Top drive torque adjusting method and system

Abstract

The invention discloses a top drive torque adjusting method and system, and belongs to the technical field of oil and gas drilling equipment, and the method comprises the steps: constructing an initial neural network model, and carrying out the training through the working condition data and operation torque data of a historical time period, and obtaining a trained neural network model; and outputting a real-time operation torque based on the trained neural network model. The mapping relation between the working condition and the torque is constructed through a device learning algorithm, the torque output of the top drive can be automatically adjusted according to the current working condition and the operation requirement, the driving torque of the top drive is adjusted according to the predicted torque value in the screwing-on and screwing-off process, the stability and safety of the buckling process are ensured, and the working efficiency is improved in the continuous operation process. And according to the predicted torque change trend, the output torque of the top drive is adjusted in real time so as to cope with working condition factors such as stratum hardness and working condition changes.

Description

Technical Field

[0001] The present application belongs to the technical field of oil and gas drilling equipment, and specifically relates to a top drive torque adjustment method and system. Background Art

[0002] In the exploration and development of oil and gas, the top drive drilling rig is one of the core equipment of modern drilling technology. Its performance is directly related to the drilling efficiency, safety and cost-effectiveness. The top drive system realizes multiple functions such as torque transmission, drilling pressure application, drill bit rotation and drill pipe connection and disconnection during the drilling process through integrated design. The top drive drilling rig is mainly composed of motor, reduction gearbox, torque transmission mechanism, main shaft, drilling control system and other parts. Its core function is to efficiently convert the rotational power of the motor into the drilling torque of the drill bit.

[0003] When the traditional top drive system is operating, the output torque adjustment of the motor mostly depends on the operating experience and cannot meet the complex and changeable drilling conditions. Summary of the invention

[0004] The purpose of the present application is to provide a top drive torque adjustment method and system to solve the problem in the above background technology that the existing top drive torque is difficult to adapt to the drilling conditions.

[0005] To achieve the above objectives, the present application provides the following technical solutions: a top drive torque adjustment method, comprising:

[0006] Obtain the working condition data and operating torque data of the historical time period;

[0007] Constructing an initial neural network model, taking the working condition data of the historical time period as the input of the initial neural network model, taking the working torque data as the output of the initial neural network model, training the initial neural network model, and obtaining a trained neural network model;

[0008] Taking the real-time working condition data as input, the real-time operating torque data is obtained through the trained neural network model, and the output torque of the motor is adjusted based on the real-time operating torque data.

[0009] Furthermore, the operating condition data include drilling pressure, bottom hole speed, mud flow, drilling tool specifications, well depth, formation characteristics, tool model and circulating pressure, and the number of input layer nodes of the initial neural network model is the same as the number of operating condition data.

[0010] Furthermore, the operating torque data includes make-up and break-out torque and continuous operating torque, and the number of output layer nodes of the initial neural network model is the same as the number of the operating torque data.

[0011] Furthermore, the initial neural network model has a first hidden layer, a second hidden layer and a third hidden layer, and the number of neurons in the first hidden layer, the second hidden layer and the third hidden layer decreases in sequence.

[0012] Furthermore, the number of neurons in the first hidden layer is 120 to 140, and the ReLU function is used as the activation function, the number of neurons in the second hidden layer is 28 to 36, and the Sigmoid function is used as the activation function, and the number of neurons in the third hidden layer is 12 to 24, and the Tanh function is used as the activation function.

[0013] Furthermore, the adjustment method also includes standardizing the operating condition data.

[0014] Furthermore, the output torque of the motor is adjusted by a frequency converter.

[0015] Furthermore, adjusting the output torque of the motor based on the real-time operating torque data includes: increasing the input voltage of the motor and reducing the input power frequency of the motor to increase the output torque of the motor.

[0016] Furthermore, the regulation method also includes monitoring the operating temperature and operating status of the motor, and cutting off the power supply of the motor or reducing the output torque of the motor when the operating temperature of the motor exceeds a set value and / or the motor is overloaded.

[0017] On the other hand, the present application discloses a top drive torque adjustment system, comprising:

[0018] An acquisition module configured to acquire operating condition data and operating torque data of a historical time period;

[0019] A construction module configured to construct an initial neural network model, using the working condition data of a historical time period as the input of the initial neural network model and the working torque data as the output of the initial neural network model, and training the initial neural network model to obtain a trained neural network model;

[0020] The adjustment module is configured to take the real-time working condition data as input, obtain the real-time working torque data through the trained neural network model, and adjust the output torque of the motor based on the real-time working torque data.

[0021] Compared with the prior art, the beneficial effects of this application are:

[0022] This application constructs a mapping relationship between working conditions and torque through a machine learning algorithm, and can automatically adjust the torque output of the top drive according to the current working conditions and operating requirements. During the make-up and make-out process, the driving torque of the top drive is adjusted according to the predicted torque value to ensure the smoothness and safety of the fastening process. During continuous operation, the output torque of the top drive is adjusted in real time according to the predicted torque change trend to cope with working conditions such as formation hardness and changes in operating conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 This is a flow chart of the application method;

[0024] Figure 2 The speed torque curve of light overload is controlled by frequency;

[0025] Figure 3 This is the current-time curve of high torque continuous drilling mode;

[0026] Figure 4 This is the current-time curve of the shackle operation mode. DETAILED DESCRIPTION

[0027] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0028] A top drive torque adjustment method, referring to Figure 1 ,include:

[0029] S100: Acquire the working condition data and operation torque data of the historical time period,

[0030] S200: constructing an initial neural network model, taking the working condition data of the historical time period as the input of the initial neural network model, taking the working torque data as the output of the initial neural network model, training the initial neural network model, and obtaining a trained neural network model;

[0031] S300: Using the real-time working condition data as input, obtaining the real-time operating torque data through the trained neural network model, and adjusting the output torque of the motor based on the real-time operating torque data.

[0032] Specifically, the above-mentioned operating condition data include drilling pressure (P), bottom hole speed (N), mud flow (Q), drilling tool specifications (D), well depth (H), formation characteristics (F), tool model (T) and circulating pressure (Cp). The operating condition data are mainly obtained through sensors (such as pressure sensors) arranged in the top drive system. Correspondingly, the number of input layer nodes of the above-mentioned initial neural network model corresponds to the number of operating condition data, that is, the number of input layer nodes of the initial neural network model is 8. At the same time, when the operating condition data is input into the initial neural network model, it is necessary to first standardize the operating condition data to eliminate dimensional differences and improve the training efficiency of the initial neural network model.

[0033] The above-mentioned operating torque data includes make-up and breakout torque and continuous operating torque. The operating torque data is mainly measured by a torque sensor arranged in the top drive system. Correspondingly, the number of output layer nodes of the above-mentioned initial neural network model corresponds to the operating torque data, that is, the number of output layer nodes of the initial neural network model is 2.

[0034] Exemplarily, the above initial neural network model is based on radial basis function neural network (RBF) to realize the processing of nonlinear and high-dimensional data.

[0035] The initial neural network model includes at least one hidden layer. In some embodiments, the initial neural network model includes a first hidden layer, a second hidden layer, and a third hidden layer. The confirmation process of the relevant parameters of the first hidden layer, the second hidden layer, and the third hidden layer is as follows:

[0036] The first hidden layer is the initial stage of feature extraction and is mainly responsible for identifying basic features from input parameters. Considering the diversity and complexity of input parameters, this layer sets a relatively large number of neurons to fully capture the detailed information of the input data. Specifically, the first hidden layer can be designed to contain 120 to 140 neurons (nodes) to ensure that the initial neural network model has sufficient flexibility and expressiveness when processing complex input data. For example, the number of neurons in the first hidden layer is 128, and the first hidden layer uses the ReLU (Rectified Linear Unit) function as the activation function, with the formula f(x) = max(0,x). At this time, the output value calculation formula of each neuron in the first hidden layer (h1_i, i = 1, 2..., 128) is as follows:

[0037] h1_i=σ(W1·X+b1_i)

[0038] Where W1 is the weight matrix from the input layer to the first hidden layer, X is the vector of the input layer (containing 8 parameters), b1_i is the bias term of the i-th neuron in the first hidden layer, and σ is the ReLU activation function.

[0039] The second hidden layer is used for dimensionality reduction and integration of features. Specifically, the second hidden layer further abstracts and combines the features extracted by the first hidden layer to form a higher-level feature representation. The second hidden layer achieves dimensionality reduction and integration of features by reducing the number of neurons, and uses nonlinear activation functions (such as ReLU, Sigmoid or Tanh) to enhance the nonlinear expression ability of the model. Specifically, the number of neurons in the second hidden layer is 28 to 36. Exemplarily, the second hidden layer can be designed to contain 32 neurons. This design can reduce computational complexity and overfitting risks while maintaining model performance.

[0040] The input variables of the second hidden layer are the neuron outputs of the first hidden layer, i.e., h1_1, h1_2, ..., h1_128.

[0041] The activation function is the Sigmoid function, and the formula is f(x)=1 / (1+exp(-x)).

[0042] Output variables: The output values ​​of 32 neurons, denoted as h2_1, h2_2, ..., h2_32. Each neuron in the second hidden layer (h2_i, i = 1, 2, ..., 32) can be calculated by the following formula:

[0043] h2_i=σ(W2·H1+b2_i)

[0044] Where W2 is the weight matrix from the first hidden layer to the second hidden layer, H1 is the output vector of the first hidden layer (containing 128 elements), b2_i is the bias term of the i-th neuron in the second hidden layer, and σ is the Sigmoid activation function.

[0045] The third hidden layer serves as a bridge between input features and final outputs and is mainly responsible for mapping high-level features to the output space. This layer can learn the mapping relationship from input parameters to make-up and make-out torque and continuous output torque to achieve accurate prediction. Specifically, the third hidden layer can be designed to include 12 to 24 neurons. Exemplarily, the third hidden layer can include 16 neurons. This design can further reduce the computational complexity and overfitting risk while maintaining the prediction accuracy of the model.

[0046] Input variables: Output values ​​h2_1, h2_2, ..., h2_32 from the 32 neurons in the second hidden layer.

[0047] Activation function: Tanh function, the formula is f(x)=(exp(x)-exp(-x)) / (exp(x)+exp(-x)).

[0048] Output variables: The output values ​​of 32 neurons, denoted as h3_1, h3_2, ..., h3_32.

[0049] The output value of each neuron (h3_i (i = 1, 2, ..., 32)) can be calculated by the following formula:

[0050] h3_i=σ(W3·H2+b3_i)

[0051] Where W3 is the weight matrix from the second hidden layer to the third hidden layer, H2 is the output vector of the second hidden layer (containing 32 elements), b3_i is the bias term of the i-th neuron in the third hidden layer, and σ is the Tanh activation function.

[0052] The output layer contains two variables: make-break torque (Tmake-break) and continuous output torque (Tcontinuous), and the expressions and formulas are:

[0053]

[0054] in, and are the weights of the shackle torque and continuous output torque from the third hidden layer to the output layer, and are the bias terms respectively.

[0055] In some embodiments, in order to achieve dynamic regulation of torque, a self-learning mechanism is introduced into the above-mentioned neural network model, which can learn new data during drilling operations online and continuously update model parameters, thereby improving the accuracy and adaptability of torque prediction. For example, sliding window technology can be used to collect operation data in real time and update the training set. At the same time, the Levenberg-Marquardt (LM) algorithm is used to optimize the neural network parameters to improve the convergence speed and prediction accuracy of the model.

[0056] The output torque regulation of the above motor is realized by the frequency converter. Specifically, the calculation formula of the power of the three-phase asynchronous motor is as follows:

[0057] P=√3U·I·cosj

[0058] P is the active power input to the motor, U is the line voltage of the motor power input, I is the line current of the motor power input, cosj is the power factor of the motor, and j is the phase difference angle between the phase voltage and the phase current.

[0059] It can be seen from the above calculation formula that by increasing the inverter output current (line current of the motor power input), the output power of the top drive motor can be increased, thereby increasing the output torque of the motor. That is, the inverter changes the frequency and voltage of the motor input power supply, thereby changing the speed and torque output characteristics of the motor. At this time, the system can quickly adjust the motor input power supply parameters through the inverter based on the drilling parameters and real-time feedback of the motor status information to achieve the output torque regulation of the motor. For example, at low frequency, by increasing the current, a large torque output can be achieved (refer to Figure 2 , which is a schematic diagram of the motor torque and speed curve under different overload conditions); at high frequencies, the stable operation of the motor can be maintained by reducing the current. This control method enables the motor to flexibly adjust the torque output without replacing or upgrading the hardware, and can ensure that the motor maintains a high torque output state for a long time, thereby improving drilling efficiency. In top drive drilling operations, the core technical indicators related to top drive torque include top drive continuous drilling torque and short-time output torque. Continuous drilling torque is the main driving force for drilling and rock breaking. When drilling in hard rock formations and other formations, continuous drilling torque must be provided to overcome the rock breaking friction at the bottom of the well; short-time output torque mainly acts in the top drive shackle operation, and it is necessary to overcome the tightening torque between the main shaft, upper and lower cocks, and the protection joint. Its torque is often much greater than the load during normal drilling operations. In response to the torque adjustment requirements of continuous drilling operations and top drive shackle operations, targeted control is performed through the frequency converter. When the top drive enters high-torque continuous drilling operations, refer to Figure 3 Based on the real-time torque data, the inverter quickly adjusts the input power parameters of the motor, for example, controlling the input current to increase to 1.1 times the rated current within a certain period of time, so that the motor outputs a continuous drilling torque exceeding its rated torque at a low speed. During the shackle operation, refer to Figure 4 , the top drive needs to provide huge torque in a short period of time to overcome the resistance between drilling tools. At this time, the input current can be controlled to increase to 2 times the rated current within a certain period of time, that is, the system realizes low-frequency and high-current output (2 times the rated current) through the inverter, so that the motor can reach and maintain the required high-torque state in a very short time. This dynamic adjustment process not only improves the efficiency of punching and buckling operations, but also significantly reduces operational risks and ensures the safety of drilling operations. It can effectively improve the torque output during the buckling process without changing the hardware structure of the top drive, which has become a technical problem that needs to be solved urgently.

[0060] At the same time, in order to ensure the stability and safety of the top drive system during the torque increase and regulation process, real-time monitoring and intelligent analysis of the motor temperature are required when adjusting the output torque of the motor. The system can detect and prevent potential faults in time, ensure the reliable operation of the motor under high torque output conditions, and set overload protection at the same time: the PLC control system monitors the operating status of the motor in real time. Once the motor is found to be overloaded, the power supply is immediately cut off or the torque output is reduced to avoid damage to the motor due to long-term overload. The implementation of overload protection depends on accurate torque sensors and efficient control systems.

[0061] For example, taking more than 60 top drives used by a company in a certain area as an example, the traditional DQ70BSC top drive outputs a continuous drilling torque of 50kNm, which basically meets the needs of fast and stable drilling on site in conventional drilling operations. In recent years, as drilling has become deeper and more complex, under complex geological conditions, the output torque of traditional top drives cannot meet actual needs. For example, when drilling in hard rock formations, the top drive system needs to be able to quickly and accurately increase the output torque. According to the torque prediction results based on the neural network model and actual operation requirements, the continuous drilling torque requirements of its drilling operations may exceed 50kNm, resulting in the existing top drive output torque value being difficult to meet actual needs. On the premise of not changing or less changing the original hardware conditions of the top drive, the structure, performance, reliability and other factors of the equipment are comprehensively considered, and the top drive output drilling torque is improved through the optimization of the transmission system and the improvement of the control system to overcome the drilling resistance and reduce the risk of stuck drill.

[0062] In some examples, in a top drive system, a torque limiter is installed between the motor and the reducer, and a torque threshold is set. When the torque applied to the transmission system exceeds this threshold, the torque limiter automatically disconnects the power transmission, thereby protecting the motor and the reducer from damage.

[0063] Flexible soft start and torque resistance control: When the drive motor starts, the flexible soft start function is realized to slowly increase the drive torque resistance to the required strength to prevent the sudden increase in torque resistance from causing fatal effects on the motor. At the same time, once the torque resistance is overloaded, the transmission phenomenon is interrupted in time through a specific control mechanism to prevent motor damage.

[0064] On the other hand, the present application discloses a top drive torque adjustment system, comprising:

[0065] An acquisition module configured to acquire operating condition data and operating torque data of a historical time period;

[0066] A construction module configured to construct an initial neural network model, using the working condition data as the input of the initial neural network model and the working torque data as the output of the initial neural network model, and training the initial neural network model to obtain a trained neural network model;

[0067] The adjustment module is configured to use the real-time working condition data as input, obtain the real-time working torque data through the neural network model data, and adjust the output torque of the motor based on the real-time working torque data

[0068] Although the embodiments of the present application have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present application, and that the scope of the present application is defined by the appended claims and their equivalents.

Claims

1. A top drive torque adjustment method, characterized in that: include: Obtain the working condition data and operating torque data of the historical time period; Constructing an initial neural network model, taking the working condition data of the historical time period as the input of the initial neural network model, taking the working torque data as the output of the initial neural network model, training the initial neural network model, and obtaining a trained neural network model; Taking the real-time working condition data as input, the real-time operating torque data is obtained through the trained neural network model, and the output torque of the motor is adjusted based on the real-time operating torque data.

2. A top drive torque adjustment method according to claim 1, characterized in that: The operating condition data include drilling pressure, bottom hole speed, mud flow, drilling tool specifications, well depth, formation characteristics, tool model and circulating pressure. The number of input layer nodes of the initial neural network model is the same as the number of operating condition data.

3. A top drive torque adjustment method according to claim 1, characterized in that: The operating torque data includes make-up and break-out torque and continuous operating torque, and the number of output layer nodes of the initial neural network model is the same as the number of the operating torque data.

4. A top drive torque adjustment method according to claim 1, characterized in that: The initial neural network model comprises a first hidden layer, a second hidden layer and a third hidden layer, and the number of neurons in the first hidden layer, the second hidden layer and the third hidden layer decreases in sequence.

5. A top drive torque adjustment method according to claim 4, characterized in that: The number of neurons in the first hidden layer is 120 to 140, and the ReLU function is used as the activation function. The number of neurons in the second hidden layer is 28 to 36, and the Sigmoid function is used as the activation function. The number of neurons in the third hidden layer is 12 to 24, and the Tanh function is used as the activation function.

6. A top drive torque adjustment method according to claim 1, characterized in that: The adjustment method also includes standardizing the operating condition data.

7. A top drive torque adjustment method according to claim 1, characterized in that: The output torque of the motor is regulated by the frequency converter.

8. A top drive torque adjustment method according to claim 1, characterized in that: Adjusting the output torque of the motor based on the real-time operating torque data includes: increasing the input voltage of the motor and reducing the input power frequency of the motor to increase the output torque of the motor.

9. A top drive torque adjustment method according to claim 1, characterized in that: The regulation method further includes monitoring the operating temperature and operating state of the motor, and cutting off the power supply of the motor or reducing the output torque of the motor when the operating temperature of the motor exceeds a set value and / or the motor is overloaded.

10. A top drive torque adjustment system, characterized in that: include: An acquisition module configured to acquire operating condition data and operating torque data of a historical time period; A construction module configured to construct an initial neural network model, using the working condition data of a historical time period as the input of the initial neural network model and the working torque data as the output of the initial neural network model, and training the initial neural network model to obtain a trained neural network model; The adjustment module is configured to take the real-time working condition data as input, obtain the real-time working torque data through the trained neural network model, and adjust the output torque of the motor based on the real-time working torque data.

Images