Electricity-to-oil power optimization control method and system for oil drilling in complex stratum

Abstract

The invention discloses an electricity-to-oil power optimization control method and system for oil drilling in a complex stratum, and belongs to the field of drilling safety prevention and control. The method comprises the steps that multi-source heterogeneous data in the oil drilling process are collected, and in an edge gateway controller, the multi-source heterogeneous data are stored in a database; carrying out fusion processing on the collected multi-source heterogeneous data to obtain a fused multi-source heterogeneous data stream; the fused multi-source heterogeneous data flow is processed through a pre-trained long-short-term memory network, and a stratum load torque change trend sequence about to be encountered by the drill bit in a preset future time window is obtained; the prediction control algorithm obtains an optimal power pre-regulation instruction sequence according to the stratum load torque change trend sequence; and the edge gateway controller receives the instruction sequence and sends the instruction sequence to a drilling fluid pump control system and a heave compensation system of the drilling platform. By fusing and analyzing the multi-source heterogeneous data in real time, the safety and efficiency of drilling operation are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of drilling safety prevention and control, and particularly relates to a complex formation oil drilling electric-oil power optimization control method and system. BACKGROUND

[0002] With the popularization of electric-oil technology on oil drilling platforms, industrial power grids are used to drive high-power variable frequency motors to replace traditional diesel generators, so that the energy cost and environmental pollution of drilling operations have been significantly reduced. However, the existing electrically driven drilling power control system still has serious technical bottlenecks when dealing with complex formation drilling. In the traditional electrically driven drilling power control system, a passive adjustment strategy based on real-time working condition parameter feedback is usually used. This strategy monitors the speed in real time through an edge gateway controller, and when the speed drops due to hardening of the formation, the output power of the frequency converter is increased to try to restore the set speed through a PID algorithm or simple logic judgment. The essence of this control method is a post-remedy measure, which is a lagging control. Moreover, in the actual drilling process, the formation structure underground is often complex and variable, and hard rock interlayers, soft mudstone formations, fractured leakage formations, or sudden interfaces between different lithologies are often encountered. When the drill bit rotates at high speed and suddenly encounters these formation mutations, the downhole load will change dramatically within milliseconds.

[0003] In addition, in the traditional drilling platform control architecture, the power control system for driving the rotary table or top drive, the pump control system for controlling the drilling fluid circulation, and the heave compensation system for offshore platforms or specific working conditions are usually independent control units. There is a lack of high-speed and effective information exchange and collaborative decision-making mechanism between systems. It is impossible to form a globally optimal control strategy, and when dealing with complex working conditions, it often loses one and gains the other, which seriously restricts the overall efficiency and intelligent level of drilling operations. SUMMARY

[0004] One of the purposes of the present application is to provide a complex formation oil drilling electric-oil power optimization control method to solve the problem of load dramatic change causing speed and torque dramatic fluctuation caused by complex formation in the prior art.

[0005] The application is realized by the technical scheme, a complex formation oil drilling electric power optimization control method, comprising the following steps: S100, collecting multi-source heterogeneous data in the oil drilling process, and performing fusion processing on the collected multi-source heterogeneous data in the edge gateway controller to obtain a fused multi-source heterogeneous data stream; S200, processing the fused multi-source heterogeneous data stream through a pre-trained long short-term memory network to obtain a preset future time window, a formation load torque change trend sequence that the drill bit will encounter; S300, a predictive control algorithm obtains an optimal power pre-regulation instruction sequence under the condition of meeting the motor safe operation constraint, taking the minimization of power grid energy consumption fluctuation, drilling tool speed deviation and torque impact as the optimization target according to the formation load torque change trend sequence; S400, the edge gateway controller receives the load torque change trend sequence and the power pre-regulation instruction sequence, takes the load torque change trend sequence as the cooperative control reference, and synchronously issues the power pre-regulation instruction sequence to the drilling fluid pump control system and the heave compensation system of the drilling platform, so that the drilling fluid pump control system and the heave compensation system execute the instruction sequence to realize the linkage optimization of the power system and the associated subsystem.

[0006] Further, the multi-source heterogeneous data includes: logging while drilling forward-looking data containing formation lithology parameters and formation drillability parameters; drilling real-time working condition data containing drilling pressure, torque and speed; power system electrical parameter data containing main drive frequency converter output current, voltage and power.

[0007] Further, the fusion processing includes: unifying the multi-source heterogeneous data to a common time reference to realize timestamp alignment; removing abnormal values and noise interference in the data through abnormal value detection to realize data abnormal value cleaning; converting data with different physical units and dimensions into data with the same scale through the Z-Score algorithm to eliminate the dimension influence.

[0008] Further, the edge gateway controller can further include a preset space-time mapping model, which can calculate in real time the formation characteristics detected by the LWD in front of the drill bit when the formation characteristics will reach the drill bit position, so as to realize the accurate alignment of the LWD forward-looking data in the time dimension, accurately and dynamically map the discrete formation information measured by the LWD in the spatial domain and the downhole depth to the time domain and the future time point required by the controller. The edge gateway controller can convert the spatial information that there is hard rock X meters in front of the drill bit into the time information that the load will sharply increase Y seconds later, which can be used for executing forward-looking control.

[0009] Further, the space-time mapping model can be as follows:

[0010] ,

[0011] wherein, t is the current time, representing the real-time system time when the calculation is performed; LWD is the LWD measurement depth, i.e. the absolute depth at which the LWD sensor measures a particular formation parameter; TLWD is the predicted arrival time, which is the future time point at which the drill bit is predicted to reach the depth of the LWD measurement point; t is the current time, representing the real-time system time when the calculation is performed; LWD is the LWD measurement depth, i.e. the absolute depth at which the LWD sensor measures a particular formation parameter; LWD is the LWD measurement depth, i.e. the absolute depth at which the LWD sensor measures a particular formation parameter; PDR is the predicted drilling rate function, which is a key embedded function, taking the downhole depth (between the current drill bit depth and the LWD measurement depth) as input, and outputting the predicted mechanical drilling rate at that depth; d is the depth infinitesimal, i.e. an infinitesimal depth increment along the integration path.

[0012] Further, the pre-trained long short-term memory network, the pre-trained data set is the historical data from the drilled section of the well, and / or the historical database of the adjacent drilled well, the long short-term memory network further includes, online updating, by comparing the actually measured torque value with the predicted value of the model, calculating the prediction error, and taking the error as a correction signal, through the back propagation algorithm of small learning rate, online fine-tuning the network weight of the long short-term memory network.

[0013] Further, the predictive control algorithm includes an optimization objective function for balancing the contradictions among performance, energy consumption and equipment smoothness, and constraint conditions for constraining the optimization objective function state.

[0014] Further, the optimization objective function is as follows:

[0015] ,

[0016] wherein, J is the objective function, x is the state vector at the current time, T is the motor output torque, T is the matrix transpose symbol; T is the entire control sequence in the future control time domain; T is a single time step; T is the prediction time domain, T is the discrete time step at the current control time; r is the reference value or desired trajectory of the state variable; W is the weight matrix of state tracking, for control in time domain, is the control input vector of current time; is the weight matrix of energy consumption control, is the change amount of control input, is the weight matrix of smooth control.

[0017] Further, the constraint conditions include:

[0018] System state space model: ;

[0019] State variable constraint: ;

[0020] Control input constraint: ;

[0021] Control input rate constraint: ;

[0022] wherein, is a discrete-time state space matrix of the drilling power system state, is a discrete-time state space matrix of the drilling power system control input, is a discrete-time state space matrix of the disturbance vector, is the measurable disturbance vector of current time; is the minimum value of state, is the maximum value of state; is the minimum value of control input, is the maximum value of control input; is the minimum value of control input change amount, is the maximum value of control input change amount.

[0023] Further, the collaborative regulation instruction of the drilling fluid pump control system is generated based on the lithology corresponding to the load torque change trend sequence: when it is predicted that the soft mudstone formation prone to mud cake is about to be entered, the collaborative instruction is to increase the drilling fluid discharge in advance; when it is predicted that the fractured formation prone to leakage is about to be entered, the collaborative instruction is to appropriately reduce the drilling fluid discharge; by embedding a lithology-drilling fluid strategy knowledge base in the drilling fluid pump control system, when the load torque change trend sequence is that the soft mudstone formation prone to mud cake will be entered in the future, the drilling fluid pump control system automatically queries the knowledge base and executes the collaborative instruction.

[0024] Further, the cooperative control instruction of the heave compensation system is generated based on the formation structure corresponding to the load torque change trend sequence, continuous short-time Fourier transform or fast Fourier transform is carried out through real-time torque data, when the analysis result shows that the signal energy continuously exceeds the preset threshold in the typical characteristic frequency band of stick-slip vibration of 0.1Hz-2Hz, it is determined that stick-slip vibration precursor exists; once the stick-slip vibration precursor is identified, the heave compensation system immediately triggers two parallel suppression actions: a high-frequency, low-amplitude sinusoidal disturbance signal is superimposed on the original speed setting value and is sent to the frequency converter; at the same time, a transient pressure increase instruction is sent to the drilling fluid pump control system, so that the displacement is increased by 5%-10% within 3-5 seconds.

[0025] In another aspect, the application provides a complex formation oil drilling electric oil power optimization control system, comprising a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the complex formation oil drilling electric oil power optimization control method as described above.

[0026] Compared with the prior art, the application has the following advantages and beneficial effects:

[0027] 1. The application realizes real-time monitoring and dynamic analysis of downhole conditions by collecting and fusing multi-source heterogeneous data in real time during drilling, overcomes the limitations of traditional well formation lithology information acquisition methods which rely on fixed data, and improves the accuracy and timeliness of formation condition identification.

[0028] 2. The application establishes a mapping model from the drill bit depth to the geological parameters based on the LWD forward-looking data, realizes accurate prediction and dynamic monitoring of complex formation conditions, and effectively solves the problem of lack of dynamic adaptability to complex formation conditions in the risk prediction of well control in the prior art.

[0029] 3. The application improves the prediction accuracy and reliability of well formation load torque changes by using a pre-set long short-term memory network combined with an online updating mechanism of actual measurement data, overcomes the poor adaptability of existing prediction models under different mine layer conditions, and realizes intelligent cooperative control of the drilling fluid pump control system and the heave compensation system based on the load torque change trend sequence as the cooperative control reference and the power pre-regulation instruction sequence generated by the prediction control algorithm, thereby improving the flexibility and response speed of the well control system in response to complex formation condition changes.

[0030] 4. The application establishes a complete downhole condition monitoring and prediction system by fusing and analyzing multi-source heterogeneous data in real time, provides a scientific basis for safety decision-making in the oil drilling process, effectively reduces the risk of overflow well blowout and other well control accidents, and improves the safety and efficiency of drilling operations. BRIEF DESCRIPTION OF DRAWINGS

[0031] The accompanying drawings, which are included to provide a further understanding of the embodiments of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the principles of the application. In the drawings:

[0032] Figure 1 The method flow chart provided for Embodiment 1 of the application.

[0033] Figure 2 The timing diagram provided for Embodiment 1 of the application. DETAILED DESCRIPTION

[0034] In order to make the objects, technical solutions and advantages of the embodiments of the application clearer, the technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are some but not all of the embodiments of the application. The components of the embodiments of the application described and shown in the drawings herein can be arranged and designed in various different configurations.

[0035] Embodiment 1

[0036] The embodiment discloses a complex formation oil drilling electric oil power optimization control method, Figure 1 The overall flow chart of the embodiment is shown, Figure 2 The overall timing diagram of the embodiment is shown, from Figure 1 It can be seen from the above that the embodiment includes the following steps:

[0037] Step 1: First, relevant data in the oil drilling process is collected, and the edge gateway controller performs real-time fusion on the collected relevant data to obtain a fused data stream. The edge gateway controller sends the data stream to a background computer for data processing.

[0038] Specifically, the relevant data includes: logging while drilling (LWD) forward-looking data containing formation lithology parameters and formation drillability parameters; drilling real-time working condition data containing drilling pressure, torque and rotating speed; power system electric parameter data containing main drive frequency converter output current, voltage and power.

[0039] Real-time fusion of the above-mentioned multi-source heterogeneous data includes: first, multi-source heterogeneous data of different sources and different frequencies are unified to a common time reference through interpolation or alignment algorithm to realize timestamp alignment; then, through abnormal value detection, abnormal values and noise interference in the data are eliminated to realize data abnormal value cleaning, improve data quality and ensure the accuracy and reliability of subsequent analysis results; finally, through Z-Score algorithm, data of different physical units and dimensions are converted into data with the same scale to eliminate the dimension influence and complete the standardization processing of the data.

[0040] In this embodiment, a mapping model from the bit depth (spatial) to the geology parameter (time series prediction input) can also be established in the edge gateway controller based on the LWD look-ahead data. This model can be a look-up table based on the neighbor well data, or a dynamic mapping model. This helps the subsequent prediction model not only rely on what just happened, but also utilize the prior knowledge of what should happen at this depth.

[0041] For the LWD look-ahead data, since the LWD measures the formation information at a certain distance ahead of the bit, in order to achieve accurate prediction (look-ahead), it is necessary to know when these formation characteristics will reach the bit. Therefore, by pre-establishing a space-time mapping model of the bit position and the LWD measurement depth in the edge gateway controller, it can be calculated in real time according to the current drilling speed that the formation characteristics detected by the LWD ahead will reach the bit position at what time, so as to achieve accurate alignment of the look-ahead data in the time dimension.

[0042] Specifically, in this embodiment, the discrete formation information measured by the LWD in the spatial domain (downhole depth) is accurately and dynamically mapped to the time domain (future time point) required by the controller. This enables the edge gateway controller to convert the spatial information "X meters ahead of the bit has hard rock" into the time information "the load will increase sharply Y seconds later" which can be used to perform look-ahead control.

[0043] Exemplarily, in this embodiment, the space-time mapping model can be as follows:

[0044] ,

[0045] wherein, is the current time, representing the real-time system time when the calculation is performed; is the LWD measurement depth, that is, the absolute depth when the LWD sensor measures a specific formation parameter; represents the predicted arrival time, which is the future time point predicted when the bit drills to the depth of the LWD measurement point; is the current time, that is, the real-time system time when the calculation is performed; is the bit depth, represents the current bit depth, that is, the real-time absolute depth of the bit at the current time; is the predicted drilling speed function, which is a key embedded function, and the input is the downhole depth (between the current bit depth and the LWD measurement depth), and the output is the predicted mechanical drilling speed at the depth; is the depth infinitesimal, that is, an infinitesimal depth increment on the integral path.

[0046] It is noted that in the above equation, is the time reference of the whole prediction, the definite integral represents the total time needed to drill the distance from the current depth to the target depth; the function is a dynamic prediction of the future drilling rate, which is a function of the formation drillability parameters (such as rock strength, abrasiveness, etc.) measured by LWD at the depth, combined with the current drilling operation parameters (such as weight on bit, rotary speed). The inverse of this function, which is the inverse of the predicted drilling rate, represents the drilling lag rate or the time needed to drill one unit of depth (in seconds / meter) at the depth . If the formation gets harder, the predicted drilling rate will decrease, and the value of this term will increase, meaning that it will take longer to drill through this layer. The model calculates the time needed to drill through each infinitesimal depth segment by dividing the whole path between the current position of the drill bit and the LWD measurement point into an infinite number of infinitesimal depth segments, then calculating the time infinitesimal needed to drill through each infinitesimal depth segment, and finally summing up all these time infinitesimals. This sum is the total time needed to drill through the whole path.

[0047] Step 2: According to the fused data stream, the future trend sequence of the formation load torque encountered by the drill bit within a preset time window (T) is calculated and output by a preset load prediction model.

[0048] The load prediction model extracts the formation lithology parameters in the LWD forward-looking data, the real-time drilling operating data, and the power system electrical parameter data from the fused data stream, aligns and samples these data according to a preset time step, and constructs a multi-dimensional time sequence feature vector.

[0049] Then, the multi-dimensional time sequence feature vector is input into a pre-trained long short-term memory network (LSTM), and the output obtained is the trend sequence of the formation load torque within the future preset time window (T).

[0050] Specifically, the pre-trained long short-term memory network in this embodiment can use a data set from the historical data of the drilled section of the well, or from the historical database of the adjacent drilled well, which contains rich information of the formation and the corresponding drilling parameters. LSTM is selected because the drilling process is a typical time sequence process, and the current load in the drilling process is related to the formation and operating conditions in a long period of time in the past. The gating mechanism of LSTM can effectively capture the long-term dependence relationship in the time sequence data.

[0051] In addition, after actual deployment, the LSTM can also be updated online. The actual measured torque value is compared with the predicted value of the model, the prediction error is calculated, and the error is used as a correction signal to fine-tune the network weights of the LSTM model online through a small learning rate back propagation algorithm, so that the LSTM model can continuously optimize itself and maintain high prediction accuracy.

[0052] Step 3: The prediction control algorithm obtains an optimal power pre-adjustment instruction sequence based on the output formation load torque trend sequence, with the optimization objectives of minimizing power grid energy consumption fluctuation, drilling tool speed deviation, and torque impact, and under the condition of meeting the motor safe operation constraint.

[0053] The instruction sequence and the formation load torque trend sequence are sent to the edge gateway controller to adjust the output power in advance, so that the edge gateway controller actively adapts to the upcoming formation load change with the optimal power output.

[0054] Specifically, in the present embodiment, the prediction control algorithm converts the complex multi-objective dynamic control problem into a series of well-conditioned quadratic programming problems solved repeatedly in a limited time domain, thereby generating proactive and forward-looking power pre-adjustment instructions. By looking forward to a limited time domain at each control time, and then solving a constrained optimization problem to obtain the optimal control input at the current time.

[0055] Exemplarily, in the present embodiment, the optimization objective function can be as follows:

[0056]

[0057] ,

[0058] wherein, is the objective function, is the state vector at the current time, for example, , is the motor speed, is the motor output torque, is the matrix transpose symbol. is the entire control sequence in the future control time domain; is a single time step; is the prediction time domain, indicating how far the edge gateway controller looks forward, i.e., how many time steps in the future are predicted; is the discrete time step at the current control time; is the reference value or expected trajectory of the state variable; is the state tracking weight matrix, is the control time domain, indicating how many control inputs the edge gateway controller calculates at a time, and generally . is the control input vector at the current time instant; usually the instructions issued by the edge gateway controller to the frequency converter, such as torque or power instructions; is the weight matrix of energy consumption control, is the change of control input, i.e. , is the weight matrix of smooth control.

[0059] It should be noted that in the above formula, is the state tracking term, which aims to penalize (minimize) the deviation between the predicted value and the expected value of the system state within the entire prediction time domain. The larger the diagonal elements of the weight matrix Q, the more closely the edge gateway controller will track the corresponding state variable (such as speed), which is directly related to drilling efficiency (maintaining optimal speed) and wellbore quality, and is the main driving term to achieve the requirements of drilling technology. is the energy consumption control term, which aims to penalize (minimize) the size of the control input itself. The larger the diagonal elements of the weight matrix R, the more the controller will tend to use smaller control instructions (such as smaller torque or power instructions) to achieve the control target, which helps to avoid unnecessary energy waste and reduce the operating cost of the system. is the smooth control term, which aims to penalize (minimize) the degree of change in control input between adjacent control time instants. The larger the diagonal elements of the weight matrix S, the smoother the control instruction sequence generated by the controller will be, directly suppressing the sharp fluctuations in power, which is a guarantee for achieving a smooth power curve to cope with load shocks. At the same time, smooth instructions can also reduce mechanical shocks to the frequency converter and transmission system, thereby prolonging the service life of the equipment.

[0060] The optimization objective function is subject to the following constraints:

[0061] System state space model: ;

[0062] State variable constraints: ;

[0063] Control input constraints: ;

[0064] Input change rate constraints: ;

[0065] wherein, is the discrete-time state space matrix of the drilling power system state, is the discrete-time state space matrix of the drilling power system control input, is the discrete-time state space matrix of the disturbance vector, and the above three state space matrices can be constructed according to system identification. is a measurable disturbance vector for the current time; that is, the sequence of formation load torque change trends in step 2; is the minimum value of the state, is the maximum value of the state; is the minimum value of the control input, is the maximum value of the control input; is the minimum value of the control input change, is the maximum value of the control input change.

[0066] It should be noted that the predictive control algorithm in the embodiment can balance the contradictions among performance, energy consumption and device smoothing under the premise of meeting all physical safety constraints. By receiving the sequence about future challenges from the LSTM, an optimal execution scheme with the highest comprehensive benefit in a future period of time is finally calculated. The prediction information is converted into specific, executable, smooth and safe device control instructions. For example, in the case of encountering a complex bottom formation interface mutation scene, the specific processing flow can be as follows: when the LWD foresight data predicts that the drill bit is about to enter the low-intensity soft mudstone formation from the high-intensity hard rock formation, the LSTM will predict that the load torque will drop abruptly. This prediction sequence is input into the predictive control algorithm as a disturbance In order to minimize the state tracking term in the objective function (prevent the motor speed from being far beyond the set value), the optimal solution of the predictive control algorithm must be to smoothly reduce the control input (i.e., power) in advance before the impact arrives, thereby generating a “pre-load reduction” ramp instruction, so that the drill bit executes a power “pre-load reduction” ramp instruction in advance before reaching the formation interface, to gently reduce the output torque and prevent the drill bit from flying away due to too fast unloading at the moment of entering the soft formation.

[0067] Step 4: The edge gateway controller receives the load torque change trend sequence and the power pre-regulation instruction sequence, takes the load torque change trend sequence as the cooperative control benchmark, and synchronously issues the power pre-regulation instruction sequence generated by the predictive control algorithm to the drilling fluid pump control system and the heave compensation system of the drilling platform. The drilling fluid pump control system and the heave compensation system execute the instruction sequence to realize the linkage optimization of the power system and the associated subsystems.

[0068] Specifically, the cooperative control instruction for the drilling fluid pump control system is generated differently based on the lithology corresponding to the load torque change trend sequence: when it is predicted that the soft mudstone formation prone to mud cake is about to be entered, the cooperative instruction is to increase the drilling fluid discharge in advance; when it is predicted that the fractured formation prone to leakage is about to be entered, the cooperative instruction is to appropriately reduce the drilling fluid discharge.

[0069] A formation lithology-drilling fluid strategy knowledge base can be built into the drilling fluid pump control system. When the LSTM predicts that a soft mudstone formation prone to mud ball is coming, the drilling fluid pump control system will automatically query the knowledge base and execute collaborative instructions. For example, 5-60 seconds in advance, gradually increase the drilling fluid displacement and / or adjust the drilling fluid performance to prevent the occurrence of mud ball. Conversely, if it is predicted that a fractured formation prone to leakage will be entered, the displacement will be appropriately reduced in advance to protect the oil layer.

[0070] The collaborative control instruction of the heave compensation system is generated based on the formation structure corresponding to the load torque change trend sequence.

[0071] For example, continuous short-time Fourier transform (STFT) or fast Fourier transform (FFT) can be performed on real-time torque data. When the analysis result shows that the signal energy continuously exceeds a preset threshold (for example, more than 3 standard deviations of normal background fluctuations) in the typical characteristic frequency band of 0.1 Hz to 2 Hz of stick-slip vibration, it is determined that there is a stick-slip vibration precursor.

[0072] Once the precursor is identified, the heave compensation system immediately triggers two parallel suppression actions: a high-frequency (for example, 5-10 Hz) and low-amplitude (for example, 1%-2% of the set speed) sinusoidal disturbance signal is superimposed on the original speed set value and sent to the frequency converter. The disturbance aims to actively destroy the stable resonance condition on which stick-slip vibration relies. At the same time, an instantaneous pressure increase instruction is sent to the drilling fluid pump control system, which increases the displacement by 5%-10% in a short time (for example, within 3-5 seconds). The increased displacement can improve the cleaning condition and hydraulic parameters at the bottom of the well, thereby changing the friction characteristics at the bottom of the well and assisting in suppressing the further development of torsional vibration.

[0073] The above specific embodiments further illustrate the purpose, technical solutions and beneficial effects of the present application. It should be understood that the above description is only a specific embodiment of the present application and is not intended to limit the protection scope of the present application. Any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for optimizing control of electric power for oil drilling in complex formations, characterized in that, The power optimization control method comprises: S100, collecting multi-source heterogeneous data in the oil drilling process, In the edge gateway controller, the collected multi-source heterogeneous data is fused to obtain a fused multi-source heterogeneous data stream; S200, processing the fused multi-source heterogeneous data stream through a pre-trained long short-term memory network to obtain a sequence of formation load torque change trends that the drill bit will encounter in a preset future time window; S300, a predictive control algorithm obtains an optimal power pre-adjustment instruction sequence according to the sequence of formation load torque change trends, with the optimization target of minimizing power consumption fluctuation, deviation of drilling tool rotating speed and torque impact, under the condition of meeting the motor safe operation constraint; S400, the edge gateway controller receives the sequence of load torque change trends and the power pre-adjustment instruction sequence, takes the sequence of load torque change trends as a cooperative control benchmark, and synchronously issues the power pre-adjustment instruction sequence to a drilling fluid pump control system and a heave compensation system of the drilling platform, so that the drilling fluid pump control system and the heave compensation system execute the instruction sequence to realize linkage optimization of the power system and the associated subsystem.

2. The method of claim 1, wherein, The multi-source heterogeneous data comprises: logging-while-drilling forward-looking data containing formation lithology parameters and formation drillability parameters; drilling real-time working condition data containing drilling pressure, torque and rotating speed; power system electrical parameter data containing output current, voltage and power of the main drive frequency converter.

3. The method of claim 1, wherein, The fusion processing comprises: aligning time stamps by unifying the multi-source heterogeneous data to a common time benchmark; cleaning data abnormal values by detecting abnormal values and noise interference in the data; eliminating dimension influence by converting data of different physical units and dimensions into data with the same scale through a Z-Score algorithm.

4. The method of claim 1, wherein, The pre-trained long short-term memory network has a pre-trained data set, which is historical data from a drilled section of the well and / or a historical database of adjacent drilled wells; The long short-term memory network further comprises online updating, calculating a prediction error by comparing an actually measured torque value with a predicted value of the model, and taking the error as a correction signal to online fine-tune network weights of the long short-term memory network through a small learning rate back propagation algorithm.

5. The method of claim 1, wherein, The predictive control algorithm comprises an optimization objective function for balancing contradictions among performance, energy consumption and equipment smoothing, and a constraint condition for constraining the optimization objective function.

6. The method of claim 5, wherein, The optimization objective function is as shown in the following formula: ， wherein, is the objective function, is the state vector at the current time instant, is the motor output torque, is the matrix transpose symbol; is the entire control sequence within the future control horizon; is a single time step; is the prediction horizon, is the discrete time step at the current control instant; is the reference value or desired trajectory of the state variable; is the weight matrix for state tracking, is the control horizon, is the control input vector at the current time instant; is the weight matrix for energy consumption control, is the change in control input, is the weight matrix for smooth control.

7. The method of claim 5, wherein, The constraint condition comprises: System state space model: ; State variable constraints: ; Control input constraints: ; Make input rate of change constraint: ; wherein, is a discrete-time state-space matrix for the state of the drilling power system, is a discrete-time state-space matrix for the control input of the drilling power system, is a discrete-time state-space matrix for the disturbance vector, is the measurable disturbance vector at the current time instant; is the minimum value of the state, is the maximum value of the state; is the minimum value of the control input, is the maximum value of the control input; is the minimum value of the control input variation, is the maximum value of the control input variation.

8. The method of claim 1, wherein, The cooperative control instruction of the drilling fluid pump control system is generated differently based on formation lithology corresponding to the sequence of load torque change trends: When it is predicted that a soft mudstone formation prone to mud cake is about to be entered, the cooperative instruction is to increase drilling fluid discharge in advance; When it is predicted that a fractured formation prone to leakage is about to be entered, the cooperative instruction is to appropriately reduce drilling fluid discharge; By embedding a formation lithology-drilling fluid strategy knowledge base in the drilling fluid pump control system, when the sequence of load torque change trends is that a soft mudstone formation prone to mud cake will be entered in the future, the drilling fluid pump control system automatically queries the knowledge base and executes the cooperative instruction.

9. The method of claim 1, wherein, The cooperative regulation instruction of the heave compensation system is differentially generated based on a formation structure corresponding to the load torque change trend sequence, Continuous short-time Fourier transform or fast Fourier transform is performed on real-time torque data, When the analysis result shows that signal energy continuously exceeds a preset threshold in a typical characteristic frequency band of stick-slip vibration, i.e. 0.1 Hz-2 Hz, it is determined that stick-slip vibration precursor exists; Once stick-slip vibration precursor is identified, the heave compensation system immediately triggers two parallel suppression actions: a high-frequency, low-amplitude sinusoidal disturbance signal is superimposed on the original speed setting value and sent to the frequency converter; Meanwhile, an instantaneous pressure boost instruction is sent to the drilling fluid pump control system, so that the displacement is increased by 5%-10% within 3-5 seconds.

10. A complex formation oil drilling electric power generation oil power optimization control system, characterized in that, The power optimization control system comprises: a processor; a memory storing a computer program, which, when executed by the processor, implements the complex formation oil drilling electric oil power optimization control method according to any one of claims 1-9.

Images