully excavating the dynamic characteristics contained in the data related to the control of the air-cushion chamber liquid level during the tunneling of the slurry shield. Moreover, the constructed dynamic characteristic model of the air-cushion chamber liquid level combines the BP network model and the ARX model, using the BP neural network to fit the regression coefficient matrix of the autoregressive with exogenous inputs (ARX) model, making the model have a strong global non-linear description ability and being more in line with the actual working characteristics of the air-cushion chamber liquid level of the shield machine. In addition, a prediction controller for the difference in the inflow and outflow slurry flow rates is constructed based on the dynamic characteristic model of the air-cushion chamber liquid level. After collecting the real-time tunneling data and optimizing the prediction control objective function, the optimal solution for the difference in the inflow and outflow slurry flow rates can be obtained. This optimal solution can keep the air-cushion chamber liquid level stable near the reference value, with a very small adjustment range for the inflow and outflow slurry pumps, a stable entire circulation system, and very small fluctuations in the pressure of the slurry chamber.

[0049] It can be understood that in step S1, the tunneling data during the tunneling process of the slurry balance shield machine is collected. The tunneling data includes the air-cushion chamber liquid level, the difference in the inflow and outflow slurry flow rates, the average propulsion speed, and the air-cushion chamber pressure value. Among them, the difference in the inflow and outflow slurry flow rates refers to the difference between the inflow slurry flow rate and the outflow slurry flow rate. Since the slurry balance shield machine has propulsion modes, bypass modes, pipeline extension modes, backwashing modes, and weekend modes, only the tunneling data in the propulsion mode of the shield machine needs to be collected as the identification data. The sampling period of the tunneling data is set to 1 second, and the tunneling data is continuously collected until the preset quantity requirement is met. In addition, during the process of obtaining the tunneling data, on the premise of ensuring the safety state of the air-cushion chamber liquid level, the difference in the inflow and outflow slurry flow rates is adjusted to cause large-range changes in the air-cushion chamber liquid level, ensuring that the collected tunneling data completely contains the dynamic characteristics of the air-cushion chamber liquid level, which is beneficial to the control accuracy of the air-cushion chamber liquid level. Then, for the collected tunneling data, the maximum-minimum value method is used for normalization processing, converting all the tunneling data to between [0, 1]. The conversion formula is as follows:

[0050]

[0051] Among them, Y represents the normalized tunneling data, and X represents the tunneling data before normalization, including the liquid level of the air cushion bin, the difference between the slurry inflow and outflow rates, the average propulsion speed, and the air cushion bin pressure value, X max represents the maximum value of the tunneling data, X min represents the minimum value of the tunneling data. Of course, in other embodiments of the present invention, other existing normalization methods can also be used, such as the Z-score normalization method, the maximum value normalization method, etc.

[0052] It can be understood that in step S2, a dynamic characteristic model of the air cushion bin liquid level is constructed based on the BP network model and the ARX model, and the normalized tunneling data is used to iteratively train the dynamic characteristic model of the air cushion bin liquid level until the iterative termination condition is met, and the trained dynamic characteristic model of the air cushion bin liquid level is saved. It can be understood that the network structure of the dynamic characteristic model of the air cushion bin liquid level is as Figure 2 shown. The BP neural network has three layers, namely the input layer, the hidden layer, and the output layer. The model extracts the non-linear features in the time series data of the air cushion bin liquid level of the slurry shield machine through the BP neural network, and uses the BP neural network to fit the regression coefficient matrix of the ARX model, so that the model has a strong global non-linear description ability and is more in line with the actual working characteristics of the air cushion bin liquid level of the shield machine.

[0053] Among them, the dynamic characteristic model of the air cushion bin liquid level can be expressed as:

[0054]

[0055] Among them, y(t) represents the model output, that is, the air cushion bin liquid level, t represents the current moment; p represents the order of the model output; q represents the order of the model input; s represents the order of the measurable disturbance of the model; f represents the delay of the model input; g represents the delay of the measurable disturbance of the model; u represents the difference between the slurry inflow and outflow rates; d represents the measurable disturbance vector of the model, d 1 represents the average propulsion speed, d 2 represents the air cushion bin pressure; T represents the transpose; ε(t) represents the Gaussian white noise signal; a 0,t-1 represents the coefficient matrix of the model bias, where a 11，0,t-1 is an element in a 0,t-1 ; a i,t-1 represents the coefficient matrix of the model output, where a 11,i,t-1 is an element in a i,t-1 ; b j,t-1 represents the coefficient matrix of the model input, where b 11,j,t-1 is an element in b j,t-1 ; c k,t-1 represents the coefficient matrix of the measurable disturbance of the model, where c 11,k,t-1 , c 12,k,t-1 are elements in ck,t-1 The elements in; a 0,t-1 、a i,t-1 、b j,t-1 、c k,t-1 are obtained by calculating the state vector through the BP neural network; x(t - 1) represents the state vector, and n w represents the order of the state vector and serves as the input to the BP neural network; represents the nth x node in the state vector; represents the weight from the lth node in the hidden layer of the BP network to the nth x node in the input layer; h l represents the output of the lth node in the hidden layer of the BP network; b l represents the bias of the lth node in the hidden layer of the BP network; n l represents the number of nodes in the hidden layer of the BP network; represents the activation function; n o represents the number of nodes in the output layer of the BP network; W m_l represents the weight from the mth node in the output layer of the BP network to the lth node in the hidden layer; O m represents the output of the mth node in the output layer of the BP network; b m represents the bias of the mth node in the output layer of the BP network; O 1 represents the output of the first node in the output layer of the BP network; O 2 represents the output of the second node in the output layer of the BP network; O 1+p represents the output of the (1 + p)th node in the output layer of the BP network; O 1+p+1 represents the output of the (1 + p + 1)th node in the output layer of the BP network; O 1+p+q represents the output of the (1 + p + q)th node in the output layer of the BP network; O 1+p+q+1 represents the output of the (1 + p + q + 1)th node in the output layer of the BP network; O 1+p+q+2s represents the output of the (1 + p + q + 2s)th node in the output layer of the BP network; θ t-1 represents the state - dependent coefficient vector.

[0056] It can be understood that after constructing the dynamic characteristic model of the air - cushion bin liquid level, taking the difference between the slurry inflow and outflow as the input variable, taking the average propulsion speed and the air - cushion bin pressure as the measurable disturbances, and taking the air - cushion bin liquid level as the output variable and state variable of the model, the dynamic characteristic model of the air - cushion bin liquid level is iteratively trained. Among them, as Figure 3 shown, the process of iteratively training the dynamic characteristic model of the air - cushion bin liquid level using the normalized tunneling data includes the following content:

[0057] Step S21: Initialize the model structure parameters;

[0058] Step S22: Using the difference between the inlet and discharge slurry flow rates as the input variable, the average propulsion speed and the air cushion bin pressure as measurable disturbances, and the air cushion bin liquid level as the output variable and state variable of the model, construct the input layer data of the BP network model and the ARX model;

[0059] Step S23: Forward calculation;

[0060] Step S24: Backpropagation to update the model parameters;

[0061] Step S25: After changing the model structure parameters, repeat steps S22, S23, and S24, and select the model structure parameters and model parameters with the minimum loss function value as the optimal model.

[0062] Specifically, the input layer data structure of the BP network model can be defined as:

[0063] Input_BP = [y(t - n w ), y(t - n w + 1),..., y(t - 2), y(t - 1)] T

[0064] where y represents the air cushion bin liquid level, and n w represents the order of Input_BP;

[0065] The input layer data structure of the ARX side can be defined as:

[0066] Input_ARX = [1, y(t - 1),..., y(t - p), u(t - f),..., u(t - f - q + 1),

[0067] d(t - g) T ,..., d(t - g - s + 1) T T

[0068] Initialize the model structure parameters p, q, s, f, g, n w and the number of nodes n in the BP hidden layer l .

[0069] Then, according to the above two data structures, construct the input layer data of the BP network model and the ARX model with the normalized air cushion bin liquid level, the difference between the inlet and discharge slurry flow rates, the average propulsion speed, and the air cushion bin pressure value.

[0070] Next, use the data of the Input_BP structure as the input of the BP network model, and through calculation, obtain the state-dependent coefficient vector θ t-1 , that is, the coefficients of the ARX model, and then the predicted output of the model can be obtained as: ​

[0071]

[0072] Among them, represents the predicted value of the liquid level in the air cushion bin.

[0073] Then, construct the loss function e, and continuously update the model parameters through the backpropagation algorithm until the loss function is minimized to obtain a model with higher accuracy. Among them, the expression of the loss function e is:

[0074]

[0075] Among them, e represents the mean square error between the predicted value and the actual value of the liquid level in the air cushion bin, N represents the number of training samples, and y represents the actual value of the liquid level in the air cushion bin.

[0076] Finally, change the model structure parameters p, q, s, f, g, n w and the number of nodes n in the hidden layer of the BP network l , repeat steps S22, S23, and S24. After traversing all the model structure parameters, compare the loss function values e under different model structure parameters, and select the model structure and model parameters with the smallest loss function value as the optimal model.

[0077] Optionally, as Figure 4 shown, the process of iteratively training the dynamic characteristic model of the air cushion bin liquid level using the normalized tunneling data further includes the following content:

[0078] Step S26: Optimize the model parameters using the particle swarm optimization algorithm.

[0079] The specific process includes the following content:

[0080] 1), Initialize the particle swarm, randomly assign an initial position and velocity to each particle, and assign the model parameters saved in step S25 to one of the particles;

[0081] 2), Calculate the cost function E of each particle cost , where the specific cost function is the prior art and the specific expression is not elaborated here;

[0082] 3), Update the velocity and position of each particle, and the update formula is:

[0083] Among them, represents the velocity of the rth particle at the (l d + 1)th iteration, represents the velocity of the rth particle at the l d th iteration, α represents the inertia weight, and β 1, β 2 represents the acceleration constant, rand 1 , rand 2 represent two random numbers between 0 and 1, pbest r represents the best position found so far by the r-th particle, and gbest represents the best position found so far by the entire particle swarm. represents the position of the r-th particle at the d l + 1-th iteration. is the position of the r-th particle at the d l-th iteration;

[0084] 4), Update the individual and global optimal solutions. For each particle, if the current position is better than the best position encountered before, update its individual optimal value. At the same time, find the position with the minimum cost function E cost from all the particles and update this position as the global optimal solution;

[0085] 5), Repeat steps 2) to 4) until the maximum number of iteration settings is reached or the predetermined threshold of the cost function is reached, then the iteration terminates. When the iteration terminates, output the global optimal solution, and save this solution as the optimal solution of the model parameters.

[0086] It can be understood that the present invention further optimizes the model parameters through the global optimization ability of the particle swarm optimization algorithm, further improves the global non-linear description ability of the model, the model has stronger generalization ability, the output of the predictive controller based on this model is more accurate, and it is beneficial to the stable control of the liquid level in the air cushion bin.

[0087] It can be understood that in step S3, a prediction controller for the difference in the inflow and outflow slurry flow rates is constructed based on the trained dynamic characteristic model of the liquid level in the air cushion bin. After collecting the real-time tunneling data and normalizing it, the data is input into the prediction controller for the difference in the inflow and outflow slurry flow rates, and the prediction control objective function of the prediction controller for the difference in the inflow and outflow slurry flow rates is optimized and solved, and then the optimal solution of the difference in the inflow and outflow slurry flow rates can be obtained. Among them, the real-time tunneling data refers to the liquid level in the air cushion bin, the difference in the inflow and outflow slurry flow rates, the average propulsion speed, and the air cushion bin pressure value at the current moment. Through the prediction controller for the difference in the inflow and outflow slurry flow rates, the optimal solution of the difference in the inflow and outflow slurry flow rates at the next moment is predicted and output. At the next moment, the inflow and outflow slurry flow rates are controlled according to this optimal solution, so that the liquid level in the air cushion bin is stabilized near the reference value.

[0088] Specifically, first convert the dynamic characteristic model of the liquid level in the air cushion bin into an intermediate model structure, which can be expressed as:

[0089]

[0090] where k nrepresents the maximum value of the input and output orders of the model. φ(t - 1) represents the sum of the bias of the model at time t - 1 and the disturbances of each order of the model, and ε(t) represents the white noise at time t. In addition, the meanings of the subscripts a and b are the same as those in the expression of the dynamic characteristic model of the air cushion bin liquid level. However, considering the actual change in the subscript range, i 1 and j 1 are used to represent.

[0091] Then, the state variables are defined as follows:

[0092]

[0093] where k 1 represents the order number; x(t) represents the state variable; x 1,t represents the state vector of order 1; represents the state vector of order k 1 ; represents the output coefficient matrix; represents the input coefficient matrix.

[0094] Based on the above-defined state variables and the intermediate model structure, the state-space model of the inlet and outlet slurry flow difference predictor controller can be obtained as:

[0095]

[0096] where y(t) represents the output variable, x(t) represents the state variable, A t , B t , Φ t , C are all coefficient matrices, a 1,t represents the coefficient matrix when the output order of the model at time t is 1, a 2,t represents the coefficient matrix when the output order of the model at time t is 2, represents the coefficient matrix when the output order of the model at time t is k n - 1, represents the coefficient matrix when the output order of the model at time t is k n ; b 1,t represents the coefficient matrix when the input order of the model at time t is 1, b 2,t represents the coefficient matrix when the input order of the model at time t is 2, represents the coefficient matrix when the input order of the model at time t is k n - 1, represents the coefficient matrix when the input order of the model at time t is k nThe coefficient matrix at that time, φ(t) represents the sum of the bias of the model and various orders of model disturbances at time t, Ξ(t + 1) represents the matrix of white noise at time t + 1, ε(t + 1) represents the white noise at time t + 1, and T represents the transpose of the vector.

[0097] Then, determine the coefficient matrices A t , B t , Φ t in the state - space model according to the tunneling data collected at time t, and obtain a locally linearized model based on this state - space model. Then, use the quadratic programming optimization algorithm to optimize the predictive control objective function of the slurry inflow - outflow difference prediction controller online. Among them, the specific quadratic programming optimization algorithm belongs to the prior art and will not be elaborated here. The predictive control objective function is:

[0098]

[0099] Among them, minJ represents the predictive control objective function, represents the 2 - norm of represents the 2 - norm of represents the future N y - step forward prediction vector of the air - cushion bin liquid level y obtained based on the state - space model, y r represents the desired vector of the air - cushion bin liquid level, y min represents the lower limit of the air - cushion bin liquid level, y max represents the upper limit of the air - cushion bin liquid level, u min represents the lower limit of the slurry inflow - outflow difference, u max represents the upper limit of the slurry inflow - outflow difference, represents the control vector solved at time t the increment vector of min represents the lower limit of the increment of the slurry inflow - outflow difference, Δu max represents the upper limit of the increment of the slurry inflow - outflow difference, is the control vector solved at time t, R 1 (t) represents the weighting coefficient. It can be understood that after solving the control vector at time t, the first term u * (t) of

[0100] is the optimal solution of the slurry inflow - outflow difference. Optionally, the weighting coefficient is adaptively adjusted based on the following formula:

[0101]

[0102] Among them, β represents the adjustment coefficient, N g represents the observation window length, Rmin represents the minimum value of the weighting coefficient, R max represents the maximum value of the weighting coefficient, y(t - i 3 ) represents the liquid level of the air cushion bin at time t - i 3 moment, y(t - i 3 - 1) represents the liquid level of the air cushion bin at time t - i 3 - 1 moment.

[0103] It can be understood that the present invention adopts a weighting coefficient adaptive adjustment mechanism to adjust the control intensity, adjusts the weighting coefficient according to the change of the actual output, and increases the weighting coefficient R 1 (t) when the output change speed is large, and conversely decreases the weighting coefficient R 1 (t).

[0104] It can be understood that the inlet and outlet slurry flow difference prediction controller of the present invention has an online rolling optimization function, not only has strong anti-interference ability, can accurately control the liquid level of the air cushion bin to be stable near the reference value, but also can add constraints and adaptively adjust the weighting coefficient through the prediction control objective function, making the liquid level control of the air cushion bin more stable and the control cost smaller.

[0105] It can be understood that after obtaining the optimal solution of the inlet and outlet slurry flow difference in step S3, in step S4, the inlet and outlet slurry flow is controlled based on the following formula:

[0106]

[0107] where u * (t) represents the optimal solution of the inlet and outlet slurry flow difference, represents the actual value of the inlet slurry pipeline flow at the current moment, represents the actual value of the outlet slurry pipeline flow at the current moment, represents the set value of the outlet slurry pipeline flow at the current moment, ΔF(t) represents the increased value of the frequency setting of the outlet slurry pipeline frequency converter at the current moment, k p 、k i 、k d respectively represent the proportional factor, integral factor, and differential factor of the outlet slurry pipeline flow controller, er(t) represents the outlet slurry pipeline flow error at the current moment, er(t - 1) represents the outlet slurry pipeline flow error at the previous moment, er(t - 2) represents the outlet slurry pipeline flow error at the moment before the previous moment, e 0 represents the preset threshold value, which can be set according to actual needs. In addition, if there are several relay stations in the outlet slurry pipeline, the increased value of the frequency setting of the relay station frequency converter increases synchronously by ΔF(t).

[0108] As can be seen from the above formula, the present invention applies all the optimal values of the obtained difference between the inflow and outflow slurry flow rates to the set value of the outflow slurry pipeline flow rate, which can reduce the disturbance to the inflow slurry of the muddy water circulation system and is beneficial to improving the stability of the circulation system.

[0109] As Figure 5 shown, another embodiment of the present invention further provides an automatic control system for the air cushion bin liquid level, including:

[0110] A tunneling data acquisition module, configured to acquire tunneling data during the tunneling process of the slurry balance shield machine and perform normalization processing on it. Among them, the tunneling data includes the air cushion bin liquid level, the difference between the inflow and outflow slurry flow rates, the average propulsion speed, and the air cushion bin pressure value;

[0111] A model construction and training module, configured to construct a dynamic characteristic model of the air cushion bin liquid level based on the BP network model and the ARX model, and use the normalized tunneling data to iteratively train the dynamic characteristic model of the air cushion bin liquid level until the iteration termination condition is met, and save the trained dynamic characteristic model of the air cushion bin liquid level;

[0112] An inflow and outflow slurry flow rate difference prediction control module, configured to construct an inflow and outflow slurry flow rate difference prediction controller based on the trained dynamic characteristic model of the air cushion bin liquid level, collect real-time tunneling data, perform normalization processing on it, and input it into the inflow and outflow slurry flow rate difference prediction controller, and optimize and solve the prediction control objective function of the inflow and outflow slurry flow rate difference prediction controller to obtain the optimal solution of the inflow and outflow slurry flow rate difference;

[0113] An inflow and outflow slurry flow rate automatic control module, configured to control the inflow and outflow slurry flow rate based on the optimal solution of the inflow and outflow slurry flow rate difference, so that the air cushion bin liquid level is stabilized at the reference value.

[0114] It can be understood that for the automatic control system of the air-cushion chamber liquid level in this embodiment, the tunneling data during the tunneling process of the slurry shield is first collected, and then a dynamic characteristic model of the air-cushion chamber liquid level is constructed based on the BP network model and the ARX model. The tunneling data is used to iteratively train the dynamic characteristic model of the air-cushion chamber liquid level. Then, a prediction controller for the difference in inflow and outflow slurry flow rates is constructed based on the trained dynamic characteristic model of the air-cushion chamber liquid level. After collecting the real-time tunneling data, the prediction control objective function of the prediction controller for the difference in inflow and outflow slurry flow rates can be optimized to obtain the optimal solution for the difference in inflow and outflow slurry flow rates. Finally, based on the optimal solution for the difference in inflow and outflow slurry flow rates, the inflow and outflow slurry flow rates are controlled to keep the air-cushion chamber liquid level stable near the reference value. The present invention is based on data-driven modeling, fully excavating the dynamic characteristics contained in the data related to the control of the air-cushion chamber liquid level during the tunneling of the slurry shield. Moreover, the constructed dynamic characteristic model of the air-cushion chamber liquid level combines the BP network model and the ARX model, using the BP neural network to fit the regression coefficient matrix of the autoregressive ARX model with exogenous variables, making the model have a strong global non-linear description ability and being more in line with the actual working characteristics of the air-cushion chamber liquid level of the shield machine. Also, a prediction controller for the difference in inflow and outflow slurry flow rates is constructed based on the dynamic characteristic model of the air-cushion chamber liquid level. After collecting the real-time tunneling data, by optimizing the prediction control objective function, the optimal solution for the difference in inflow and outflow slurry flow rates can be obtained. This optimal solution can keep the air-cushion chamber liquid level stable near the reference value, with a very small adjustment range for the inflow and outflow slurry pumps, a stable entire circulation system, and very small fluctuations in the slurry chamber pressure.

[0115] It can be understood that each module in the embodiment of the system corresponds to each step in the above method embodiment. Therefore, the specific working process of each module will not be elaborated here, and reference can be made to the above method embodiment accordingly.

[0116] In addition, another embodiment of the present invention further provides an electronic device, including a processor and a memory. A computer program is stored in the memory, and the processor is used to execute the steps of the method as described above by calling the computer program stored in the memory.

[0117] In addition, another embodiment of the present invention further provides a computer-readable storage medium for storing a computer program for the automatic control of the air-cushion chamber liquid level. When the computer program runs on a computer, it executes the steps of the method as described above.

[0118] The forms of computer-readable storage media generally include: floppy disks, flexible disks, hard disks, magnetic tapes, any other magnetic media, CD-ROMs, any other optical media, punch cards, paper tapes, any other physical media with a pattern of holes, random access memories (RAMs), programmable read-only memories (PROMs), erasable programmable read-only memories (EPROMs), flash erasable programmable read-only memories (FLASH-EPROMs), any other memory chips or cartridges, or any other media readable by a computer. Instructions can further be transmitted or received by a transmission medium. The term transmission medium can include any tangible or intangible medium that can be used to store, encode, or carry instructions for execution by a machine, and includes digital or analog communication signals or the intangible media that facilitate the communication of the above instructions. Transmission media includes coaxial cables, copper wires, and optical fibers, which include the wires of a bus used to transmit a computer data signal.

[0119] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memories, CD-ROMs, optical memories, etc.) containing computer-usable program code. The solutions in the embodiments of the present application can be implemented using various computer languages. For example, object-oriented programming languages such as Java and interpreted scripting languages such as JavaScript.

[0120] The present application is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of the present application. It should be understood that each flow and / or block in the flowchart and / or block diagram, as well as the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate means for implementing the functions specified in Figure 1 one or more flows or multiple flows and / or blocks Figure 1 one or more blocks or multiple blocks.

[0121] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to operate in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including an instruction means that implements the function specified in one or more of the processes and / or blocks Figure 1 one or more of the processes and / or blocks Figure 1 specified in the block or blocks.

[0122] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the function specified in one or more of the processes and / or blocks Figure 1 one or more of the processes and / or blocks Figure 1 specified in the block or blocks.

[0123] Although the preferred embodiments of the present application have been described, additional changes and modifications can be made by those skilled in the art once they learn of the basic inventive concept. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications falling within the scope of the present application.

[0124] It is obvious that those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the claims of the present application and their equivalent technologies, the present application is also intended to include these modifications and variations.

[0125] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention can have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A method for automatically controlling the liquid level of an air cushion tank, characterized in that: Includes the following: The excavation data of the slurry shield machine during excavation is obtained and normalized, wherein the excavation data includes the air cushion chamber liquid level, the difference between the inlet and outlet slurry flow, the average value of the propulsion speed and the air cushion chamber pressure value; Based on the BP network model and the ARX model, the dynamic characteristic model of the air cushion tank liquid level is constructed, and the normalized excavation data is used to iteratively train the dynamic characteristic model of the air cushion tank liquid level until the iteration termination condition is met, and the trained dynamic characteristic model of the air cushion tank liquid level is saved; Based on the trained air cushion tank liquid level dynamic characteristic model, a prediction controller for the difference between the inlet and outlet flow rates is constructed. Real-time excavation data is collected, normalized, and then input into the prediction controller for the difference between the inlet and outlet flow rates. The prediction control objective function of the prediction controller for the difference between the inlet and outlet flow rates is optimized and solved to obtain the optimal solution for the difference between the inlet and outlet flow rates. The inlet and outlet flow rates are controlled based on the optimal solution of the difference between the inlet and outlet flow rates, so that the liquid level in the air cushion bin is stabilized at the reference value.

2. The air cushion tank liquid level automatic control method according to claim 1, characterized in that: The inlet and outlet flow rates are controlled based on the following formula: Among them, u * (t) represents the optimal solution for the difference between the inlet and outlet flow rates, Indicates the actual flow rate of the slurry inlet pipeline at the current moment. Indicates the actual flow value of the slurry discharge pipeline at the current moment. represents the flow setting value of the slurry discharge pipeline at the current moment, ΔF(t) represents the frequency setting increase value of the slurry discharge pipeline inverter at the current moment, k p , k i , k d They respectively represent the proportional factor, integral factor and differential factor of the slurry discharge pipeline flow controller, er(t) represents the slurry discharge pipeline flow error at the current moment, er(t-1) represents the slurry discharge pipeline flow error at the previous moment, er(t-2) represents the slurry discharge pipeline flow error at the previous moment, and e0 represents the preset threshold.

3. The air cushion tank liquid level automatic control method according to claim 1, characterized in that: The dynamic characteristic model of the air cushion tank liquid level can be expressed as: Among them, y(t) represents the model output, that is, the air cushion tank liquid level, t represents the current time, p represents the order of the model output, q represents the order of the model input, s represents the order of the model measurable interference, f represents the delay of the model input, g represents the delay of the model measurable interference, u represents the difference between the inlet and outlet flow rates, d represents the model measurable interference vector, d1 represents the average propulsion speed, and d2 represents the air cushion tank pressure, T represents the transpose, ε(t) represents the Gaussian white noise signal, and a represents the Gaussian white noise signal. 0,t-1 The coefficient matrix representing the model bias, where a 11，0,t-1 for a 0,t-1 The elements in a i,t-1 Represents the coefficient matrix of the model output, where a 11,i,t-1 for a i,t-1 Elements in b j,t-1 Represents the coefficient matrix of the model input, where b 11,j,t-1 for b j,t-1 Elements in; c k,t-1 The coefficient matrix representing the measurable disturbance of the model, where c 11,k,t-1 、c 12,k,t-1 c k,t-1 The elements in a 0,t-1 、a i,t-1 , b j,t-1 、c k,t-1 is calculated by the state vector through the BP neural network; x(t-1) represents the state vector, n w Represents the order of the state vector, which serves as the input of the BP neural network; Represents the nth x nodes; Represents the distance from the lth node in the hidden layer of the BP network to the nth node in the input layer x The weight of the node; h l represents the output of the lth node in the hidden layer of the BP network; b l represents the bias of the lth node in the hidden layer of the BP network; n l Indicates the number of nodes in the hidden layer of the BP network; represents the activation function; n o Indicates the number of nodes in the output layer of the BP network; W m_l represents the weight from the mth node in the output layer to the lth node in the hidden layer of the BP network; m represents the output of the mth node in the output layer of the BP network; b m represents the bias of the mth node in the output layer of the BP network; O1 represents the output of the first node in the output layer of the BP network; O2 represents the output of the second node in the output layer of the BP network; 1+p Represents the output of the 1+pth node in the output layer of the BP network; 1+p+1 Represents the output of the 1+p+1th node of the BP network output layer; 1+p+q Represents the output of the 1+p+qth node in the output layer of the BP network; 1+p+q+1 Represents the output of the 1+p+q+1th node of the BP network output layer; 1+p+q+2s represents the output of the 1+p+q+2sth node in the output layer of the BP network; θ t-1 represents the state dependency coefficient vector.

4. The air cushion tank liquid level automatic control method according to claim 3, characterized in that: The state space model of the inlet and outlet flow difference prediction controller is: in, y(t) represents the output variable, x(t) represents the state variable, A t , B t , Φ t , C are coefficient matrices, a 1,t represents the coefficient matrix when the model output order is 1 at time t, a 2,t represents the coefficient matrix when the model output order is 2 at time t, a kn-1,t Indicates that the model output order at time t is k n The coefficient matrix when -1, a kn,t Indicates that the model output order at time t is k n The coefficient matrix when b 1,t represents the coefficient matrix when the model input order is 1 at time t, b 2,t represents the coefficient matrix when the model input order is 2 at time t, b kn-1,t Indicates that the model input order at time t is k n The coefficient matrix when -1, b kn,t Indicates that the model input order at time t is k n , φ(t) represents the coefficient matrix at time t and the sum of the bias of the model and the interference of each order model, Ξ(t+1) represents the matrix of white noise at time t+1, ε(t+1) represents the white noise at time t+1, and T represents the transpose of the vector.

5. The air cushion tank liquid level automatic control method according to claim 4, characterized in that: The predictive control objective function is: Among them, minJ represents the predictive control objective function, express The 2-norm of express The 2-norm of represents the future N of the air cushion tank liquid level y obtained based on the state space model y One step ahead prediction vector, y r represents the expected vector of the air cushion tank liquid level, y min Indicates the lower limit of the air cushion tank liquid level, y max Indicates the upper limit of the air cushion tank liquid level, u min Indicates the lower limit of the difference between the inlet and outlet flow rates, u max Indicates the upper limit of the difference between the inlet and outlet flow rates. represents the control vector obtained at time t The increment vector, Δu min Indicates the lower limit of the difference between the inlet and outlet flow rates, Δu max Indicates the incremental upper limit of the difference between the inlet and outlet flow rates. is the control vector solved at time t, and R1(t) represents the weighting coefficient.

6. The air cushion tank liquid level automatic control method according to claim 5, characterized in that: The weighting coefficient is adaptively adjusted based on the following formula: Among them, β represents the adjustment coefficient, N g represents the observation window length, R min Represents the minimum value of the weighting coefficient, R max It represents the maximum value of the weighted coefficient, y(t-i3) represents the liquid level of the air cushion tank at the moment t-i3, and y(t-i3-1) represents the liquid level of the air cushion tank at the moment t-i3-1.

7. The air cushion tank liquid level automatic control method according to claim 1, characterized in that: In the process of obtaining excavation data, under the premise of ensuring the safe state of the air cushion tank liquid level, the air cushion tank liquid level can be changed over a large range by adjusting the difference in the inlet and outlet slurry flow rates.

8. An automatic control system for liquid level in an air cushion tank, characterized in that: include: The excavation data acquisition module is used to obtain the excavation data of the slurry shield machine during the excavation process and normalize it, wherein the excavation data includes the air cushion chamber liquid level, the difference between the inlet and outlet slurry flow, the average value of the propulsion speed and the air cushion chamber pressure value; The model building and training module is used to build the dynamic characteristic model of the air cushion tank liquid level based on the BP network model and the ARX model, and iteratively train the dynamic characteristic model of the air cushion tank liquid level using the normalized excavation data until the iteration termination condition is met, and save the trained dynamic characteristic model of the air cushion tank liquid level; The inlet and outlet flow difference prediction control module is used to build an inlet and outlet flow difference prediction controller based on the trained air cushion tank liquid level dynamic characteristic model, collect real-time excavation data, normalize it and input it into the inlet and outlet flow difference prediction controller, optimize and solve the prediction control objective function of the inlet and outlet flow difference prediction controller, and obtain the optimal solution for the inlet and outlet flow difference; The automatic control module for the inlet and outlet slurry flow is used to control the inlet and outlet slurry flow based on the optimal solution of the inlet and outlet slurry flow difference, so that the liquid level in the air cushion bin is stabilized at a reference value.

9. An electronic device, characterized in that: The method comprises a processor and a memory, wherein the memory stores a computer program, and the processor executes the steps of the method according to any one of claims 1 to 7 by calling the computer program stored in the memory.

10. A computer-readable storage medium for storing a computer program for automatic control of liquid level in an air cushion tank, characterized in that: When the computer program is run on a computer, the steps of the method according to any one of claims 1 to 7 are executed.

Images