An Adaptive Optimization Control Method and System for Intelligent Drilling Parameters in Kilometer-Scale Vertical Shafts
By establishing an initial population in kilometer-level vertical shaft drilling and fusing parameter similarity and historical global optimal information, and adjusting the crossover probability in combination with the drilling rate evaluation function value, the problem of limited optimization accuracy and convergence effect in the iterative optimization of parameters in kilometer-level vertical shaft drilling was solved, realizing adaptive optimization of parameter groups and safe and efficient drilling control.
Patent Information
- Application Number
- CN202611618380.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-10-20
- Publication Date
- 2026-11-17
AI Technical Summary
In existing technologies, the cross probability adjustment method is fixed during the iterative optimization of parameters for kilometer-level vertical shaft drilling. It is difficult to adaptively adjust the parameters according to individual performance differences, which limits the optimization accuracy and convergence effect. In particular, the drilling stability and construction efficiency are insufficient under variable geological conditions.
By acquiring drilling parameter types and geological data measured while drilling, an initial population is established. Based on parameter similarity and historical global optimal information, iterative optimization is performed to generate a composite steering parameter set. The crossover probability is determined by combining the drilling rate evaluation function value, thereby achieving adaptive optimization of the parameter set.
It improves the adaptability and construction efficiency of parameter control for kilometer-level vertical shaft drilling, ensures that the parameter set is optimized within the safety boundary, and enhances the stability and efficiency of drilling.
Smart Images

Figure CN123113406A_ABST
Abstract
Description
Technical Field
[0001] This application belongs to the field of kilometer-level vertical shaft drilling, and in particular relates to an adaptive optimization control method and system for intelligent drilling parameters of kilometer-level vertical shafts. Background Technology
[0002] During the drilling of deep vertical shafts at depths of several kilometers, the formation mechanical characteristics, such as ground stress, rock hardness, drillability, and abrasiveness, continuously change with increasing depth. If drilling parameters cannot be adapted to the current geological conditions in a timely manner, drilling stability and construction efficiency can be easily affected. To maintain optimal drilling conditions, it is usually necessary to adjust key drilling parameters such as drill pressure, rotation speed, and mud pump volume in real time, based on geological data obtained from measurements while drilling. Existing technologies have incorporated swarm intelligence optimization methods into the drilling parameter optimization process. By constructing a drilling performance evaluation model, iterative searches are conducted within preset safety constraints to find the optimal parameter combination for drilling efficiency. However, in existing parameter search processes, the crossover probability often adopts a fixed adjustment strategy that decreases with the number of iterations. This makes it difficult to adjust the search intensity in real time based on the performance differences between the current individual parameters and the historical globally optimal parameter set. This results in insufficient coordination between local search and global guidance, affecting the later optimization accuracy and convergence effect. Therefore, how to integrate local similar parameter characteristics and historical global optimal state information during the iterative optimization of drilling parameters under varying geological conditions, improve the guidance of individual searches, and adaptively adjust the cross-optimization behavior according to the performance differences between individual parameters to obtain a drilling parameter combination more suitable for the current working conditions is a technical problem that needs to be solved in the field of kilometer-level vertical shaft drilling parameter control. Summary of the Invention
[0003] To address the problem that existing technologies use fixed cross-probability adjustment methods, which are difficult to adaptively adjust based on individual parameter performance differences, thus limiting optimization accuracy and convergence performance, this invention proposes an adaptive optimization control method and system for intelligent drilling parameters in kilometer-level vertical shafts.
[0004] According to one aspect of the present invention, an adaptive optimization control method for intelligent drilling parameters of kilometer-scale vertical shafts is proposed, comprising the following steps: Obtain drilling parameter types and geological data measured while drilling, and establish an initial population containing multiple drilling parameter groups based on the drilling parameter types; The initial population is used as the current population. An iterative optimization operation is performed on each target parameter group in the current population. The parameter similarity between other parameter groups and the target parameter group is calculated based on the parameter values. The neighboring reference parameter group is determined. The guidance weight is determined based on the parameter similarity between the target parameter group and the population's historical global optimal parameter group. The neighboring reference parameter group and the population's historical global optimal parameter group are weighted and fused according to the guidance weight to generate a composite guidance parameter group. The crossover probability is determined based on the difference between the drilling rate evaluation function value of the target parameter set and the population historical global optimal parameter set. The target parameter set and the composite steering parameter set are cross-referenced based on the crossover probability to generate an experimental parameter set. The drilling rate evaluation function value of the experimental parameter set determines whether to replace the target parameter set with the experimental parameter set. The iterative optimization operation is repeated until the preset convergence condition is met and the optimal drilling parameters are output.
[0005] Optionally, the step of acquiring drilling parameter types and measurement-while-drilling geological data, and establishing an initial population containing multiple drilling parameter groups based on the drilling parameter types, includes: Determine the safe operating control boundaries for drilling pressure, rotation speed, and mud pump volume; Collect geological data measured while drilling during the drilling process to obtain the lithological response sequence matrix corresponding to the current well depth; Within the safe operation control boundary, multiple drilling parameter combinations are randomly generated using a uniform distribution generation algorithm. Each drilling parameter combination is treated as an individual, and multiple individuals constitute the initial population.
[0006] Optionally, the step of using the initial population as the current population, performing iterative optimization operations on each target parameter set in the current population, calculating the parameter similarity between other parameter sets and the target parameter set based on parameter values, and determining the nearest reference parameter set includes: Extract the target parameter value sequence of the target parameter group and the reference parameter value sequence of other parameter groups in the population; The target parameter value sequence and the reference parameter value sequence are dimensionless. Calculate the sum of squared differences between corresponding elements in the dimensionless sequence of target parameter values and the sequence of reference parameter values; The parameter similarity is calculated based on the reciprocal of the sum of squared differences. The parameter group with the highest parameter similarity is determined as the nearest reference parameter group.
[0007] Optionally, the step of determining the guidance weight based on the parameter similarity between the target parameter set and the population's historical global optimal parameter set, and then weighting and fusing the neighboring reference parameter set and the population's historical global optimal parameter set according to the guidance weight to generate a composite guidance parameter set, includes: Obtain the first parameter vector corresponding to the target parameter set and the second parameter vector corresponding to the population's historical global optimal parameter set, and perform dimensionless processing on them respectively; Calculate the cosine similarity between the dimensionless first parameter vector and the second parameter vector, and use the normalized cosine similarity as the guiding weight. The first fusion component is obtained by multiplying the guiding weight by the parameter value of the population's historical global optimal parameter set; The difference between 1 and the guiding weight is multiplied by the parameter value of the neighboring reference parameter group to obtain the second fusion component; The first fusion component and the second fusion component are added together to obtain the composite guidance parameter set.
[0008] Optionally, the step of determining the crossover probability based on the difference between the drilling rate evaluation function value of the target parameter set and the population's historical global optimal parameter set, and performing parameter crossover on the target parameter set and the composite steering parameter set based on the crossover probability to generate an experimental parameter set, includes: Obtain the first drilling rate evaluation function value of the target parameter set and the second drilling rate evaluation function value of the population's historical global optimal parameter set; Calculate the absolute value of the difference between the second drilling speed evaluation function value and the first drilling speed evaluation function value, and map the absolute value of the difference to a preset probability interval to obtain the cross probability; For each parameter dimension, a random number between 0 and 1 is generated. If the random number for the corresponding dimension is less than the crossover probability, the parameter value of the target parameter group is replaced with the parameter value of the composite guiding parameter group in that dimension. If the random number in the corresponding dimension is greater than or equal to the crossover probability, then the parameter values of the target parameter set in that dimension are retained, thereby generating the experimental parameter set.
[0009] Optionally, determining whether to replace the target parameter set with the test parameter set based on the drilling rate evaluation function value of the test parameter set includes: The test parameter set and the geological data measured while drilling are substituted into the pre-constructed drilling speed prediction model to calculate and output the corresponding expected drilling speed, which is used as the drilling speed evaluation function value of the test parameter set. Compare the drilling rate evaluation function values of the test parameter set with those of the target parameter set; When the drilling rate evaluation function value of the experimental parameter group is greater than the drilling rate evaluation function value of the target parameter group, the experimental parameter group is determined to be superior to the target parameter group, and the individual at the corresponding position in the next generation population is updated to the experimental parameter group and the target parameter group is discarded.
[0010] According to another aspect of the present invention, a kilometer-level intelligent drilling parameter adaptive optimization control system for vertical shafts is proposed, comprising the following modules: The acquisition module is used to acquire drilling parameter types and geological data measured while drilling, and to establish an initial population containing multiple drilling parameter groups based on the drilling parameter types. The calculation module is used to take the initial population as the current population, perform iterative optimization operation on each target parameter group in the current population, calculate the parameter similarity between other parameter groups and the target parameter group based on the parameter values, determine the neighboring reference parameter group, determine the guidance weight based on the parameter similarity between the target parameter group and the population's historical global optimal parameter group, and perform weighted fusion of the neighboring reference parameter group and the population's historical global optimal parameter group based on the guidance weight to generate a composite guidance parameter group; The generation module is used to determine the crossover probability based on the difference between the drilling rate evaluation function value of the target parameter set and the population historical global optimal parameter set, perform parameter crossover on the target parameter set and the composite steering parameter set according to the crossover probability, generate a test parameter set, determine whether to replace the target parameter set with the test parameter set according to the drilling rate evaluation function value of the test parameter set, and repeat the iterative optimization operation until the preset convergence condition is met and the optimal drilling parameters are output.
[0011] Preferably, the step of acquiring drilling parameter types and geological data measured while drilling, and establishing an initial population containing multiple drilling parameter groups based on the drilling parameter types, includes: Determine the safe operating control boundaries for drilling pressure, rotation speed, and mud pump volume; Collect geological data measured while drilling during the drilling process to obtain the lithological response sequence matrix corresponding to the current well depth; Within the safe operation control boundary, multiple drilling parameter combinations are randomly generated using a uniform distribution generation algorithm. Each drilling parameter combination is treated as an individual, and multiple individuals constitute the initial population.
[0012] Preferably, the step of taking the initial population as the current population, performing an iterative optimization operation on each target parameter set in the current population, calculating the parameter similarity between other parameter sets and the target parameter set based on the parameter values, and determining the nearest reference parameter set includes: Extract the target parameter value sequence of the target parameter group and the reference parameter value sequence of other parameter groups in the population; The target parameter value sequence and the reference parameter value sequence are dimensionless. Calculate the sum of squared differences between corresponding elements in the dimensionless sequence of target parameter values and the sequence of reference parameter values; The parameter similarity is calculated based on the reciprocal of the sum of squared differences. The parameter group with the highest parameter similarity is determined as the nearest reference parameter group.
[0013] Preferably, the step of determining the guidance weight based on the parameter similarity between the target parameter set and the population's historical global optimal parameter set, and then weighting and fusing the neighboring reference parameter set and the population's historical global optimal parameter set according to the guidance weight to generate a composite guidance parameter set, includes: Obtain the first parameter vector corresponding to the target parameter set and the second parameter vector corresponding to the population's historical global optimal parameter set, and perform dimensionless processing on them respectively; Calculate the cosine similarity between the dimensionless first parameter vector and the second parameter vector, and use the normalized cosine similarity as the guiding weight. The first fusion component is obtained by multiplying the guiding weight by the parameter value of the population's historical global optimal parameter set; The difference between 1 and the guiding weight is multiplied by the parameter value of the neighboring reference parameter group to obtain the second fusion component; The first fusion component and the second fusion component are added together to obtain the composite guidance parameter set.
[0014] Preferably, the step of determining the crossover probability based on the difference between the drilling rate evaluation function value of the target parameter set and the population's historical global optimal parameter set, and performing parameter crossover on the target parameter set and the composite steering parameter set based on the crossover probability to generate an experimental parameter set includes: Obtain the first drilling rate evaluation function value of the target parameter set and the second drilling rate evaluation function value of the population's historical global optimal parameter set; Calculate the absolute value of the difference between the second drilling speed evaluation function value and the first drilling speed evaluation function value, and map the absolute value of the difference to a preset probability interval to obtain the cross probability; For each parameter dimension, a random number between 0 and 1 is generated. If the random number for the corresponding dimension is less than the crossover probability, the parameter value of the target parameter group is replaced with the parameter value of the composite guiding parameter group in that dimension. If the random number in the corresponding dimension is greater than or equal to the crossover probability, then the parameter values of the target parameter set in that dimension are retained, thereby generating the experimental parameter set.
[0015] Preferably, the step of determining whether to replace the target parameter set with the test parameter set based on the drilling rate evaluation function value of the test parameter set includes: The test parameter set and the geological data measured while drilling are substituted into the pre-constructed drilling speed prediction model to calculate and output the corresponding expected drilling speed, which is used as the drilling speed evaluation function value of the test parameter set. Compare the drilling rate evaluation function values of the test parameter set with those of the target parameter set; When the drilling rate evaluation function value of the experimental parameter group is greater than the drilling rate evaluation function value of the target parameter group, the experimental parameter group is determined to be superior to the target parameter group, and the individual at the corresponding position in the next generation population is updated to the experimental parameter group and the target parameter group is discarded.
[0016] The present invention has the following beneficial effects: Acquire geological data from drilling measurements and establish an initial population so that drilling parameter optimization can be carried out in conjunction with current geological conditions; During the iteration process, the neighboring reference parameter group is determined based on parameter similarity, and the guidance weight is determined according to the similarity relationship between the target parameter group and the historical global optimal parameter group. The neighboring reference parameter group and the historical global optimal parameter group are weighted and fused to generate a composite guidance parameter group, thereby taking into account both local similarity reference and global guidance. The crossover probability is determined based on the difference in drilling speed evaluation function values, thereby improving the targeting of parameter crossover adjustments; Since both the target parameter set and the composite steering parameter set are located within the safe operation control boundary, and the parameter crossover adopts the corresponding dimension parameter value selection method, the cross-generated test parameter set can be kept within the safe boundary range. This is beneficial for obtaining better drilling parameters under the variable working conditions of kilometer-level vertical shafts, thereby improving the adaptability of drilling control and construction efficiency. Attached Figure Description
[0017] Figure 1 A flowchart of the first embodiment; Figure 2 This is a schematic diagram of the fusion distribution of composite guidance parameters. Detailed Implementation
[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.
[0019] In the first embodiment, the present invention proposes an adaptive optimization control method for intelligent drilling parameters in kilometer-scale vertical shafts, such as... Figure 1 As shown, it includes the following steps: S1, acquire drilling parameter types and geological data measured while drilling, and establish an initial population containing multiple drilling parameter groups based on the drilling parameter types.
[0020] Geological data is collected in real time by the measurement-while-drilling system configured on a kilometer-class vertical shaft drilling rig. The geological data includes the uniaxial compressive strength of the rock, the rock drillability grade, the formation abrasiveness index, as well as natural gamma, sonic transit time, and formation resistivity. The drilling parameter types to be optimized are determined, including drilling pressure, rotational speed, and mud pump rate.
[0021] Based on the performance limitations of the drilling rig and the current geological data, upper and lower thresholds for drilling pressure, rotation speed, and mud pump rate are determined. The initial population size, i.e., the total number of parameter sets, is then set. Using a sampling algorithm with uniform coverage characteristics, such as the Latin hypercube sampling algorithm, multiple numerical sequences are randomly generated within the determined upper and lower threshold ranges. Each numerical sequence serves as a drilling parameter set, and all drilling parameter sets together constitute the initial population. Each drilling parameter set contains a specific drilling pressure value, rotation speed value, and mud pump rate value.
[0022] In an optional embodiment, the step of acquiring drilling parameter types and measurement-while-drilling geological data, and establishing an initial population containing multiple drilling parameter groups based on the drilling parameter types, includes: Determine the safe operating control boundaries for drilling pressure, rotation speed, and mud pump volume; Collect geological data measured while drilling during the drilling process to obtain the lithological response sequence matrix corresponding to the current well depth; Within the safe operation control boundary, multiple drilling parameter combinations are randomly generated using a uniform distribution generation algorithm. Each drilling parameter combination is treated as an individual, and multiple individuals constitute the initial population.
[0023] According to the drilling rig equipment manual and downhole safety specifications, the safe operation control boundaries for drilling parameters are set as follows: drilling pressure boundary is [100, 300] kN, rotation speed boundary is [40, 120] r / min, and mud pump flow rate boundary is [20, 50] L / s.
[0024] Collect drilling geological data from N depth nodes within a preset depth window near the current well depth of 1500 meters. For example, the natural gamma value corresponding to one depth node is 85 API, the rock sonic transit time is 75 μs / ft, and the formation resistivity is 25 Ω·m. Construct a lithological response sequence matrix with dimension N×3 according to the depth nodes.
[0025] Within the safe operation control boundary, the initial population size is set to 50. Multiple drilling parameter combinations are randomly generated using a uniform distribution generation algorithm, with the formula being parameter value = lower limit + rand(0,1) × (upper limit - lower limit). When the three randomly generated coefficients are 0.45, 0.60, and 0.30, the generated drilling parameter combinations are: drilling pressure 190kN (100 + 0.45 × 200), rotation speed 88r / min (40 + 0.60 × 80), and mud pump flow rate 29L / s (20 + 0.30 × 30).
[0026] By repeating the above operations, each drilling parameter is combined as an individual, and the initial population consists of 50 individuals.
[0027] S2, taking the initial population as the current population, performing an iterative optimization operation on each target parameter group in the current population, calculating the parameter similarity between other parameter groups and the target parameter group based on the parameter values, determining the neighboring reference parameter group, determining the guidance weight based on the parameter similarity between the target parameter group and the population's historical global optimal parameter group, and generating a composite guidance parameter group by weighted fusion of the neighboring reference parameter group and the population's historical global optimal parameter group based on the guidance weight.
[0028] During the initial execution, the parameter values of each drilling parameter group in the initial population are normalized to obtain the normalized values corresponding to each parameter dimension. The absolute difference between each normalized value of each drilling parameter group and the preset mid-center benchmark value is calculated, wherein the preset mid-center benchmark value is preferably 0.5. The mid-center proximity deviation value is obtained by summing the absolute differences. The drilling parameter group with the smallest mid-center proximity deviation value is selected as the initial guiding parameter group for the first iteration and is temporarily used as the initial value of the population's historical global optimal parameter group. In subsequent iterations, the population's historical global optimal parameter group is updated based on the drilling rate evaluation results.
[0029] In each iteration, the current individual in the current population is taken as the target parameter set. The normalized sum of squared differences between the other parameter sets in the current population and the target parameter set in each dimension is calculated, and the reciprocal of the normalized sum of squared differences is taken as the parameter similarity. The parameter set with the highest parameter similarity is selected as the neighboring reference parameter set.
[0030] Calculate the global guidance similarity between the target parameter set and the population's historical global optimal parameter set, and determine the guidance weight based on the global guidance similarity. A higher global guidance similarity results in a higher guidance weight. Weight the population's historical global optimal parameter set according to the guidance weight, and weight the neighboring reference parameter set by subtracting the guidance weight from 1. Add the two weighted results according to their corresponding parameter dimensions to obtain a composite guidance parameter set including new drilling pressure, new rotation speed, and new mud pump volume.
[0031] In an optional embodiment, the step of using the initial population as the current population, performing an iterative optimization operation on each target parameter set in the current population, calculating the parameter similarity between other parameter sets and the target parameter set based on parameter values, and determining the nearest reference parameter set includes: Extract the target parameter value sequence of the target parameter group and the reference parameter value sequence of other parameter groups in the population; The target parameter value sequence and the reference parameter value sequence are dimensionless. Calculate the sum of squared differences between corresponding elements in the dimensionless sequence of target parameter values and the sequence of reference parameter values; The parameter similarity is calculated based on the reciprocal of the sum of squared differences. The parameter group with the highest parameter similarity is determined as the nearest reference parameter group.
[0032] Extract the target parameter value sequence of the target parameter set The parameters are [190kN, 88r / min, 29L / s], and a sequence of reference parameter values for another parameter group in the population. The target parameter value sequence is [200kN, 80r / min, 32L / s]. The maximum-minimum normalization method is used to process the target parameter value sequence and the reference parameter value sequence. The processed target parameter value sequence is... The reference parameter value sequence is [0.45, 0.60, 0.30]. [0.50, 0.50, 0.40].
[0033] The sum of squared differences between corresponding elements in the dimensionless target parameter value sequence and the reference parameter value sequence is calculated as D = 0.0225. Introducing a minimum constant ε of 1e-5, the parameter similarity S is calculated based on the reciprocal of the sum of squared differences: S = 1 / (D+ε) = 1 / 0.02251 ≈ 44.42. The remaining parameter groups in the current population are traversed, and their parameter similarities are calculated for each group. The parameter group with the highest similarity is determined as the nearest reference parameter group.
[0034] In an optional embodiment, the step of determining the guidance weight based on the parameter similarity between the target parameter set and the population's historical global optimal parameter set, and then weighting and fusing the neighboring reference parameter set and the population's historical global optimal parameter set according to the guidance weight to generate a composite guidance parameter set, includes: Obtain the first parameter vector corresponding to the target parameter set and the second parameter vector corresponding to the population's historical global optimal parameter set, and perform dimensionless processing on them respectively; Calculate the cosine similarity between the dimensionless first parameter vector and the second parameter vector, and use the normalized cosine similarity as the guiding weight. The first fusion component is obtained by multiplying the guiding weight by the parameter value of the population's historical global optimal parameter set; The difference between 1 and the guiding weight is multiplied by the parameter value of the neighboring reference parameter group to obtain the second fusion component; The first fusion component and the second fusion component are added together to obtain the composite guidance parameter set.
[0035] Obtain the first parameter vector corresponding to the target parameter set and perform normalization. =[0.45,0.60,0.30], obtain the second parameter vector corresponding to the global optimal parameter set in the history of the population, and perform dimensionless processing to obtain the second dimensionless parameter vector. =[0.80,0.70,0.50]. The dot product of the dimensionless first parameter vector and the second parameter vector is calculated as 0.45×0.80+0.60×0.70+0.30×0.50=0.93. The mold length is approximately 0.8078 mm. The modulus is approximately 1.1747, and the calculated cosine similarity is 0.93 / (0.8078×1.1747)≈0.98. The normalized cosine similarity of 0.98 is used as the guiding weight.
[0036] The parameters obtained from the historical global optimal parameter set of the population are: drilling pressure 260 kN, rotation speed 104 r / min, and mud pump flow rate 35 L / s. The parameters obtained from the neighboring reference parameter set are: drilling pressure 220 kN, rotation speed 95 r / min, and mud pump flow rate 30 L / s. Multiplying the steering weight by the parameters of the historical global optimal parameter set yields the first fusion component [254.8, 101.92, 34.3]. Multiplying the difference between 1 and the steering weight by the parameters of the neighboring reference parameter set yields the second fusion component [4.4, 1.9, 0.6]. Adding the first and second fusion components yields the composite steering parameter set [259.2 kN, 103.82 r / min, 34.9 L / s]. The fusion distribution of the composite steering parameters is as follows: Figure 2 As shown.
[0037] S3. Determine the crossover probability based on the difference between the drilling rate evaluation function value of the target parameter group and the population historical global optimal parameter group. Perform parameter crossover on the target parameter group and the composite steering parameter group according to the crossover probability to generate a test parameter group. Determine whether to replace the target parameter group with the test parameter group based on the drilling rate evaluation function value of the test parameter group. Repeat the iterative optimization operation until the preset convergence condition is met and the optimal drilling parameters are output.
[0038] The input characteristics and parameter constraints of the drilling rate evaluation function are determined in advance based on geological data obtained from drilling measurements and the rock breaking mechanism. A drilling rate prediction model is trained based on historical drilling samples, and the predicted drilling rate output by the model is used as the drilling rate evaluation function value. The target parameter set and the population's historical globally optimal parameter set are substituted into the drilling rate evaluation function to calculate the corresponding drilling rate evaluation function values. The absolute difference between the two drilling rate evaluation function values is calculated, and this absolute difference is mapped to a preset probability interval according to a preset upper limit to obtain the cross-probability.
[0039] For each dimension of the target parameter set, i.e., each drilling parameter, a uniformly distributed random number between 0 and 1 is generated. If the random number is less than the crossover probability, the value of the experimental parameter set in that dimension is set to the value of the composite steering parameter set in that dimension. If the random number is greater than or equal to the crossover probability, the value of the experimental parameter set in that dimension is set to the value of the target parameter set in that dimension, thus completing parameter crossover and generating the experimental parameter set. Since both the target parameter set and the composite steering parameter set are located within the safe operation control boundary, and the parameter crossover is a parameter value selection operation in the corresponding dimension, the generated experimental parameter set is still located within the safe operation control boundary.
[0040] The experimental parameter set is substituted into the drilling rate evaluation function to calculate the corresponding experimental drilling rate evaluation function value, which is then compared with the original drilling rate evaluation function value of the target parameter set. If the experimental drilling rate evaluation function value is greater than the original drilling rate evaluation function value, the target parameter set is replaced with the experimental parameter set in the next generation of the population; otherwise, the original target parameter set is retained in the next generation of the population.
[0041] After completing the parameter replacement judgment, the parameter set that enters the next generation of the population is used as the current retained parameter set; if the drilling rate evaluation function value of the current retained parameter set is greater than the drilling rate evaluation function value of the current population's historical global optimal parameter set, then the current retained parameter set is updated to the new population's historical global optimal parameter set; otherwise, the current population's historical global optimal parameter set remains unchanged.
[0042] After completing the update judgment of the current target parameter set, the same iterative optimization operation is performed on other target parameter sets in the current population. After one round of updates, a next-generation population is formed, and the next-generation population is used as the new current population. The current iteration number is updated, and it is determined whether the termination condition is met. For example, the preset maximum number of iterations is set to 100, the preset number of consecutive convergences is 5, and the preset accuracy threshold is 0.01 m / h. If the current iteration number reaches 100, or if the increment of the drilling rate evaluation function value of the population's historical global optimal parameter set relative to the previous iteration is less than 0.01 m / h in 5 consecutive iterations, the iterative optimization operation is stopped. The population's historical global optimal parameter set at this time is output as the optimal drilling parameters under the current optimization conditions, and the drilling pressure, rotation speed, and mud pump volume are sent to the bottom control system of the kilometer-level vertical shaft drilling rig to execute the actual drilling operation.
[0043] In an optional embodiment, the step of determining the crossover probability based on the difference between the drilling rate evaluation function value of the target parameter set and the population's historical global optimal parameter set, and performing parameter crossover on the target parameter set and the composite steering parameter set based on the crossover probability to generate an experimental parameter set, includes: Obtain the first drilling rate evaluation function value of the target parameter set and the second drilling rate evaluation function value of the population's historical global optimal parameter set; Calculate the absolute value of the difference between the second drilling speed evaluation function value and the first drilling speed evaluation function value, and map the absolute value of the difference to a preset probability interval to obtain the cross probability; For each parameter dimension, a random number between 0 and 1 is generated. If the random number for the corresponding dimension is less than the crossover probability, the parameter value of the target parameter group is replaced with the parameter value of the composite guiding parameter group in that dimension. If the random number in the corresponding dimension is greater than or equal to the crossover probability, then the parameter values of the target parameter set in that dimension are retained, thereby generating the experimental parameter set.
[0044] The first drilling rate evaluation function value of the target parameter set is 4.2 m / h, and the second drilling rate evaluation function value of the population's historical global optimal parameter set is 6.8 m / h. The absolute value of the difference between the second and first drilling rate evaluation function values is calculated as |6.8 - 4.2| = 2.6 m / h. The maximum difference is set to 5.0 m / h, and the preset probability interval is [0.1, 0.9]. A linear mapping is used to map the absolute value of the difference to the preset probability interval, resulting in a crossover probability CR = 0.1 + (2.6 / 5.0) × (0.9 - 0.1) = 0.516.
[0045] For the dimensions of drilling pressure, rotation speed, and mud pump rate, a random number between 0 and 1 was generated, with values of 0.35, 0.72, and 0.48 respectively. For the drilling pressure dimension, since the random number 0.35 is less than the crossover probability of 0.516, the drilling pressure parameter value of 259.2 kN from the composite steering parameter group was used to replace the parameter value of the target parameter group in the corresponding dimension. For the rotation speed dimension, since the random number 0.72 is greater than or equal to the crossover probability of 0.516, the rotation speed parameter value of 88 r / min from the target parameter group in the corresponding dimension was retained. For the mud pump rate dimension, since the random number 0.48 is less than the crossover probability of 0.516, the mud pump rate parameter value of 34.9 L / s from the composite steering parameter group was used to replace the parameter value of the target parameter group in the corresponding dimension. The final generated experimental parameter group is [259.2 kN, 88 r / min, 34.9 L / s].
[0046] In an optional embodiment, determining whether to replace the target parameter set with the test parameter set based on the drilling rate evaluation function value of the test parameter set includes: The test parameter set and the geological data measured while drilling are substituted into the pre-constructed drilling speed prediction model to calculate and output the corresponding expected drilling speed, which is used as the drilling speed evaluation function value of the test parameter set. Compare the drilling rate evaluation function values of the test parameter set with those of the target parameter set; When the drilling rate evaluation function value of the experimental parameter group is greater than the drilling rate evaluation function value of the target parameter group, the experimental parameter group is determined to be superior to the target parameter group, and the individual at the corresponding position in the next generation population is updated to the experimental parameter group and the target parameter group is discarded.
[0047] The geological feature vector corresponding to the current well depth is extracted from the lithological response sequence matrix. The experimental parameter set, drilling pressure 259.2kN, rotation speed 88r / min, mud pump flow rate 34.9L / s, and geological data measured while drilling, natural gamma 85API, sonic transit time 75μs / ft, and formation resistivity 25Ω·m are spliced into an input vector and substituted into the pre-constructed BP neural network drilling rate prediction model.
[0048] The BP neural network drilling rate prediction model is pre-trained using historical drilling samples. These historical samples include drill pressure, rotational speed, mud pump volume, and corresponding measurement-while-drilling (MWD) geological data recorded during historical drilling operations. The MWD geological data includes natural gamma ray, sonic transit time, and formation resistivity. The sample labels are the measured mechanical drilling rates under the corresponding operating conditions. Before training, the historical drilling samples are screened and feature-constructed based on rock breaking mechanisms. Samples with drill pressure, rotational speed, or mud pump volume exceeding safe operating control boundaries are removed. Drill pressure, rotational speed, mud pump volume, and MWD geological data characterizing lithological changes are used together as model input features, ensuring that the model input reflects the correspondence between drilling parameters, formation conditions, and mechanical drilling rate.
[0049] The BP neural network drilling rate prediction model comprises an input layer, a hidden layer, and an output layer. The input layer receives an input vector formed by concatenating drilling pressure, rotational speed, mud pump volume, and geological data obtained during drilling. The hidden layer performs nonlinear feature mapping on the input vector to extract the correlation features between drilling parameters and geological conditions. The output layer outputs the corresponding predicted mechanical drilling rate, which is used as the drilling rate evaluation function value. During training, each set of historical drilling parameters is concatenated with the corresponding geological data obtained during drilling to form a training input vector. The measured mechanical drilling rate is used as the training output target. The network weights and biases are adjusted using an error backpropagation algorithm, gradually bringing the model's output predicted mechanical drilling rate closer to the measured mechanical drilling rate.
[0050] The model's calculated expected drilling rate is 5.4 m / h, which is used as the drilling rate evaluation function value for the experimental parameter group. Comparing the drilling rate evaluation function value of the experimental parameter group with the target parameter group's value of 4.2 m / h, if the experimental parameter group's value of 5.4 m / h is greater than the target parameter group's value of 4.2 m / h, the experimental parameter group is deemed superior to the target parameter group. The individual at the corresponding position in the next generation population is then updated to the experimental parameter group, and the target parameter group is discarded.
[0051] Since the drilling rate evaluation function value of the experimental parameter group is 5.4 m / h, which is less than the drilling rate evaluation function value of the current population's historical global optimal parameter group (6.8 m / h), the population's historical global optimal parameter group will not be updated. If the drilling rate evaluation function value of a certain experimental parameter group is greater than the drilling rate evaluation function value of the current population's historical global optimal parameter group, then the experimental parameter group will be updated synchronously to the new population's historical global optimal parameter group.
[0052] In the second embodiment, the present invention also proposes a kilometer-level intelligent drilling parameter adaptive optimization control system for vertical shafts, comprising the following modules: The acquisition module is used to acquire drilling parameter types and geological data measured while drilling, and to establish an initial population containing multiple drilling parameter groups based on the drilling parameter types. The calculation module is used to take the initial population as the current population, perform iterative optimization operation on each target parameter group in the current population, calculate the parameter similarity between other parameter groups and the target parameter group based on the parameter values, determine the neighboring reference parameter group, determine the guidance weight based on the parameter similarity between the target parameter group and the population's historical global optimal parameter group, and perform weighted fusion of the neighboring reference parameter group and the population's historical global optimal parameter group based on the guidance weight to generate a composite guidance parameter group; The generation module is used to determine the crossover probability based on the difference between the drilling rate evaluation function value of the target parameter set and the population historical global optimal parameter set, perform parameter crossover on the target parameter set and the composite steering parameter set according to the crossover probability, generate a test parameter set, determine whether to replace the target parameter set with the test parameter set according to the drilling rate evaluation function value of the test parameter set, and repeat the iterative optimization operation until the preset convergence condition is met and the optimal drilling parameters are output.
[0053] The above description represents the preferred embodiments of the present invention. It should be noted that, for those skilled in the art, various improvements and modifications can be made without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.
Claims
1. A method for adaptive optimization control of intelligent drilling parameters for kilometer-level vertical shafts, characterized in that, Includes the following steps: Obtain drilling parameter types and geological data measured while drilling, and establish an initial population containing multiple drilling parameter groups based on the drilling parameter types; The initial population is used as the current population. An iterative optimization operation is performed on each target parameter group in the current population. The parameter similarity between other parameter groups and the target parameter group is calculated based on the parameter values. The neighboring reference parameter group is determined. The guidance weight is determined based on the parameter similarity between the target parameter group and the population's historical global optimal parameter group. The neighboring reference parameter group and the population's historical global optimal parameter group are weighted and fused according to the guidance weight to generate a composite guidance parameter group. The crossover probability is determined based on the difference between the drilling rate evaluation function value of the target parameter set and the population historical global optimal parameter set. The target parameter set and the composite steering parameter set are cross-referenced based on the crossover probability to generate an experimental parameter set. The drilling rate evaluation function value of the experimental parameter set determines whether to replace the target parameter set with the experimental parameter set. The iterative optimization operation is repeated until the preset convergence condition is met and the optimal drilling parameters are output.
2. The method according to claim 1, characterized in that, The acquisition of drilling parameter types and geological data obtained during drilling, and the establishment of an initial population containing multiple drilling parameter groups based on the drilling parameter types, include: Determine the safe operating control boundaries for drilling pressure, rotation speed, and mud pump volume; Collect geological data measured while drilling during the drilling process to obtain the lithological response sequence matrix corresponding to the current well depth; Within the safe operation control boundary, multiple drilling parameter combinations are randomly generated using a uniform distribution generation algorithm. Each drilling parameter combination is treated as an individual, and multiple individuals constitute the initial population.
3. The method according to claim 1, characterized in that, The step of taking the initial population as the current population, performing an iterative optimization operation on each target parameter set in the current population, calculating the parameter similarity between other parameter sets and the target parameter set based on the parameter values, and determining the nearest reference parameter set includes: Extract the target parameter value sequence of the target parameter group and the reference parameter value sequence of other parameter groups in the population; The target parameter value sequence and the reference parameter value sequence are dimensionless. Calculate the sum of squared differences between corresponding elements in the dimensionless sequence of target parameter values and the sequence of reference parameter values; The parameter similarity is calculated based on the reciprocal of the sum of squared differences. The parameter group with the highest parameter similarity is determined as the nearest reference parameter group.
4. The method according to claim 3, characterized in that, The step of determining the guidance weight based on the parameter similarity between the target parameter set and the population's historical global optimal parameter set, and then weighting and fusing the neighboring reference parameter set and the population's historical global optimal parameter set according to the guidance weight to generate a composite guidance parameter set includes: Obtain the first parameter vector corresponding to the target parameter set and the second parameter vector corresponding to the population's historical global optimal parameter set, and perform dimensionless processing on them respectively; Calculate the cosine similarity between the dimensionless first parameter vector and the second parameter vector, and use the normalized cosine similarity as the guiding weight. The first fusion component is obtained by multiplying the guiding weight by the parameter value of the population's historical global optimal parameter set; The difference between 1 and the guiding weight is multiplied by the parameter value of the neighboring reference parameter group to obtain the second fusion component; The first fusion component and the second fusion component are added together to obtain the composite guidance parameter set.
5. The method according to claim 1, characterized in that, The step of determining the crossover probability based on the difference between the drilling rate evaluation function value of the target parameter set and the population's historical global optimal parameter set, and performing parameter crossover on the target parameter set and the composite steering parameter set based on the crossover probability to generate an experimental parameter set includes: Obtain the first drilling rate evaluation function value of the target parameter set and the second drilling rate evaluation function value of the population's historical global optimal parameter set; Calculate the absolute value of the difference between the second drilling speed evaluation function value and the first drilling speed evaluation function value, and map the absolute value of the difference to a preset probability interval to obtain the cross probability; For each parameter dimension, a random number between 0 and 1 is generated. If the random number for the corresponding dimension is less than the crossover probability, the parameter value of the target parameter group is replaced with the parameter value of the composite guiding parameter group in that dimension. If the random number in the corresponding dimension is greater than or equal to the crossover probability, then the parameter values of the target parameter set in that dimension are retained, thereby generating the experimental parameter set.
6. The method according to claim 1, characterized in that, The step of determining whether to replace the target parameter set with the test parameter set based on the drilling rate evaluation function value of the test parameter set includes: The test parameter set and the geological data measured while drilling are substituted into the pre-constructed drilling speed prediction model to calculate and output the corresponding expected drilling speed, which is used as the drilling speed evaluation function value of the test parameter set. Compare the drilling rate evaluation function values of the test parameter set with those of the target parameter set; When the drilling rate evaluation function value of the experimental parameter group is greater than the drilling rate evaluation function value of the target parameter group, the experimental parameter group is determined to be superior to the target parameter group, and the individual at the corresponding position in the next generation population is updated to the experimental parameter group and the target parameter group is discarded.
7. A kilometer-level intelligent drilling parameter adaptive optimization control system for vertical shafts, characterized in that, Includes the following modules: The acquisition module is used to acquire drilling parameter types and geological data measured while drilling, and to establish an initial population containing multiple drilling parameter groups based on the drilling parameter types. The calculation module is used to take the initial population as the current population, perform iterative optimization operation on each target parameter group in the current population, calculate the parameter similarity between other parameter groups and the target parameter group based on the parameter values, determine the neighboring reference parameter group, determine the guidance weight based on the parameter similarity between the target parameter group and the population's historical global optimal parameter group, and perform weighted fusion of the neighboring reference parameter group and the population's historical global optimal parameter group based on the guidance weight to generate a composite guidance parameter group; The generation module is used to determine the crossover probability based on the difference between the drilling rate evaluation function value of the target parameter set and the population historical global optimal parameter set, perform parameter crossover on the target parameter set and the composite steering parameter set according to the crossover probability, generate a test parameter set, determine whether to replace the target parameter set with the test parameter set according to the drilling rate evaluation function value of the test parameter set, and repeat the iterative optimization operation until the preset convergence condition is met and the optimal drilling parameters are output.
8. The system according to claim 7, characterized in that, The acquisition of drilling parameter types and geological data obtained during drilling, and the establishment of an initial population containing multiple drilling parameter groups based on the drilling parameter types, include: Determine the safe operating control boundaries for drilling pressure, rotation speed, and mud pump volume; Collect geological data measured while drilling during the drilling process to obtain the lithological response sequence matrix corresponding to the current well depth; Within the safe operation control boundary, multiple drilling parameter combinations are randomly generated using a uniform distribution generation algorithm. Each drilling parameter combination is treated as an individual, and multiple individuals constitute the initial population.
9. The system according to claim 7, characterized in that, The step of taking the initial population as the current population, performing an iterative optimization operation on each target parameter set in the current population, calculating the parameter similarity between other parameter sets and the target parameter set based on the parameter values, and determining the nearest reference parameter set includes: Extract the target parameter value sequence of the target parameter group and the reference parameter value sequence of other parameter groups in the population; The target parameter value sequence and the reference parameter value sequence are dimensionless. Calculate the sum of squared differences between corresponding elements in the dimensionless sequence of target parameter values and the sequence of reference parameter values; The parameter similarity is calculated based on the reciprocal of the sum of squared differences. The parameter group with the highest parameter similarity is determined as the nearest reference parameter group.
10. The system according to claim 7, characterized in that, The step of determining the guidance weight based on the parameter similarity between the target parameter set and the population's historical global optimal parameter set, and then weighting and fusing the neighboring reference parameter set and the population's historical global optimal parameter set according to the guidance weight to generate a composite guidance parameter set includes: Obtain the first parameter vector corresponding to the target parameter set and the second parameter vector corresponding to the population's historical global optimal parameter set, and perform dimensionless processing on them respectively; Calculate the cosine similarity between the dimensionless first parameter vector and the second parameter vector, and use the normalized cosine similarity as the guiding weight. The first fusion component is obtained by multiplying the guiding weight by the parameter value of the population's historical global optimal parameter set; The difference between 1 and the guiding weight is multiplied by the parameter value of the neighboring reference parameter group to obtain the second fusion component; The first fusion component and the second fusion component are added together to obtain the composite guidance parameter set.
