Air cushion warehouse liquid level automatic control method and system, electronic equipment and storage medium

Abstract

The application discloses a kind of air cushion storehouse liquid level automatic control method and system, electronic equipment, storage medium, the method is based on data-driven modeling, fully excavates the dynamic characteristics contained in the mud shield tunneling air cushion storehouse liquid level control related data, the air cushion storehouse liquid level dynamic characteristic model constructed combines BP network model and ARX model, regression coefficient matrix of ARX model is fitted using BP neural network, so that the model has very strong global nonlinear description ability, more close to shield machine air cushion storehouse liquid level actual work characteristics;And, also based on air cushion storehouse liquid level dynamic characteristic model constructs the difference value prediction controller of in and out slurry flow, after collecting real-time tunneling data, by optimizing and solving to prediction control objective function, the optimal solution of the difference value of in and out slurry flow can be obtained, the optimal solution can make air cushion storehouse liquid level stabilize around reference value, the regulation range of in and out slurry pump is very small, the whole circulation system is stable, and mud storehouse pressure fluctuation is very small.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of slurry balance shield machines, in particular, relates to a kind of air cushion bin liquid level automatic control method and system, electronic equipment, computer readable storage medium. BACKGROUND

[0002] The shield machine is a special construction equipment for tunneling engineering, which has the functions of cutterhead excavation, hydraulic propulsion, segment assembly, spoil conveying and guidance measurement, and is a complex system integrating advanced technologies such as machine, electricity, fluid and light. The slurry balance shield machine accounts for a large proportion in the entire shield market. The excavation bin of the slurry balance shield machine is divided into a slurry bin and an air cushion bin. The slurry bin is filled with slurry, the lower part of the air cushion bin is filled with slurry, and the upper part of the air cushion bin is filled with compressed air. When the slurry shield machine is normally excavating, the slurry level in the air cushion bin needs to be ensured on the central axis. At present, the air cushion bin liquid level control is mostly manually controlled, and a dedicated person is configured to monitor the change of the air cushion bin liquid level on the slurry operation panel interface all day and manually adjust it. This manual control method requires a high level of operation of the operator. If the operation is not proper, the large range fluctuation of the air cushion bin liquid level will have an adverse effect on the slurry bin pressure and the air cushion bin automatic pressure maintenance, and more seriously, will cause the ground settlement. Therefore, the air cushion bin liquid level automatic control of the slurry balance shield machine has important practical significance. The Chinese patent application with the publication number CN114810104A discloses an air cushion bin liquid level adjusting system and method of a slurry air balance shield. The air cushion bin liquid level is monitored by an air cushion bin liquid level monitoring system, a feedback liquid level change signal is fed to a shield machine slurry circulating system, and the slurry circulating system adjusts the opening and closing of a ball valve or the flow of a slurry pump according to the received signal, so as to realize the function of automatically adjusting the stable air cushion bin liquid level. However, the scheme of the patent application limits the air cushion bin liquid level between a high liquid level and a low liquid level, resulting in a large adjustment range of the in-out slurry pump and poor stability of the entire circulating system, and large pressure fluctuation of the slurry bin. SUMMARY

[0003] The present application provides an air cushion bin liquid level automatic control method and system, electronic equipment and computer readable storage medium, which can stabilize the air cushion bin liquid level near a reference value, the adjustment range of the in-out slurry pump is very small, the entire circulating system is stable, and the pressure fluctuation of the slurry bin is very small.

[0004] According to one aspect of the present application, an air cushion bin liquid level automatic control method is provided, which includes the following contents:

[0005] Excavation data in the tunneling process of the slurry balance shield machine are acquired and normalized, wherein the excavation data include the air cushion bin liquid level, the in-out slurry flow difference value, the average value of the propulsion speed and the air cushion bin pressure value;

[0006] The gas cushion chamber liquid level dynamic characteristic model is constructed based on a BP network model and an ARX model, and the normalized tunneling data are used for iterative training of the gas cushion chamber liquid level dynamic characteristic model until the iteration termination condition is met, and the trained gas cushion chamber liquid level dynamic characteristic model is saved;

[0007] Based on the trained gas cushion chamber liquid level dynamic characteristic model, a difference value prediction controller of the inlet and outlet slurry flow is constructed, real-time tunneling data are collected and normalized, and then input into the difference value prediction controller of the inlet and outlet slurry flow, and an optimal solution of the difference value of the inlet and outlet slurry flow is obtained by optimizing and solving the prediction control objective function of the difference value prediction controller of the inlet and outlet slurry flow;

[0008] Based on the optimal solution of the difference value of the inlet and outlet slurry flow, the inlet and outlet slurry flow is controlled, so that the gas cushion chamber liquid level is stabilized at the reference value.

[0009] Further, the inlet and outlet slurry flow is controlled based on the following formula:

[0010]

[0011] Wherein, u * (t) represents the optimal solution of the difference value of the inlet and outlet slurry flow, 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 frequency set increase value 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, and e0 represents a preset threshold.

[0012] Further, the gas cushion chamber liquid level dynamic characteristic model can be represented as:

[0013]

[0014] Wherein, y(t) represents the model output, i.e. the gas cushion chamber 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 value of the inlet and outlet slurry flow; d represents the measurable disturbance vector, d1 represents the average value of the propulsion speed, and d2 represents the gas cushion chamber pressure; T represents transposition; ε(t) represents a Gaussian white noise signal; a 0,t-1 represents the coefficient matrix of the model bias, wherein a 11，0,t-1 is a0,t-1 Elements in; a i,t-1 This 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 This 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 This represents the coefficient matrix of the model's measurable disturbances, where c 11,k,t-1 c 12,k,t-1 For c k,t-1 Elements in; a 0,t-1 a i,t-1 b j,t-1 c k,t-1 It is calculated from the state vector using a BP neural network; x(t-1) represents the state vector, n w The order of the state vector is used as the input to the BP neural network. Represents the nth state in the state vector x One node; This represents the distance from the l-th node in the hidden layer of a BP network to the n-th node in the input layer. x The weights of each node; h l This represents the output of the l-th node in the hidden layer of the BP network; b l This represents the bias of the l-th node in the hidden layer of the BP network; n l This indicates the number of nodes in the hidden layer of the BP network; n represents the activation function; o W represents the number of nodes in the output layer of a BP network. m_l This represents the weights from the m-th node in the output layer to the l-th node in the hidden layer of the BP network; O m b represents the output of the m-th node in the output layer of a BP network. m O represents the bias of the m-th node in the output layer of the BP network; O1 represents the output of the 1-th node in the output layer of the BP network; O2 represents the output of the 2-th node in the output layer of the BP network; O 1+p This represents the output of the (1+p)th node in the output layer of a BP network; O 1+p+1 This represents the output of the (1+p+1)th node in the output layer of a BP network; O 1+p+q This represents the output of the (1+p+q)th node in the output layer of a BP network; O 1+p+q+1 This represents the output of the (1+p+q+1)th node in the output layer of a BP network; O 1+p+q+2s θ represents the output of the (1+p+q+2s)th node in the output layer of the BP network; t-1 This represents the state dependency coefficient vector.

[0015] Furthermore, the state-space model of the inlet and outlet slurry flow rate difference prediction controller is as follows:

[0016]

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

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

[0019]

[0020] wherein min J represents the predictive control objective function, represents the 2-norm of , represents the 2-norm of , represents a future N y step forward prediction vector of the air cushion bin liquid level y based on the state space model, y r represents an expected vector of the air cushion bin liquid level, y min represents a lower limit of the air cushion bin liquid level, y max represents an upper limit of the air cushion bin liquid level, u min represents a lower limit of the difference value of the pulp feeding and discharging flow rate, u max represents an upper limit of the difference value of the pulp feeding and discharging flow rate, represents an increment vector of the control vector solved at time t, Δumin an increment lower limit of the difference between the inlet and outlet slurry flow rates, Δu max an increment upper limit of the difference between the inlet and outlet slurry flow rates, is a control vector solved at time t, R1(t) represents a weighting coefficient.

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

[0022]

[0023] wherein β represents an adjustment coefficient, N g represents an observation window length, R min represents a minimum value of the weighting coefficient, R max represents a maximum value of the weighting coefficient, y(t-i3) represents the air cushion chamber liquid level at time t-i3, and y(t-i3-1) represents the air cushion chamber liquid level at time t-i3-1.

[0024] Further, in the process of obtaining the tunneling data, the air cushion chamber liquid level is changed in a large range by adjusting the difference between the inlet and outlet slurry flow rates under the premise of ensuring the safety state of the air cushion chamber liquid level.

[0025] In addition, the present application also provides an air cushion chamber liquid level automatic control system, comprising:

[0026] a tunneling data acquisition module, configured to acquire tunneling data in a tunneling process of a slurry balance shield tunneling machine and perform normalization processing on the tunneling data, wherein the tunneling data comprises an air cushion chamber liquid level, a difference between inlet and outlet slurry flow rates, an average value of a pushing speed, and an air cushion chamber pressure value;

[0027] a model construction and training module, configured to construct an air cushion chamber liquid level dynamic characteristic model based on a BP network model and an ARX model, and perform iterative training on the air cushion chamber liquid level dynamic characteristic model by using the normalized tunneling data until an iterative termination condition is met, and save the trained air cushion chamber liquid level dynamic characteristic model;

[0028] a difference between inlet and outlet slurry flow rates prediction control module, configured to construct a difference between inlet and outlet slurry flow rates prediction controller based on the trained air cushion chamber liquid level dynamic characteristic model, input normalized real-time tunneling data into the difference between inlet and outlet slurry flow rates prediction controller, and optimize and solve a prediction control objective function of the difference between inlet and outlet slurry flow rates prediction controller to obtain an optimal solution of the difference between inlet and outlet slurry flow rates;

[0029] a difference between inlet and outlet slurry flow rates automatic control module, configured to control the difference between inlet and outlet slurry flow rates based on the optimal solution of the difference between inlet and outlet slurry flow rates, so that the air cushion chamber liquid level is stabilized at a reference value.

[0030] In addition, the application further provides an electronic device comprising a processor and a memory, wherein the memory stores a computer program, and the processor is configured to execute the steps of the method by invoking the computer program stored in the memory.

[0031] In addition, the application further provides a computer readable storage medium for storing a computer program for automatic control of the air cushion chamber liquid level, wherein the computer program executes the steps of the method when running on a computer.

[0032] The application has the following beneficial effects:

[0033] The automatic control method for the air cushion chamber liquid level of the application is based on data-driven modeling, fully excavates the dynamic characteristics contained in the data related to the air cushion chamber liquid level control of the slurry shield tunneling, and combines the BP network model and the ARX model to build the air cushion chamber liquid level dynamic characteristic model, so that the model has strong global nonlinear description capability and is more suitable for the actual working characteristics of the air cushion chamber liquid level of the shield machine. In addition, the air cushion chamber liquid level dynamic characteristic model is used to build a difference prediction controller for the inflow and outflow slurry flow, and the optimal solution of the difference between the inflow and outflow slurry flow can be obtained by optimizing and solving the prediction control objective function after collecting real-time tunneling data, which can make the air cushion chamber liquid level stable around the reference value, the adjustment range of the inflow and outflow slurry pump is small, the entire circulation system is stable, and the slurry chamber pressure fluctuation is small.

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

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

[0036] The accompanying drawings, which form a part of the present application, are intended to provide further understanding of the application and are incorporated herein for a purpose of explanations of the application. The schematic embodiments of the application and their descriptions are used to explain the application and do not constitute an improper limitation on the application. In the drawings:

[0037] Figure 1 is a flowchart of the automatic control method for the air cushion chamber liquid level of the preferred embodiment of the application;

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

[0039] Figure 3 is Figure 1 is a subflowchart of step S2 in the preferred embodiment of the application;

[0040] Figure 4 is Figure 1 Another sub-flowchart of step S2 in the method is shown in Figure 2.

[0041] Figure 5 Figure 3 is a schematic diagram of a module structure of an automatic control system for a gas cushion chamber liquid level according to another embodiment of the present application. DETAILED DESCRIPTION

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

[0043] With reference to Figure 1 The preferred embodiments of the present application provide an automatic control method for a gas cushion chamber liquid level, which comprises the following contents:

[0044] Step S1: obtaining tunneling data in a tunneling process of a slurry balance shield tunneling machine and performing normalization processing on the tunneling data, wherein the tunneling data comprises a gas cushion chamber liquid level, a difference between an inlet slurry flow and an outlet slurry flow, an average value of a pushing speed, and a gas cushion chamber pressure value;

[0045] Step S2: constructing a dynamic characteristic model of the gas cushion chamber liquid level based on a BP network model and an ARX model, and performing iterative training on the dynamic characteristic model of the gas cushion chamber liquid level by using the normalized tunneling data until an iterative termination condition is met, and saving the trained dynamic characteristic model of the gas cushion chamber liquid level;

[0046] Step S3: constructing a difference between an inlet slurry flow and an outlet slurry flow prediction controller based on the trained dynamic characteristic model of the gas cushion chamber liquid level, inputting normalized real-time tunneling data into the difference between an inlet slurry flow and an outlet slurry flow prediction controller, and optimizing and solving a prediction control objective function of the difference between an inlet slurry flow and an outlet slurry flow prediction controller to obtain an optimal solution of the difference between an inlet slurry flow and an outlet slurry flow;

[0047] Step S4: controlling the difference between an inlet slurry flow and an outlet slurry flow based on the optimal solution of the difference between an inlet slurry flow and an outlet slurry flow, so that the gas cushion chamber liquid level is stabilized at a reference value.

[0048] It can be understood that the air cushion chamber liquid level automatic control method of the embodiment first collects tunneling data in the slurry shield tunneling process, then constructs an air cushion chamber liquid level dynamic characteristic model based on a BP network model and an ARX model, and iteratively trains the air cushion chamber liquid level dynamic characteristic model using the tunneling data, then constructs a difference value prediction controller of the slurry inlet and outlet flow based on the trained air cushion chamber liquid level dynamic characteristic model, and after collecting real-time tunneling data, the prediction control objective function of the difference value prediction controller of the slurry inlet and outlet flow is optimized and solved to obtain the optimal solution of the difference value of the slurry inlet and outlet flow, and finally, the optimal solution of the difference value of the slurry inlet and outlet flow is used to control the slurry inlet and outlet flow, so that the air cushion chamber liquid level is stabilized near the reference value. The present application is based on data-driven modeling, fully excavates the dynamic characteristics contained in the relevant data of the slurry shield tunneling air cushion chamber liquid level control, and the constructed air cushion chamber liquid level dynamic characteristic model combines the BP network model and the ARX model, uses the BP neural network to fit the regression coefficient matrix of the autoregressive ARX model with external variables, so that the model has strong global nonlinear description ability and is more suitable for the actual working characteristics of the shield machine air cushion chamber liquid level. Moreover, based on the air cushion chamber liquid level dynamic characteristic model, a difference value prediction controller of the slurry inlet and outlet flow is constructed, and after collecting real-time tunneling data, the optimal solution of the difference value of the slurry inlet and outlet flow can be obtained by optimizing and solving the prediction control objective function, which can make the air cushion chamber liquid level stable near the reference value, the adjustment range of the slurry inlet and outlet pump is small, the entire circulation system is stable, and the slurry chamber pressure fluctuation is small.

[0049] It can be understood that in the step S1, tunneling data in the slurry balance shield tunneling process is collected, and the tunneling data includes the air cushion chamber liquid level, the difference value of the slurry inlet and outlet flow, the average value of the advancing speed and the air cushion chamber pressure value, wherein the difference value of the slurry inlet and outlet flow refers to the difference between the slurry inlet flow and the slurry outlet flow. Since the slurry balance shield machine has advancing mode, bypass mode, pipeline extension mode, backwashing mode and weekend mode, only the tunneling data of the advancing mode of the shield machine is collected as identification data, the sampling period of the tunneling data is set to 1 second, and the tunneling data is continuously collected until the tunneling data reaches the preset quantity requirement. In addition, in the process of obtaining the tunneling data, under the premise of ensuring the safety of the air cushion chamber liquid level, the air cushion chamber liquid level is changed in a large range by adjusting the difference value of the slurry inlet and outlet flow, so that the dynamic characteristics of the air cushion chamber liquid level are completely contained in the collected tunneling data, which is beneficial to the control precision of the air cushion chamber liquid level. Then, the maximum and minimum value method is used to normalize the collected tunneling data, and the tunneling data is converted to [0, 1], and the conversion formula is as follows:

[0050]

[0051] wherein Y represents normalized tunneling data, X represents tunneling data before normalization, including air cushion chamber liquid level, difference between slurry feeding and discharging flow, average advancing speed and air cushion chamber pressure value, X max represents maximum value of tunneling data, X min represents minimum value of tunneling data. Of course, in other embodiments of the present application, other normalization methods can also be used, such as Z-score normalization method, maximum value normalization method, etc.

[0052] It can be understood that in the step S2, the air cushion chamber liquid level dynamic characteristic model is constructed based on the BP network model and the ARX model, and the air cushion chamber liquid level dynamic characteristic model is iteratively trained using the normalized tunneling data until the iteration termination condition is met, and the trained air cushion chamber liquid level dynamic characteristic model is saved. It can be understood that the network structure of the air cushion chamber liquid level dynamic characteristic model is as shown in Figure 2 The BP neural network has three layers, namely input layer, hidden layer and output layer. The model extracts nonlinear features in the air cushion chamber liquid level time series data of the slurry shield tunneling 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 strong global nonlinear description ability and is more suitable for the actual working characteristics of the shield tunneling machine.

[0053] The air cushion chamber liquid level dynamic characteristic model can be represented as:

[0054]

[0055] wherein y(t) represents model output, i.e. air cushion chamber liquid level, t represents current time; p represents order of model output; q represents order of model input; s represents order of measurable disturbance of the model; f represents delay of model input; g represents delay of measurable disturbance of the model; u represents difference between slurry feeding and discharging flow; d represents measurable disturbance vector, d1 represents average advancing speed, and d2 represents air cushion chamber pressure; T represents transposition; ε(t) represents Gaussian white noise signal; a 0,t-1 represents coefficient matrix of model bias, wherein a 11，0,t-1 is an element in a 0,t-1 ; a i,t-1 represents coefficient matrix of model output, wherein a 11,i,t-1 is an element in a i,t-1 ; b j,t-1 represents coefficient matrix of model input, wherein b 11,j,t-1 is an element in b j,t-1 ; c k,t-1 represents coefficient matrix of measurable disturbance of the model, wherein c 11,k,t-1 , c 12,k,t-1 are elements in c k,t-1 .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, as the input of the BP neural network; represents the nth x node in the state vector; represents the weight of the lth x node in the hidden layer of the BP network to the nth l node in the input layer; h l represents the output of the lth l node in the hidden layer of the BP network; b 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 of the mth m node in the output layer of the BP network to the lth m node in the hidden layer; O m represents the bias of the mth 1+p node in the output layer of the BP network; O1 represents the output of the 1st 1+p+1 node in the output layer of the BP network; O2 represents the output of the 2nd 1+p+q node in the output layer of the BP network; O 1+p+q+1 represents the output of the 1+p+1th 1+p+q+2s 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 the air cushion bin liquid level dynamic characteristic model is constructed, the difference between the inlet and discharge pulp flow is used as an input variable, the average value of the advancing speed and the air cushion bin pressure are used as measurable disturbances, and the air cushion bin liquid level is used as the output variable and the state variable of the model. The air cushion bin liquid level dynamic characteristic model is iteratively trained. As shown in Figure 3 , the process of iteratively training the air cushion bin liquid level dynamic characteristic model using the normalized tunneling data includes the following contents:

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

[0058] Step S22: Taking the difference between the pulp feeding and discharging flow rates as an input variable, taking the average propelling speed and the air cushion bin pressure as measurable disturbances, and taking the air cushion bin liquid level as an output variable and a state variable of the model, input layer data of the BP network model and the ARX model are constructed;

[0059] Step S23: Forward calculation;

[0060] Step S24: Model parameter update through back propagation;

[0061] Step S25: After changing the model structure parameters, steps S22, S23 and S24 are repeatedly executed, and the model structure parameters and the model parameters with the minimum loss function value are selected as 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] wherein y represents the air cushion bin liquid level, n w represents the order of Input_BP;

[0065] And the input layer data structure of the ARX end 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] The model structure parameters p, q, s, f, g, n w and the number of nodes n l of the BP hidden layer are initialized.

[0069] Then, the normalized air cushion bin liquid level, the difference between the pulp feeding and discharging flow rates, the average propelling speed and the air cushion bin pressure value are constructed into the input layer data of the BP network model and the ARX model according to the above two data structures.

[0070] Next, the data in the Input_BP structure is taken as the input of the BP network model, and the state-dependent coefficient vector θ t-1 , i.e., the coefficient of the ARX model, is obtained through calculation, and the prediction output of the model is:

[0071]

[0072] wherein, represents the air cushion bin liquid level prediction value.

[0073] Then, a loss function e is constructed, and the model parameters are constantly updated by a back propagation algorithm until the loss function is minimized, so as to obtain a model with high precision, wherein the expression of the loss function e is:

[0074]

[0075] wherein e represents the mean square error of the air cushion bin liquid level prediction value and the actual value, N represents the number of training samples, and y represents the actual value of the air cushion bin liquid level.

[0076] Finally, the model structure parameters p, q, s, f, g, n w and the number of nodes n of the BP network hidden layer are changed by a grid search method l , steps S22, S23 and S24 are repeatedly executed, after all the model structure parameters are traversed, the loss function values e under different model structure parameters are compared, and the model structure and model parameters with the minimum loss function value are selected as the optimal model.

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

[0078] Step S26: optimizing the model parameters by using a particle swarm optimization algorithm.

[0079] The specific process includes the following contents:

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

[0081] 2) calculating the cost function E of each particle cost , wherein the specific cost function is a prior art, and the specific expression is not described herein again;

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

[0083] wherein, represents the speed of the rth particle at the l d +1th iteration, represents the speed of the rth particle at the l dThe velocity at the next iteration, where α represents the inertia weight, β1 and β2 represent acceleration constants, and rand1 and rand2 represent two random numbers between 0 and 1, pbest r Let represent the best position found so far by the r-th particle, and gbest represent the best position found so far by the entire particle swarm. Indicates that the r-th particle is in l d The position at +1 iteration, For the r-th particle in l d The position at the next iteration;

[0084] 4) Update individual and global optimal solutions. For each particle, if the current position is better than the previously encountered best position, update its individual optimal value, and simultaneously find the minimum cost function E from all particles. cost The position is determined and updated to the global optimal solution using this position.

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

[0086] It is understood that this invention further optimizes the model parameters by leveraging the global optimization capability of the particle swarm optimization algorithm, thereby improving the model's global nonlinear description capability, enhancing its generalization ability, and making the output of the predictive controller based on this model more accurate, which is beneficial for the stable control of the air cushion chamber liquid level.

[0087] It is understood that in step S3, a predictive controller for the difference in slurry flow rate is constructed based on the trained dynamic characteristic model of the air cushion chamber liquid level. Real-time tunneling data is collected, normalized, and then input into the predictive controller. The predictive control objective function of the predictive controller is optimized to obtain the optimal solution for the difference in slurry flow rate. Here, real-time tunneling data refers to the current air cushion chamber liquid level, the difference in slurry flow rate, the average propulsion speed, and the air cushion chamber pressure. The predictive controller predicts and outputs the optimal solution for the difference in slurry flow rate at the next moment. At the next moment, the slurry flow rate is controlled according to this optimal solution to stabilize the air cushion chamber liquid level near the reference value.

[0088] Specifically, the dynamic characteristic model of the air cushion chamber liquid level is first transformed into an intermediate model structure, which can be represented as:

[0089]

[0090] Where, k nφ(t) represents the sum of the bias of the model at time t and the interference of each order model, and ε(t) represents the white noise at time t. In addition, the meanings of the characters a and b are consistent with the expression of the dynamic characteristic model of the air cushion bin liquid level, but considering that the actual range of the subscript has changed, i1 and j1 are used here.

[0091] Then, the state variable is defined as follows:

[0092]

[0093] where k1 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 k1; represents the output coefficient matrix; represents the input coefficient matrix.

[0094] According to the state space model of the difference value prediction controller of the pulp feeding and discharging flow rate obtained according to the above defined state variable and intermediate model structure, the following can be obtained:

[0095]

[0096] where, 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 of the model output order 1 at time t, a 2,t represents the coefficient matrix of the model output order 2 at time t, represents the coefficient matrix of the model output order k n -1 at time t, represents the coefficient matrix of the model output order k n at time t, b 1,t represents the coefficient matrix of the model input order 1 at time t, b 2,t represents the coefficient matrix of the model input order 2 at time t, represents the coefficient matrix of the model input order k n -1 at time t, represents the coefficient matrix of the model input order k n at time t, φ(t) represents the sum of the bias of the model at time t and the interference of each order model, Ξ(t+1) represents the matrix of the 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, the coefficient matrix A in the state space model is determined according to the tunneling data collected at time t t ,B t ,Φ t , a local linearization model based on the state space model is obtained, and a quadratic programming optimization algorithm is used to optimize the prediction control objective function of the difference value prediction controller of the pulp feeding and discharging flow rate online, wherein the specific quadratic programming optimization algorithm belongs to the prior art and will not be described here. The prediction control objective function is:

[0098]

[0099] Wherein, minJ represents the prediction 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 based on the state space model, y r represents the air cushion bin liquid level expected vector, 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 difference value of the pulp feeding and discharging flow rate, u max represents the upper limit of the difference value of the pulp feeding and discharging flow rate, represents the increment vector of the control vector solved at time t, Δu min represents the increment lower limit of the difference value of the pulp feeding and discharging flow rate, Δu max represents the increment upper limit of the difference value of the pulp feeding and discharging flow rate, is the control vector solved at time t, and R1(t) represents the weighting coefficient. It can be understood that after the control vector solved at time t is obtained, the first term u * (t) of is the optimal solution of the difference value of the pulp feeding and discharging flow rate.

[0100]

[0101]

[0102] Wherein, β represents the adjustment coefficient, N g represents the observation window length, R min represents the minimum value of the weighting coefficient, R max represents the maximum value of the weighting coefficient, y(t-i3) represents the air cushion bin liquid level at time t-i3, and y(t-i3-1) represents the air cushion bin liquid level at time t-i3-1.

[0103] It can be understood that the application adopts a weighting coefficient adaptive adjustment mechanism to adjust the control strength, and adjusts the weighting coefficient according to the actual output change, and increases the weighting coefficient R1(t) when the output change speed is large, and vice versa.

[0104] It can be understood that the feed and discharge pulp flow difference prediction controller has an online rolling optimization function, has strong anti-interference ability, can accurately control the air cushion bin liquid level to be stable near the reference value, and through the prediction control target function, constraints can be added, the weighting coefficient can be adaptively adjusted, the air cushion bin liquid level control is more stable, and the control cost is smaller.

[0105] It can be understood that after the optimal solution of the feed and discharge pulp flow difference is solved in step S3, in step S4, the feed and discharge pulp flow is controlled based on the following formula:

[0106]

[0107] Wherein, u * (t) represents the optimal solution of the feed and discharge pulp flow difference, represents the actual value of the feed pulp pipeline flow at the current moment, represents the actual value of the discharge pulp pipeline flow at the current moment, represents the set value of the discharge pulp pipeline flow at the current moment, and ΔF(t) represents the frequency setting increase value of the frequency converter of the discharge pulp pipeline at the current moment, k p , k i , and k d respectively represent the proportional factor, the integral factor and the differential factor of the discharge pulp pipeline flow controller, er(t) represents the discharge pulp pipeline flow error at the current moment, er(t-1) represents the discharge pulp pipeline flow error at the last moment, er(t-2) represents the discharge pulp pipeline flow error at the moment before the last moment, and e0 represents a preset threshold value, which can be set according to actual needs. In addition, if there are several relay stations in the discharge pulp pipeline, the frequency setting increase value of the relay station frequency converter is increased synchronously by ΔF(t).

[0108] As can be seen from the above formula, the optimal value of the feed and discharge pulp flow difference solved by the application is applied to the set value of the discharge pulp pipeline flow, which can reduce the disturbance to the slurry circulating system, and is beneficial to improve the stability of the circulating system.

[0109] As Figure 5 shown, another embodiment of the application further provides an air cushion bin liquid level automatic control system, which comprises:

[0110] The tunneling data acquisition module is configured to acquire tunneling data in the slurry balance shield tunneling process and perform normalization processing on the tunneling data, wherein the tunneling data includes the air cushion chamber liquid level, the difference between the inlet and outlet slurry flow, the average value of the advancing speed, and the air cushion chamber pressure value.

[0111] The model construction and training module is configured to construct an air cushion chamber liquid level dynamic characteristic model based on a BP network model and an ARX model, and perform iterative training on the air cushion chamber liquid level dynamic characteristic model using the normalized tunneling data until an iteration termination condition is met, and save the trained air cushion chamber liquid level dynamic characteristic model.

[0112] The inlet and outlet slurry flow difference value prediction control module is configured to construct an inlet and outlet slurry flow difference value prediction controller based on the trained air cushion chamber liquid level dynamic characteristic model, input the normalized real-time tunneling data into the inlet and outlet slurry flow difference value prediction controller, and optimize and solve a prediction control objective function of the inlet and outlet slurry flow difference value prediction controller to obtain an optimal solution of the inlet and outlet slurry flow difference value.

[0113] The inlet and outlet slurry flow automatic control module is configured to control the inlet and outlet slurry flow based on the optimal solution of the inlet and outlet slurry flow difference value, so that the air cushion chamber liquid level is stabilized at the reference value.

[0114] It can be understood that the air cushion chamber liquid level automatic control system of the embodiment first acquires tunneling data in the slurry shield tunneling process, then constructs an air cushion chamber liquid level dynamic characteristic model based on a BP network model and an ARX model, and performs iterative training on the air cushion chamber liquid level dynamic characteristic model using the tunneling data, then constructs an inlet and outlet slurry flow difference value prediction controller based on the trained air cushion chamber liquid level dynamic characteristic model, and after acquiring real-time tunneling data, optimizes and solves a prediction control objective function of the inlet and outlet slurry flow difference value prediction controller to obtain an optimal solution of the inlet and outlet slurry flow difference value, and finally controls the inlet and outlet slurry flow based on the optimal solution of the inlet and outlet slurry flow difference value, so that the air cushion chamber liquid level is stabilized near the reference value. The present application is based on data-driven modeling, fully excavates the dynamic characteristics contained in the relevant data of the slurry shield tunneling air cushion chamber liquid level control, and the constructed air cushion chamber liquid level dynamic characteristic model combines the BP network model and the ARX model, uses the BP neural network to fit the regression coefficient matrix of the autoregressive ARX model with an external variable, so that the model has strong global nonlinear description capability and is more suitable for the actual working characteristics of the shield machine air cushion chamber liquid level. Furthermore, the inlet and outlet slurry flow difference value prediction controller is constructed based on the air cushion chamber liquid level dynamic characteristic model, and after acquiring real-time tunneling data, the optimal solution of the inlet and outlet slurry flow difference value can be obtained by optimizing and solving the prediction control objective function, which can stabilize the air cushion chamber liquid level near the reference value, the adjustment range of the inlet and outlet slurry pump is small, the entire circulating system is stable, and the slurry chamber pressure fluctuation is small.

[0115] It can be understood that each module of the system embodiment corresponds to each step of the method embodiment, and thus the specific working process of each module will not be described here again, and the above method embodiment can be referred to.

[0116] In addition, another embodiment of the present application further provides an electronic device, comprising a processor and a memory, wherein the memory stores a computer program, and the processor is configured to execute the steps of the method by invoking the computer program stored in the memory.

[0117] In addition, another embodiment of the present application further provides a computer readable storage medium for storing a computer program for automatically controlling the liquid level of the air cushion bin, wherein the computer program executes the steps of the method when running on a computer.

[0118] The forms of the general computer readable storage medium 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 patterns of holes, random access memories (RAM), programmable read-only memories (PROM), erasable programmable read-only memories (EPROM), FLASH-EPROM, any other memory chips or cartridges, or any other computer readable media. The instructions can be further transmitted or received by a transmission medium. The term transmission medium includes any tangible or non-tangible medium that can be used to store, encode, or carry the instructions for execution by a machine, and includes digital or analog communication signals transmitted via wired or wireless communication links.

[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 an entirely hardware embodiment, an entirely 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 storage, CD-ROMs, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present application can be implemented in various computer languages, such as the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0120] The 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 such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flowcharts and / or blocks Figure 1 means for functionally implementing the steps listed in the flowchart block or 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 function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the function specified in the flowchart block or blocks. Figure 1 one or more flowcharts and / or blocks Figure 1 means for functionally implementing the steps listed in the flowchart block or blocks.

[0122] The 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 such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flowcharts and / or blocks Figure 1 means for functionally implementing the steps listed in the flowchart block or blocks.

[0123] While the preferred embodiments of the application have been described, additional variations and modifications can be made to the embodiments by those of skill in the art once they have the benefit of the present disclosure without departing from the spirit and scope of the application. Accordingly, the attached claims are intended to cover all such modifications and variations as falling within the scope of the application.

[0124] It is apparent that a person skilled in the art could make various changes and modifications to the application without departing from the spirit and scope of the application. Therefore, if these modifications and changes fall within the scope of the claims and their equivalents, it is intended to include them in the protection provided by this patent.

[0125] The application described above is merely preferred embodiments of the application and is not intended to limit the application to the embodiments described. The application can be modified and adapted in various ways and should be understood accordingly. Within the scope of the application, any modification, equivalent replacement, improvement, etc. should be included in the protection of the application.

Claims

1. An automatic control method of a gas cushion silo liquid level, characterized by, The method comprises the following steps: The tunneling data in the process of tunneling of the slurry balance shield tunneling machine is acquired and normalized, wherein the tunneling data comprises the air cushion bin liquid level, the difference between the feeding and discharging slurry flow, the average value of the pushing speed and the air cushion bin pressure value; A dynamic characteristic model of the air cushion bin liquid level is constructed based on a BP network model and an ARX model, and the dynamic characteristic model of the air cushion bin liquid level is iteratively trained by using the normalized tunneling data until an iteration termination condition is met, and the trained dynamic characteristic model of the air cushion bin liquid level is saved; A difference between feeding and discharging slurry flow prediction controller is constructed based on the trained dynamic characteristic model of the air cushion bin liquid level, real-time tunneling data is collected and normalized and then input into the difference between feeding and discharging slurry flow prediction controller, and an optimal solution of the difference between feeding and discharging slurry flow is obtained by optimizing and solving a prediction control objective function of the difference between feeding and discharging slurry flow prediction controller; The feeding and discharging slurry flow is controlled based on the optimal solution of the difference between feeding and discharging slurry flow, so that the air cushion bin liquid level is stabilized at a reference value; The dynamic characteristic model of the air cushion bin liquid level is represented as: Where y(t) represents the model output, i.e., the liquid level in the air cushion chamber, and 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 disturbance; f represents the delay of the model input; g represents the delay of the model-measurable disturbance; u represents the difference between the inlet and outlet slurry flow rates; d represents the model-measurable disturbance vector, d1 represents the average propulsion speed, and d2 represents the air cushion chamber pressure; T represents transpose; ε(t) represents the Gaussian white noise signal; a 0,t-1 The coefficient matrix represents the model bias, where a 11，0,t-1 For a 0,t-1 Elements in; a i,t-1 This 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 This 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 This represents the coefficient matrix of the model's measurable disturbances, where c 11,k,t-1 c 12,k,t-1 For c k,t-1 Elements in; a 0,t-1 a i,t-1 b j,t-1 ,ck ,t-1 It is calculated from the state vector using a BP neural network; x(t-1) represents the state vector, n w The order of the state vector is used as the input to the BP neural network. Represents the nth state in the state vector x One node; This represents the distance from the l-th node in the hidden layer of a BP network to the n-th node in the input layer. x The weights of each node; h l This represents the output of the l-th node in the hidden layer of the BP network; b l This represents the bias of the l-th node in the hidden layer of the BP network; n l This indicates the number of nodes in the hidden layer of the BP network; n represents the activation function; o W represents the number of nodes in the output layer of a BP network. m_l This represents the weights from the m-th node in the output layer to the l-th node in the hidden layer of the BP network; O m b represents the output of the m-th node in the output layer of a BP network. m O represents the bias of the m-th node in the output layer of the BP network; O1 represents the output of the 1-th node in the output layer of the BP network; O2 represents the output of the 2-th node in the output layer of the BP network; O 1+p This represents the output of the (1+p)th node in the output layer of a BP network; O 1+p+1 This represents the output of the (1+p+1)th node in the output layer of a BP network; O 1+p+q denotes the output of the (1 + p + q)th node of the output layer of the BP network; O 1+p+q+1 denotes the output of the (1 + p + q + 1)th node of the output layer of the BP network; O 1+p+q+2s denotes the output of the (1 + p + q + 2s)th node of the output layer of the BP network; θ t-1 denotes the state-dependent coefficient vector.

2. The air cushion bin level automatic control method according to claim 1, wherein The feeding and discharging slurry flow is controlled based on the following formula: wherein, u * (t) represents the optimal solution of the difference of the inlet and outlet flow rates, represents the actual value of the inlet flow rate at the current time, represents the actual value of the outlet flow rate at the current time, represents the set value of the outlet flow rate at the current time, ΔF(t) represents the set value of the frequency increase of the frequency converter of the outlet flow rate at the current time, k p , k i , k d respectively represent the proportional factor, the integral factor and the differential factor of the outlet flow rate controller, er(t) represents the error of the outlet flow rate at the current time, er(t-1) represents the error of the outlet flow rate at the previous time, er(t-2) represents the error of the outlet flow rate at the time before the previous time, and e0 represents a preset threshold value.

3. The air cushion bin level automatic control method according to claim 1, wherein The state space model of the difference between feeding and discharging slurry flow prediction controller is: wherein y(t) represents an output variable, x(t) represents a state variable, A t , B t , Φ t , C are coefficient matrices, a 1,t represents a coefficient matrix of a model output order of 1 at time t, a 2,t represents a coefficient matrix of a model output order of 2 at time t, a kn-1,t represents a coefficient matrix of a model output order of k n -1 at time t, a kn,t represents a coefficient matrix of a model output order of k n at time t, b 1,t represents a coefficient matrix of a model input order of 1 at time t, b 2,t represents a coefficient matrix of a model input order of 2 at time t, b kn-1,t represents a coefficient matrix of a model input order of k n -1 at time t, b kn,t represents a coefficient matrix of a model input order of k n at time t, φ(t) represents a sum of a bias of a model at time t and a disturbance of each order model, Ξ(t+1) represents a matrix of white noise at time t+1, ε(t+1) represents white noise at time t+1, and T represents a transpose of a vector.

4. The air cushion bin level automatic control method according to claim 3, wherein The prediction control objective function is: wherein minJ represents a prediction control target function, represents 2-norm of represents 2-norm of represents a future N y step forward prediction vector of the air cushion chamber liquid level, y r represents an air cushion chamber liquid level expected vector, y min represents a lower limit of the air cushion chamber liquid level, y max represents an upper limit of the air cushion chamber liquid level, u min represents a lower limit of the pulp flow difference, u max represents an upper limit of the pulp flow difference, represents a control vector solved at time t is an increment vector of the control vector, Δu min represents a lower limit of the pulp flow difference increment, Δu max represents an upper limit of the pulp flow difference increment, is a control vector solved at time t, and R1(t) represents a weighting coefficient.

5. The air cushion bin level automatic control method according to claim 4, wherein The weighting coefficient is adaptively adjusted based on the following formula: where β represents an adjustment coefficient, N g represents an observation window length, R min represents a minimum value of a weighting coefficient, R max represents a maximum value of a weighting coefficient, y(t-i3) represents a gas cushion tank liquid level at t-i3, and y(t-i3-1) represents a gas cushion tank liquid level at t-i3-1.

6. The air cushion bin level automatic control method of claim 1, wherein, In the process of acquiring the tunneling data, the air cushion bin liquid level is changed in a large range by adjusting the difference between feeding and discharging slurry flow under the premise of ensuring the safety state of the air cushion bin liquid level.

7. An automatic control system for the liquid level of a gas cushion silo, which employs the automatic control method for the liquid level of a gas cushion silo according to any one of claims 1 to 6, characterized by The method comprises the following steps: The tunneling data in the process of tunneling of the slurry balance shield tunneling machine is acquired and normalized, wherein the tunneling data comprises the air cushion bin liquid level, the difference between the feeding and discharging slurry flow, the average value of the pushing speed and the air cushion bin pressure value; A dynamic characteristic model of the air cushion bin liquid level is constructed based on a BP network model and an ARX model, and the dynamic characteristic model of the air cushion bin liquid level is iteratively trained by using the normalized tunneling data until an iteration termination condition is met, and the trained dynamic characteristic model of the air cushion bin liquid level is saved; A difference between feeding and discharging slurry flow prediction controller is constructed based on the trained dynamic characteristic model of the air cushion bin liquid level, real-time tunneling data is collected and normalized and then input into the difference between feeding and discharging slurry flow prediction controller, and an optimal solution of the difference between feeding and discharging slurry flow is obtained by optimizing and solving a prediction control objective function of the difference between feeding and discharging slurry flow prediction controller; The feeding and discharging slurry flow is controlled based on the optimal solution of the difference between feeding and discharging slurry flow, so that the air cushion bin liquid level is stabilized at a reference value.

8. An electronic device, comprising: The computer program runs on the computer to execute the steps of the method of any one of claims 1-6.

9. A computer readable storage medium for storing a computer program for automatic control of the level of a stock in an air cushion store, characterized in that The computer program runs on the computer to execute the steps of the method of any one of claims 1-6.

Images