Shield tunneling machine total thrust control method and system based on neural network

Abstract

The invention discloses a method and system for controlling the total thrust of a shield tunneling machine based on a neural network, and the method comprises the steps: obtaining the expected total thrust of the shield tunneling machine, and controlling the shield tunneling machine to carry out the propelling operation according to the expected total thrust of the shield tunneling machine; the actual working total thrust and the cutterhead torque of the shield tunneling machine are obtained; calculating an expected thrust adjustment coefficient according to the actual working total thrust, the cutterhead torque and the cutterhead torque threshold range; whether the cutterhead torque value exceeds the cutterhead torque threshold value range or not is judged; under the condition that the cutterhead torque exceeds the cutterhead torque threshold value range, the expected total propelling thrust of the shield tunneling machine is adjusted according to the thrust adjusting coefficient; and controlling the shield tunneling machine to adjust the actual working total thrust according to the adjusted expected total thrust. By means of the method, damage caused by too large torque of the cutterhead can be effectively avoided, the propelling speed is increased, and the tunneling efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of tunnel construction, and in particular to a total thrust control method and system for a shield machine based on a neural network. Background Art

[0002] The propulsion control of existing shield machines is mainly based on PID control strategy. However, under complex geological conditions, the dynamic relationship between the total propulsion thrust and the cutterhead torque is relatively complex. Traditional PID control cannot cope with the fluctuation of the cutterhead torque, resulting in low tunneling efficiency and the cutterhead is easily damaged due to excessive cutterhead torque. In addition, the current propulsion control system lacks the ability to adaptively adjust the cutterhead torque and cannot maximize the propulsion speed without damaging the equipment. Summary of the invention

[0003] One of the embodiments of this specification provides a total thrust control method for a shield machine based on a neural network. The method comprises:

[0004] Obtain the expected total thrust of the shield machine;

[0005] Controlling the shield machine to perform propulsion operations according to the expected total propulsion thrust;

[0006] Obtaining the actual total working thrust and cutter head torque of the shield machine;

[0007] Obtaining a desired thrust adjustment coefficient according to the actual total working thrust, the cutter head torque, and the cutter head torque threshold range;

[0008] Determining whether the cutter disc torque exceeds the cutter disc torque threshold range;

[0009] When the cutter head torque exceeds the cutter head torque threshold range, adjusting the expected total thrust of the shield machine according to the thrust adjustment coefficient;

[0010] The shield machine is controlled to adjust the actual working total thrust according to the adjusted expected total thrust.

[0011] In some embodiments, the thrust adjustment coefficient calculation model includes an input layer, a multi-head self-attention layer, a multi-layer perceptron, and an output layer.

[0012] One of the embodiments of this specification provides a total thrust control system of a shield machine based on a neural network, comprising:

[0013] The first data acquisition module is used to obtain the expected total thrust of the shield machine;

[0014] A control module, used for controlling the shield machine to perform a propulsion operation according to the expected total propulsion thrust;

[0015] The second data acquisition module is used to obtain the actual working total thrust and cutter head torque of the shield machine;

[0016] A thrust adjustment coefficient calculation module, used for obtaining a desired thrust adjustment coefficient according to the actual total working thrust, the cutter head torque and the cutter head torque threshold range;

[0017] A judgment module, used for judging whether the cutter disc torque exceeds the cutter disc torque threshold range;

[0018] An expected total thrust adjustment module, used for adjusting the expected total thrust of the shield machine according to the thrust adjustment coefficient when the cutter head torque exceeds the cutter head torque threshold range;

[0019] The total thrust adjustment module is used to control the shield machine to adjust the actual working total thrust according to the adjusted expected propulsion total thrust.

[0020] One of the embodiments of the present specification provides a total thrust control device for a shield machine based on a neural network, comprising a processor, wherein the processor is used to execute a total thrust control method for a shield machine based on a neural network.

[0021] One of the embodiments of this specification provides a computer-readable storage medium, which stores computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes a total thrust control method for a shield machine based on a neural network.

[0022] The beneficial effects of the present invention include:

[0023] (1) By adjusting the total thrust in real time to ensure that the cutter head torque is within a predetermined threshold range, damage caused by excessive cutter head torque can be effectively avoided.

[0024] (2) The adaptive adjustment capability of the cutterhead torque can maximize the propulsion speed and improve the excavation efficiency without damaging the equipment. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] This specification will be further described in the form of exemplary embodiments, which will be described in detail by the accompanying drawings. These embodiments are not restrictive, and in these embodiments, the same number represents the same structure, wherein:

[0026] Figure 1 is an exemplary system block diagram of the overall process of the shield machine propulsion system according to some embodiments of this specification;

[0027] Figure 2 is an exemplary flow chart of the operation of the shield machine propulsion system according to some embodiments of this specification;

[0028] Figure 3 is an exemplary flow chart of a total thrust control method of a shield machine based on a neural network according to some embodiments of this specification;

[0029] Figure 4 is an exemplary structural diagram of a thrust adjustment coefficient calculation model according to some embodiments of this specification;

[0030] Figure 5 is an exemplary flow chart of initial thrust adjustment coefficient calculation model training according to some embodiments of this specification;

[0031] Figure 6 It is an exemplary structural diagram of a shield machine total thrust control system based on a neural network as shown in some embodiments of this specification. DETAILED DESCRIPTION

[0032] In order to more clearly illustrate the technical solutions of the embodiments of this specification, the following is a brief introduction to the drawings required for the description of the embodiments. Obviously, the drawings described below are only some examples or embodiments of this specification. For ordinary technicians in this field, this specification can also be applied to other similar scenarios based on these drawings without creative work. Unless it is obvious from the language environment or otherwise explained, the same reference numerals in the figures represent the same structure or operation.

[0033] It should be understood that the "system", "device", "unit" and / or "module" used herein are a method for distinguishing different components, elements, parts, portions or assemblies at different levels. However, if other words can achieve the same purpose, the words can be replaced by other expressions.

[0034] As shown in this specification and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "comprise" and "include" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0035] Flowcharts are used in this specification to illustrate the operations performed by the system according to the embodiments of this specification. It should be understood that the preceding or following operations are not necessarily performed precisely in order. Instead, the steps may be processed in reverse order or simultaneously. At the same time, other operations may also be added to these processes, or one or more operations may be removed from these processes.

[0036] Figure 1 It is an exemplary system block diagram of the overall process of the shield machine propulsion system shown in some embodiments of this specification.

[0037] Shield machine, also known as shield tunnel boring machine, is a special engineering machinery for tunnel excavation. Modern shield tunnel boring machine integrates optical, mechanical, electrical, hydraulic, sensor and information technology. It has the functions of excavating and cutting soil, conveying soil ballast, assembling tunnel lining, measuring and guiding deviation correction, etc. It involves geology, civil engineering, machinery, mechanics, hydraulics, electricity, control, measurement and other disciplines and technologies. It must be designed and manufactured according to different geology, and the reliability requirements are extremely high. Shield tunnel boring machine has been widely used in subway, railway, highway, municipal, hydropower and other tunnel projects.

[0038] In some embodiments, the present specification is further improved based on a shield machine propulsion system based on a three-ring nested control architecture, and the shield machine propulsion system includes an outermost ring - a total thrust control ring, an intermediate ring - a posture change and an innermost ring - a cylinder control ring.

[0039] The outermost loop - total thrust control loop: is responsible for dynamically adjusting the expected total thrust of the total thrust PID control loop based on the cutter head torque of the shield machine, and realizing real-time adjustment of the total thrust through the PID controller. Among them, the total thrust control loop is directly related to the cutter head torque, and the value of the cutter head torque affects the adjustment of the total thrust.

[0040] Middle ring - attitude ring: This control ring adjusts the pressure of each group of cylinders according to the attitude deviation (horizontal deviation and vertical deviation) during the tunneling process of the shield machine to keep the attitude of the shield machine stable. The attitude deviation is monitored in real time by the sensor. When the attitude of the shield machine deviates during the tunneling process, the attitude ring will automatically adjust the corresponding cylinder pressure to ensure that the shield machine is advanced according to the designed axis.

[0041] Innermost ring - Cylinder control ring: This control ring is responsible for the specific ABCD four-group partition cylinder action to ensure accurate control of the propulsion posture. The pressure adjustment of the ABCD four-group partition cylinder is controlled by the posture control ring command to ensure the stability of the shield machine posture under different geological conditions.

[0042] Specifically, the neural network-based shield machine total thrust control method and system provided by the present invention is an optimization technical solution for adjusting the total thrust of the outermost loop - the total thrust control loop based on the above-mentioned shield machine propulsion system.

[0043] Figure 2 is an exemplary flow chart of the operation of a shield machine propulsion system according to some embodiments of the present specification.

[0044] During the tunneling process of the shield machine, the shield machine propulsion system first monitors the attitude deviation of the shield machine in real time through the attitude sensor. If there is a deviation in the horizontal or vertical direction of the shield machine, the attitude control loop will adjust the corresponding partition cylinder pressure according to the deviation information to ensure the stability of the shield machine's attitude. At the same time, the total thrust control loop will adjust the total thrust based on the change of the cutter head torque to ensure uniformity during the propulsion process. Finally, the cylinder control loop is responsible for executing the instructions of the attitude control loop and the total thrust control loop, and accurately controlling the ABCD partition cylinders, thereby realizing the automatic propulsion of the shield machine under complex geological conditions. The specific work flow is as follows:

[0045] (1) System initialization:

[0046] Start the shield machine automatic control system and check whether all sensors and control modules are working properly;

[0047] Initialize the initial pressure value of each group of zone cylinders;

[0048] Read and display the current posture of the shield machine, total thrust, pressure and stroke of the four groups of ABCD cylinders, and cutter head torque data.

[0049] (2) Read the cutter head torque:

[0050] The system displays the value of the cutter disc torque in real time;

[0051] If the cutter head torque exceeds a preset threshold, adjusting the desired total thrust in the total thrust control loop;

[0052] The adjusted total thrust setpoint is input to the total thrust control loop.

[0053] (3) Attitude control loop:

[0054] The system reads the data of the shield machine's attitude sensor in real time and monitors the horizontal and vertical attitude deviations of the shield machine;

[0055] If attitude deviation (including horizontal and vertical deviation) is detected, the system adjusts the desired attitude setting value of the attitude loop;

[0056] Adjust the pressure setting value of each group of zone cylinders according to the posture deviation to keep the shield machine posture stable;

[0057] Output attitude adjustment instructions to the cylinder control loop.

[0058] (4) Cylinder control ring

[0059] The system reads the data of the shield machine's attitude sensor in real time and monitors the horizontal and vertical attitude deviations of the shield machine;

[0060] The cylinder control loop adjusts the pressure of the four-group ABCD zone cylinders in real time according to the output instructions of the attitude control loop;

[0061] Adjust the posture of the shield machine to ensure that the shield machine excavates along the designed axis;

[0062] At the same time, the feedback of the cylinder pressure is detected to ensure the accurate execution of the control instructions.

[0063] (5) Promoting process monitoring

[0064] The system continuously monitors the total thrust, cutterhead torque, TBM posture, and the pressure and stroke of each group of cylinders;

[0065] If the posture, cutter torque and total thrust are detected to be beyond the preset range, an alarm will be issued immediately and the control parameters will be adjusted;

[0066] During the advancement process, the system checks and updates the control parameters at regular intervals to cope with changes in different geological conditions.

[0067] (6) End of excavation section

[0068] When the shield machine completes the advancement task of the preset excavation section, the system automatically stops the advancement operation;

[0069] The system enters standby mode, waiting for the start command of the next excavation task.

[0070] Figure 3 This is an exemplary flow chart of a total thrust control method of a shield machine based on a neural network according to some embodiments of this specification. The process specifically includes the following steps:

[0071] S310, obtaining the expected total thrust of the shield machine.

[0072] In some embodiments, the expected total thrust of the shield machine is the expected total thrust input by the operator. The operator can input the expected total thrust through the intelligent terminal interface integrated in the shield machine.

[0073] S320: Control the shield machine to perform propulsion operations according to the desired total propulsion thrust.

[0074] In some embodiments, the system will control the total thrust controller to operate the shield machine to start the propulsion operation according to the expected total thrust initially set by the operator. During the propulsion of the shield machine, the cutter head speed remains fixed according to the total thrust. In some cases, the control performance of the total thrust controller is unstable and is susceptible to sudden changes due to environmental changes, resulting in the actual working total thrust of the shield machine being unable to maintain the expected total thrust. Therefore, the system needs to obtain the actual working total thrust and cutter head torque during the operation of the shield machine in real time.

[0075] S330, obtaining the actual total working thrust and cutter head torque of the shield machine.

[0076] In some embodiments, the control performance of the total thrust controller is unstable and is susceptible to sudden changes due to environmental changes. The actual total working thrust of the shield machine will change according to environmental changes, and the actual total working thrust is positively correlated with the cutter head torque. When the cutter head torque is too large, it is easy to cause damage to the cutter head. When the cutter head torque is too small, the shield machine excavation speed is reduced and the working efficiency is low.

[0077] In some embodiments, the system can obtain the actual total working thrust and cutterhead torque of the shield machine in real time.

[0078] S340: Obtain a desired thrust adjustment coefficient according to the actual total working thrust, the cutter head torque, and the cutter head torque threshold range.

[0079] In some embodiments, the expected thrust adjustment coefficient can be obtained by calculating the thrust adjustment coefficient calculation model, wherein the thrust adjustment coefficient calculation model is obtained by training the initial thrust adjustment coefficient calculation model. For details about the calculation process of the thrust adjustment coefficient calculation model, please refer to Figure 4 and its details.

[0080] S350, determining whether the cutter disc torque exceeds the cutter disc torque threshold range.

[0081] In some embodiments, the cutter head torque threshold range can be input by the operator through the intelligent terminal interface. The actual working total thrust is positively correlated with the cutter head torque. By judging the cutter head torque, the cutter head equipment of the shield machine can be monitored in real time, and when the cutter head torque exceeds the threshold range, the total thrust of the shield machine can be adjusted in time.

[0082] S360: When the cutter head torque exceeds the cutter head torque threshold range, adjust the expected total propulsion thrust of the shield machine according to the thrust adjustment coefficient.

[0083] S370: Control the shield machine to adjust the actual total working thrust according to the adjusted expected total propulsion thrust.

[0084] In some embodiments, the system can adjust the actual total working thrust of the shield machine through feedforward control. For example, the expected propulsion adjustment coefficient is adjusted according to the operator input conditions (expected total propulsion thrust, cutter head torque range threshold, etc.) and the working condition of the shield machine, and the actual total working thrust of the shield machine is further adjusted.

[0085] In some embodiments, the system may also monitor the thrust deviation in real time through feedback control, and correct possible errors in the feedforward control when the control performance of the total thrust controller is poor or the environment changes suddenly.

[0086] For example, the feedforward part obtains the thrust adjustment coefficient k through the thrust adjustment coefficient calculation model of the attention mechanism; during the actual propulsion process, the thrust of the shield machine will be fed back to the control system in real time through the sensor; the feedback part calculates the error between the actual thrust and the expected thrust, and outputs a corrected thrust adjustment coefficient k through PID or other feedback algorithms. feedback .

[0087] Furthermore, the adjustment coefficients of the feedforward control and the feedback control are added together to obtain the final corrected thrust adjustment coefficient: k final =k+k feedback .

[0088] Because the control performance of the total thrust controller is relatively poor or the environment changes suddenly, there may be errors in the total thrust control. This correction thrust adjustment coefficient can offset this error. It can ensure both rapid response and precise thrust adjustment, and is suitable for shield machine propulsion tasks under complex geological conditions.

[0089] Figure 4 is an exemplary structural diagram of a thrust adjustment coefficient calculation model according to some embodiments of this specification.

[0090] In some embodiments, obtaining the expected thrust adjustment coefficient may include: inputting the actual total working thrust, the cutter disc torque, and the cutter disc torque threshold range into a thrust adjustment coefficient calculation model to obtain the expected thrust adjustment coefficient, wherein the thrust adjustment coefficient calculation model is obtained by training an initial thrust adjustment coefficient calculation model.

[0091] The thrust adjustment coefficient calculation model includes an input layer, a multi-head self-attention layer, a multi-layer perceptron, and an output layer. The calculation process of the thrust adjustment coefficient calculation model in step S340 specifically includes:

[0092] (1) Input layer

[0093] The input layer constructs the input features into a four-dimensional vector. For example, the actual working total thrust F, the cutter torque threshold range [p1, p2], and the cutter torque P constitute a four-dimensional vector X = [P, F, p1, p2], where each input feature is a scalar, and the thought vector is input into the multi-head self-attention layer.

[0094] (2) Multi-head self-attention layer

[0095] A multi-head attention mechanism is used to capture the interactions between different inputs (key control quantities, namely, input cutter disc torque P, total propulsion thrust F, and cutter disc torque threshold range [p1, p2]), which can dynamically assign the importance of each input (key control quantity).

[0096] In some embodiments, the three key control quantities of the input layer have different influences on the adjustment coefficient K of the output desired total thrust. The greatest influence is the total thrust F, followed by the cutter head torque P. These two values ​​determine the nonlinear relationship between the total thrust F and the cutter head torque P, and the upper and lower limits of the cutter head torque threshold [p1, p2] are supplementary reference factors in order to reduce or even eliminate the adjustment error.

[0097] The four-dimensional vector X is transformed linearly to generate query (Q), key (K) and value (V) vectors, randomly taking values ​​in a randomly selected range, such as [-0.1, 0.1], and uniformly randomly initializing the weight matrix W with a dimension of 4×4 Q , W K and W V :

[0098] The attention score is obtained by calculating the dot product of Q and K, and then the attention weight is generated by normalization through the softmax function: where d k is the dimension of the K vector, used for scaling.

[0099] Further, using the attention weights and value vectors, the output O of each head is calculated: O i =Attention(Q,K)·V (3).

[0100] In some embodiments, the output results of each attention head are merged (by splicing or averaging) to obtain a richer information representation. This merged output can better describe the operating state of the shield machine in different environments and the dynamic relationship between the cutter head torque and the propulsion thrust. The neural network of the multi-head self-attention layer provided in this specification has 3 heads, and the attention score of each head is calculated separately, and then their outputs are spliced ​​or averaged.

[0101] The concatenated attention score output: O=Concat(O1,O2,O3) (4). The averaged attention score output is:

[0102] (3) Multilayer Perceptron

[0103] In some embodiments, the input after the attention mechanism is processed by a multi-layer perceptron for modeling complex nonlinear relationships. This part may include two to three hidden layers, each layer has 16 to 32 neurons, and the activation function uses ReLU.

[0104] The output O of the multi-head self-attention layer is combined and input into a feedforward neural network (MLP) with multiple hidden layers to model the nonlinear relationship between the cutter head torque and the total thrust. Each layer of the network can extract complex features through activation functions (such as ReLU) and finally predict the expected thrust or cutter head torque adjustment coefficient.

[0105] The first hidden layer H1 is used to extract preliminary features of the input data, which may be low-level feature representations of input parameters such as cutter head torque and total propulsion thrust, aiming to capture their potential nonlinear relationships.

[0106] The first hidden layer has 16 neurons, the size of the weight matrix W1 is 16×3 (the input has 3 features), the size of the bias term b1 is 16×1, and a normal distribution is used. Generate random numbers and initialize the weight matrix W1 and bias term b1. Calculate the output of the first hidden layer: H1 = Re LU (W1·O+b1).

[0107] The second hidden layer H2 performs further deep feature modeling on the data after preliminary feature extraction, further optimizes the feature representation of the data, and provides input for the model to output the final thrust regulation decision.

[0108] The second layer has 32 neurons, the size of the weight matrix W2 is 32×16, the size of the bias term b2 is 32×1, and the normal distribution is used. Generate random numbers and initialize the weight matrix W2 and bias term b2. Perform the same calculation: H2 = ReLU (W2·H1+b2).

[0109] Final output adjustment factor:

[0110] k=W3·H2+b3 (6).

[0111] Among them, W3 represents the weight matrix from the second hidden layer H2 to the output layer, which maps the output of the hidden layer to the final output; b3 represents the bias term of the output layer, which combines with the weight matrix to determine the final result of the output layer.

[0112] The final output of the model (expected thrust adjustment coefficient) can be used in real time to adjust the operating parameters of the shield machine. During the actual advancement process, the system continuously predicts and adjusts based on historical data to ensure a smooth excavation process and maximize efficiency.

[0113] Figure 5 is an exemplary flow chart of initial thrust adjustment coefficient calculation model training according to some embodiments of the present specification.

[0114] In some embodiments, the thrust adjustment coefficient calculation model is obtained through initial thrust adjustment coefficient calculation model training, and the initial thrust adjustment coefficient calculation model training process includes:

[0115] Step S510: constructing a training data set of the shield machine, including the total thrust, cutterhead torque, cutterhead torque threshold, and ideal thrust adjustment coefficient in historical mining data.

[0116] In some embodiments, the training data of the initial thrust adjustment coefficient calculation model comes from the historical operation data of the actual excavation process of the shield machine. The neural network is trained using a supervised learning method. The training data includes the actual value of the cutter head torque, the total thrust and the ideal thrust adjustment coefficient under different geological conditions.

[0117] Step S520: Input the training data set into the initial thrust adjustment coefficient calculation model to obtain the thrust adjustment coefficient output by the model. For the specific steps of step 520, please refer to Figure 4 Detailed description of step S320.

[0118] Step S530: According to the ideal thrust adjustment coefficient and the thrust adjustment coefficient output by the model, a loss function is used to adjust the network weight and the bias coefficient.

[0119] In some embodiments, the goal of the initial thrust adjustment coefficient calculation model is to predict a continuous thrust adjustment coefficient, and the loss function may use a mean square error, which is as follows: Where N is the number of samples; is the thrust adjustment coefficient predicted by the neural network; is the target adjustment coefficient, which serves as the goal of supervised learning.

[0120] In some embodiments, the shield driver or operator uses the intelligent terminal to automatically control the shield machine during excavation, and continuously adjusts and sets the expected total thrust F according to the change of the cutter head torque. 1desiret , record the history of these operations, and according to the total thrust control strategy, the real thrust adjustment coefficient K can be extracted from these historical data according to the cutter head torque threshold true .

[0121] For example: If the current cutter torque P exceeds the set upper limit p2, the target adjustment coefficient is directly calculated according to the following formula: If the cutter head torque P is lower than the threshold value p1, the following is calculated similarly:

[0122] Step S540: When the error between the ideal thrust adjustment coefficient and the thrust adjustment coefficient output by the model is less than a threshold, a trained thrust adjustment coefficient calculation model is obtained.

[0123] In some embodiments, when the cutter head torque value exceeds the cutter head torque threshold range, step S360 may include multiple embodiments of adjusting the total thrust of the shield machine according to the thrust adjustment coefficient. The specific adjustment method is as follows:

[0124] Embodiment 1

[0125] When the cutter disc torque value is less than the lower limit value of the cutter disc torque threshold range, the total propulsion thrust is increased according to the thrust adjustment coefficient.

[0126] For example, when the cutter head torque P is less than the set cutter head torque lower limit threshold p1, the total thrust is increased accordingly to improve the excavation efficiency. The calculation formula is:

[0127] F1=F+k(p1-P) (11);

[0128] When the cutter head torque value is greater than the upper limit value of the cutter head torque threshold range, reducing the total propulsion thrust according to the thrust adjustment coefficient;

[0129] For example: when the cutter disc torque value P is greater than the set cutter disc torque threshold upper limit p2, in order to prevent the cutter disc from being damaged due to excessive torque, the system can automatically reduce the total thrust F. The calculation formula is:

[0130] F1=Fk(P-p2) (10),

[0131] Among them, F1 is the expected total thrust, P is the current cutterhead torque, F is the actual total working thrust of the shield machine, [p1, p2] is the cutterhead torque threshold range, p1 is the lower limit of the cutterhead torque threshold range, and p2 is the upper limit of the cutterhead torque threshold range.

[0132] Embodiment 2

[0133] In some cases, the actual working conditions of the shield machine are very complex and changeable, and different geological and stratigraphic conditions can be considered in the total thrust control strategy. The shield machine driver can set the address adjustment index in the intelligent terminal excavation parameter control interface according to different geological and stratigraphic conditions, so that the thrust control has different sensitivities to the cutterhead torque under different geological conditions, which is suitable for coping with different geological conditions and working conditions.

[0134] In some embodiments, adjusting the total propulsion thrust of the shield machine according to the thrust adjustment coefficient may include: setting a geological adjustment index based on geological conditions when the cutter head torque exceeds the cutter head torque threshold range; and adjusting the total propulsion thrust of the shield machine in combination with the geological adjustment index and according to the expected thrust adjustment coefficient.

[0135] For example, when the cutter head torque P is less than the set cutter head torque lower limit threshold p1, the expected total thrust is increased accordingly to improve the excavation efficiency. The calculation formula is:

[0136] F1=F+k(p1-P) p (11)

[0137] Among them, F1 is the expected total thrust, P is the current cutterhead torque, F is the actual total working thrust of the shield machine, [p1, p2] is the cutterhead torque threshold range, p1 is the lower limit of the cutterhead torque threshold range, p2 is the upper limit of the cutterhead torque threshold range, and m is the geological adjustment index, which is the nonlinear adjustment coefficient for geological conditions.

[0138] When the cutter disc torque P is greater than the set cutter disc torque upper limit threshold p2, in order to prevent the cutter disc from being damaged due to excessive torque, the system can automatically reduce the expected total thrust F1. The calculation formula is:

[0139] F1=Fk(p-p2) n (12)

[0140] Among them, F1 is the expected total thrust, P is the current cutterhead torque, F is the actual total working thrust of the shield machine, [p1, p2] is the cutterhead torque threshold range, p1 is the lower limit of the cutterhead torque threshold range, p2 is the upper limit of the cutterhead torque threshold range, and n is the geological adjustment index, which is the nonlinear adjustment coefficient for geological conditions.

[0141] Embodiment 3

[0142] In some embodiments, adjusting the total propulsion thrust of the shield machine according to the thrust adjustment coefficient may include: obtaining the target cutter head torque and geological condition parameters when the cutter head torque exceeds the cutter head torque threshold range; optimizing the expected thrust adjustment coefficient based on the target cutter head torque and geological condition parameters; and adjusting the expected total propulsion thrust of the shield machine according to the adjusted expected thrust adjustment coefficient.

[0143] This embodiment is a method for adjusting the expected thrust adjustment coefficient based on an adaptive control strategy. The adaptive control can dynamically adjust the expected thrust adjustment coefficient output by the neural network according to the operating state of the shield machine and the geological conditions and environment, so that the control system is more adaptable. The calculation formula for optimizing the expected thrust adjustment coefficient based on the target cutter head torque and the geological condition parameters is as follows:

[0144]

[0145] Among them, k0 is the coefficient of thrust adjustment coefficient k after adaptive adjustment, P is the actual cutter head torque, P target is the target cutter head torque, i.e., the midpoint of the cutter head torque threshold range [p1, p2] (the midpoint of p1 and p2).

[0146] Furthermore, the expected total thrust of the shield machine is adjusted according to the adjusted expected thrust adjustment coefficient, and the calculation is as follows:

[0147]

[0148] Among them, F1 is the expected total thrust, and F is the actual working total thrust.

[0149] Embodiment 4

[0150] In some embodiments, adjusting the total propulsion thrust of the shield machine according to the thrust adjustment coefficient may include: when the cutter head torque value exceeds the cutter head torque threshold range, obtaining the target cutter head torque based on the error value between the target cutter head torque and the cutter head torque, and adjusting the expected total propulsion thrust of the shield machine through a PID closed-loop control coefficient algorithm and in combination with the expected thrust adjustment coefficient.

[0151] In some embodiments, the PID control strategy is a classic closed-loop control coefficient algorithm that can be used to dynamically adjust the control variable according to the system error, for example, according to the cutter head torque P and the target cutter head torque P t arg et The core idea of ​​PID control is to eliminate system errors by continuously adjusting the control variables so that the system output is stable near the target value. The specific calculation method is as follows:

[0152]

[0153] Where: k p , k i , k d is the initial PID parameter, k is the expected thrust adjustment coefficient output by the thrust adjustment coefficient calculation model, F1 is the expected total thrust, F is the actual working total thrust, P is the actual cutter head torque, P t arg et is the target cutter head torque, i.e., the midpoint of the cutter head torque threshold range [p1, p2] (the midpoint of p1 and p2).

[0154] It should be noted that the above description of the relevant process is only for example and explanation, and does not limit the scope of application of this specification. For those skilled in the art, various modifications and changes can be made to the process under the guidance of this specification. However, these modifications and changes are still within the scope of this specification.

[0155] Figure 6 is an exemplary structural diagram of a shield machine total thrust control system based on a neural network according to some embodiments of this specification. Figure 6 As shown, the shield machine total thrust control system 600 based on neural network includes:

[0156] The first data acquisition module 610 is used to acquire the expected total thrust of the shield machine;

[0157] The control module 620 is used to control the shield machine to perform propulsion operations according to the expected total propulsion thrust.

[0158] The second data acquisition module 630 is used to obtain the actual total working thrust and cutter head torque of the shield machine.

[0159] The thrust adjustment coefficient calculation module 640 is used to obtain the expected thrust adjustment coefficient according to the actual total working thrust, the cutter head torque and the cutter head torque threshold range.

[0160] The judgment module 650 is used to judge whether the cutter disc torque exceeds the cutter disc torque threshold range.

[0161] The expected total propulsion thrust adjustment module 660 is used to adjust the expected total propulsion thrust of the shield machine according to the expected thrust adjustment coefficient when the cutter head torque exceeds the cutter head torque threshold range.

[0162] The total thrust adjustment module 670 is used to control the shield machine to adjust the actual working total thrust according to the adjusted expected total thrust.

[0163] In some embodiments, the present application also provides a neural network-based shield machine total thrust control device, including a processor, and the processor is used to execute the above-mentioned neural network-based shield machine total thrust control method.

[0164] In some embodiments, the present application also provides a computer-readable storage medium, which stores computer instructions. When the computer reads the computer instructions in the storage medium, the computer executes the above-mentioned neural network-based shield machine total thrust control method.

[0165] The basic concepts have been described above. Obviously, for those skilled in the art, the above detailed disclosure is only for example and does not constitute a limitation of this specification. Although not explicitly stated here, those skilled in the art may make various modifications, improvements and corrections to this specification. Such modifications, improvements and corrections are suggested in this specification, so such modifications, improvements and corrections still belong to the spirit and scope of the exemplary embodiments of this specification.

[0166] At the same time, this specification uses specific words to describe the embodiments of this specification. For example, "one embodiment", "an embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of this specification. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. In addition, certain features, structures or characteristics in one or more embodiments of this specification can be appropriately combined.

[0167] In addition, unless explicitly stated in the claims, the order of the processing elements and sequences described in this specification, the use of alphanumeric characters, or the use of other names are not intended to limit the order of the processes and methods of this specification. Although the above disclosure discusses some invention embodiments that are currently considered useful through various examples, it should be understood that such details are only for illustrative purposes, and the attached claims are not limited to the disclosed embodiments. On the contrary, the claims are intended to cover all modifications and equivalent combinations that are consistent with the essence and scope of the embodiments of this specification. For example, although the system components described above can be implemented by hardware devices, they can also be implemented only by software solutions, such as installing the described system on an existing server or mobile device.

[0168] Similarly, it should be noted that in order to simplify the description disclosed in this specification and thus help understand one or more embodiments of the invention, in the above description of the embodiments of this specification, multiple features are sometimes combined into one embodiment, figure or description thereof. However, this disclosure method does not mean that the features required by the subject matter of this specification are more than the features mentioned in the claims. In fact, the features of the embodiments are less than all the features of the single embodiment disclosed above.

[0169] In some embodiments, numbers describing the number of components and attributes are used. It should be understood that such numbers used in the description of the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Unless otherwise specified, "about", "approximately" or "substantially" indicate that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which may change according to the required features of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and adopt the general method of retaining digits. Although the numerical domains and parameters used to confirm the breadth of their range in some embodiments of this specification are approximate values, in specific embodiments, the setting of such numerical values ​​is as accurate as possible within the feasible range.

[0170] Each patent, patent application, patent application publication, and other materials, such as articles, books, specifications, publications, documents, etc., cited in this specification are hereby incorporated by reference in their entirety. Except for application history documents that are inconsistent with or conflicting with the contents of this specification, documents that limit the broadest scope of the claims of this specification (currently or later attached to this specification) are also excluded. It should be noted that if the descriptions, definitions, and / or use of terms in the materials attached to this specification are inconsistent or conflicting with the contents described in this specification, the descriptions, definitions, and / or use of terms in this specification shall prevail.

[0171] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other variations may also fall within the scope of this specification. Therefore, as an example and not a limitation, alternative configurations of the embodiments of this specification may be considered consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.

Claims

1. A total thrust control method for a shield machine based on a neural network, characterized in that: include: Obtain the expected total thrust of the shield machine; Controlling the shield machine to perform propulsion operations according to the expected total propulsion thrust; Obtaining the actual total working thrust and cutter head torque of the shield machine; According to the actual working total thrust, the cutter head torque and the cutter head torque threshold range, obtaining the expected thrust adjustment coefficient; judging whether the cutter head torque exceeds the cutter head torque threshold range; When the cutter head torque exceeds the cutter head torque threshold range, adjusting the expected total thrust of the shield machine according to the thrust adjustment coefficient; The shield machine is controlled to adjust the actual working total thrust according to the adjusted expected total thrust.

2. The total thrust control method of a shield machine based on a neural network according to claim 1 is characterized in that: The obtaining of the expected thrust adjustment coefficient comprises: Inputting the actual working total thrust, the cutter head torque and the cutter head torque threshold range into a thrust adjustment coefficient calculation model to obtain a desired thrust adjustment coefficient, wherein the thrust adjustment coefficient calculation model is obtained by training an initial thrust adjustment coefficient calculation model; The thrust adjustment coefficient calculation model includes an input layer, a multi-head self-attention layer, a multi-layer perceptron and an output layer; wherein the calculation process of the thrust adjustment coefficient calculation model includes: The input layer uses the actual working total thrust, the cutter head torque and the cutter head torque threshold range as input features to form a four-dimensional input vector; The multi-head self-attention layer obtains a multi-head attention representation of the importance of each input feature by performing multiple attention calculations on the input vector; The multi-layer perceptron takes the multi-head attention representation as input and calculates the nonlinear relationship representation of each input feature through multiple nonlinear processing units; The output layer calculates the desired thrust adjustment coefficient based on the weight matrix and the bias term.

3. The total thrust control method of a shield machine based on a neural network according to claim 2 is characterized in that: The initial thrust adjustment coefficient calculation model training includes: Constructing a training data set of the shield machine, including total thrust, cutterhead torque, cutterhead torque threshold, and ideal thrust adjustment coefficient in historical mining data; Inputting the training data set into an initial thrust adjustment coefficient calculation model to obtain a thrust adjustment coefficient output by the model; According to the ideal thrust adjustment coefficient and the thrust adjustment coefficient output by the model, a loss function is used to adjust the network weight and the bias coefficient; When the error between the ideal thrust adjustment coefficient and the thrust adjustment coefficient output by the model is less than a threshold, a trained thrust adjustment coefficient calculation model is obtained.

4. The total thrust control method of a shield machine based on a neural network according to claim 1, characterized in that: When the cutter head torque exceeds the cutter head torque threshold range, adjusting the expected total thrust of the shield machine according to the expected thrust adjustment coefficient includes: When the cutter disc torque is less than the lower limit of the cutter disc torque threshold range, the expected total propulsion thrust is increased: F1=F+k(p1-P); When the cutter disc torque is greater than the upper limit of the cutter disc torque threshold range, reducing the expected total propulsion thrust: F1=Fk(P-p2), Among them, F1 is the expected total thrust, P is the current cutterhead torque, F is the actual total working thrust of the shield machine, [p1, p2] is the cutterhead torque threshold range, p1 is the lower limit of the cutterhead torque threshold range, and p2 is the upper limit of the cutterhead torque threshold range.

5. The total thrust control method of a shield machine based on a neural network according to claim 1, characterized in that: Also includes: wherein, when the cutter head torque exceeds the cutter head torque threshold range, a geological adjustment index is set based on geological conditions; The total thrust of the shield machine is adjusted in combination with the geological adjustment index and according to the expected thrust adjustment coefficient.

6. The total thrust control method of a shield machine based on a neural network according to claim 1, characterized in that: Also includes: When the cutter head torque exceeds the cutter head torque threshold range, obtaining a target cutter head torque and a geological condition parameter; Optimizing the desired thrust adjustment coefficient based on the target cutterhead torque and geological condition parameters; The expected total propulsion thrust of the shield machine is adjusted according to the adjusted expected thrust adjustment coefficient.

7. The total thrust control method of a shield machine based on a neural network according to claim 1, characterized in that: Also includes: When the cutter head torque value exceeds the cutter head torque threshold range, obtaining a target cutter head torque; According to the error value between the target cutter head torque and the cutter head torque, the expected total thrust of the shield machine is adjusted through a PID closed-loop control coefficient algorithm in combination with the expected thrust adjustment coefficient.

8. A total thrust control system for a shield machine based on a neural network, characterized in that: include: The first data acquisition module is used to obtain the expected total thrust of the shield machine; A control module, used for controlling the shield machine to perform a propulsion operation according to the expected total propulsion thrust; The second data acquisition module is used to obtain the actual working total thrust and cutter head torque of the shield machine; A thrust adjustment coefficient calculation module, used for obtaining a desired thrust adjustment coefficient according to the actual total working thrust, the cutter head torque and the cutter head torque threshold range; A judgment module, used for judging whether the cutter disc torque exceeds the cutter disc torque threshold range; An expected total propulsion thrust adjustment module, used for adjusting the expected total propulsion thrust of the shield machine according to the expected thrust adjustment coefficient when the cutter head torque exceeds the cutter head torque threshold range; The total thrust adjustment module is used to control the shield machine to adjust the actual working total thrust according to the adjusted expected propulsion total thrust.

9. A shield machine total thrust control device based on a neural network, comprising a processor, wherein the processor is used to execute the shield machine total thrust control method based on a neural network according to any one of claims 1 to 7.

10. A computer-readable storage medium storing computer instructions. When a computer reads the computer instructions in the storage medium, the computer executes the shield machine total thrust control method based on a neural network as claimed in any one of claims 1 to 7.

Images