Indoor large-scale static pressure pile hydraulic control method based on artificial intelligence

Abstract

The present application relates to the technical field of hydraulic control of static pressure pile, in particular to an indoor large-scale static pressure pile hydraulic control method based on artificial intelligence; the method cooperates through environment perception, AI stratum prediction, pile body state monitoring, hydraulic dynamic control and central control unit, first collects basic parameters to train LSTM model, then collects environment and pile body data in real time, calculates hydraulic target pressure, correction coefficient and real-time flow through multiple formula progressive calculation to form closed-loop control; the method solves the problem that the prior art does not couple indoor wall constraint, hydraulic oil temperature and humidity and stratum mutation, improves pile pressing precision and system stability, and reduces overload failure rate and energy consumption.

Description

Technical Field

[0001] This invention relates to the field of static pressure pile construction technology, specifically to an artificial intelligence-based hydraulic control method for large-scale indoor static pressure piles. Background Technology

[0002] In indoor large-scale static pressure pile construction, the hydraulic control system is the core to ensure the smooth driving of the pile. However, existing technologies have significant shortcomings in this scenario. Current technologies typically focus only on the single control relationship between the vertical pressure of the pile and the driving depth, neglecting the multi-factor coupling effects unique to the indoor environment. Firstly, the confined indoor space causes non-uniform lateral stress due to lateral compression from surrounding walls during pile driving. This stress leads to abnormal fluctuations in the hydraulic system load. Existing solutions assume no lateral stress interference, easily causing hydraulic system overload and equipment damage. Secondly, poor indoor ventilation results in slow heat dissipation from the hydraulic system during pile driving. Simultaneously, changes in indoor humidity alter the viscosity of the hydraulic oil. The coupling effect of temperature and humidity causes lag in the response of the hydraulic actuator. Existing solutions only compensate for single temperature or humidity factors, failing to eliminate the response delay caused by coupling effects. Furthermore, indoor large-scale piles often need to traverse different strata such as backfill and original soil layers. Sudden changes in stratum interfaces cause a sharp increase or decrease in pile end resistance. Existing PID control, due to its slow response speed, cannot adjust the hydraulic output in time, easily leading to pile cracking. The aforementioned multi-factor coupling problem overlaps with each other, seriously affecting the construction quality and safety of large-scale static pressure piles indoors.

[0003] Based on the above problems, there is an urgent need for a hydraulic control method that can comprehensively address the issues affecting the construction quality and safety of large-scale static pressure piles indoors. Summary of the Invention

[0004] This invention provides an artificial intelligence-based hydraulic control method for large-scale indoor static pressure piles, including the steps of collecting basic indoor environmental data, pile parameters, and hydraulic system parameters and inputting them into the control unit, and training an LSTM network using historical pile driving data; it also includes a real-time control stage and a closed-loop adjustment stage; in the real-time control stage, the environmental perception module collects indoor temperature difference, relative humidity, and wall lateral stress at a preset period, the pile condition monitoring module collects pile strain, pile end resistance, and real-time hydraulic system pressure, and the AI ​​stratum prediction module outputs the probability of stratum mutation and the real-time change in pile end resistance; the central control unit first calculates the target pressure of the hydraulic system based on the collected data, then calculates the target pressure correction coefficient based on the output of the AI ​​stratum prediction module, and finally calculates the real-time output flow rate of the hydraulic pump based on the target pressure, correction coefficient, and real-time collected hydraulic system pressure, and sends the flow rate command to the hydraulic dynamic control module to drive the cylinder to drive the pile; in the closed-loop adjustment stage, the actual pile strain is compared with the design threshold at a preset period, and the flow rate calculation parameters are adjusted when the threshold is exceeded, and the correction coefficient calculation parameters are temporarily adjusted when the probability of stratum mutation output by the AI ​​stratum prediction module exceeds the preset value.

[0005] Preferably, the number of historical pile driving data collected is no less than 500 sets, and the prediction accuracy of ground change after the LSTM network training is no less than 92%; the acquisition cycle of the environmental perception module and the pile condition monitoring module is set to 0.1 seconds, the cycle of comparing the actual strain of the pile body with the design threshold in the closed-loop adjustment stage is set to 1 second, and the design threshold of pile body strain is set to 1500με.

[0006] Further preferred indoor environmental basic data include wall stiffness and initial temperature and humidity; pile parameters include pile material and side surface area; hydraulic system parameters include effective working area of ​​hydraulic cylinder and pressure loss coefficient of hydraulic pipeline; after the central control unit inputs the parameters, the parameters are normalized. The normalization process adopts the linear normalization method to map the parameter values ​​to the 0-1 range. The mapping formula is parameter normalized value = (parameter actual value - parameter minimum value) / (parameter maximum value - parameter minimum value).

[0007] In a further optimized manner, when training the AI ​​formation prediction module, the input historical pile driving data includes driving depth, real-time pressure, and pile strain. During the training process, cross-validation is used, and the historical data is divided into training set and validation set in a 7:3 ratio. The training set is used for iterative optimization of model parameters, and the validation set is used to evaluate the model's generalization ability. When the prediction error on the validation set is less than the preset error threshold for 5 consecutive training rounds, the model training is stopped.

[0008] Furthermore, the central control unit uses the following formula to calculate the target pressure of the hydraulic system:

[0009] ;

[0010] In the formula, The target pressure for the hydraulic system is expressed in MPa. This is the reference force required for the vertical driving of the pile, expressed in kN. The wall constraint influence coefficient is obtained by fitting the indoor wall stiffness with the pile spacing. The lateral stress generated by the compression of the pile by the wall is expressed in MPa. The lateral surface area of ​​the pile is expressed in units of... ; This refers to the dynamic viscosity of the hydraulic oil, expressed in Pa·s. The effective working area of ​​the hydraulic cylinder, in units of ; The temperature compensation coefficient is determined by the hydraulic oil type. This represents the difference between the real-time temperature and the initial temperature of the hydraulic oil, in °C. The humidity influence coefficient is obtained from an indoor humidity sensor. The value represents indoor relative humidity, expressed as %.

[0011] A further preferred embodiment is that the central control unit uses the following formula when calculating the target pressure correction factor:

[0012] ;

[0013] In the formula, This is a correction factor for the target pressure, and has no unit. The weight coefficients are predicted by the AI ​​and determined by the training accuracy of the LSTM model. The output value of the LSTM network represents the probability of abrupt changes in the formation, and its value ranges from 0 to 1. This includes the past 5 sets of historical pile driving data, including driving depth, real-time pressure, and pile strain. This refers to the density of the soil layer, in units of... ; This refers to the pile driving speed, expressed in m / min. This is the sensitivity coefficient for sudden changes in pile end resistance; This represents the real-time change in pile end resistance, expressed in kN.

[0014] A further preferred embodiment is that the central control unit uses the following formula to calculate the real-time output flow of the hydraulic pump:

[0015] ;

[0016] In the formula, This refers to the real-time output flow rate of the hydraulic pump, expressed in L / min. This refers to the real-time pressure of the hydraulic system, expressed in MPa. The volume of the hydraulic cylinder is expressed in liters (L). The control period is measured in seconds (s). The density of hydraulic oil is expressed in units of... ; This refers to the pressure loss coefficient of the hydraulic pipeline. This is the pile strain compensation coefficient; The value represents the real-time strain of the pile body, expressed in με.

[0017] In a further preferred embodiment, after receiving the flow command, the hydraulic dynamic control module first verifies the validity of the command. The verification includes whether the flow command value is within the rated flow range of the hydraulic pump. If it exceeds the range, an alarm signal is output and the upper limit of the rated flow is adopted. After the verification is passed, the hydraulic dynamic control module drives the hydraulic pump motor through the PWM signal and adjusts the motor speed to achieve real-time flow output. The motor speed and flow output have a linear relationship, and the relationship expression is motor speed = (flow command / hydraulic pump displacement) × 60, with the unit being r / min.

[0018] In a further preferred embodiment, during the closed-loop adjustment phase, when the actual strain of the pile exceeds the design threshold, the pile strain compensation coefficient in the flow calculation formula is increased by 0.01, and this coefficient is maintained until the next comparison cycle after each adjustment; when the probability of a sudden change in the formation output by the AI ​​formation prediction module exceeds 80%, the AI ​​prediction weight coefficient in the target pressure correction coefficient calculation formula is increased by 0.05, and when the probability of a sudden change in the formation drops below 50%, the initial value of the AI ​​prediction weight coefficient is restored.

[0019] Further preferred features include a fault diagnosis phase. During this phase, the central control unit monitors the operating status parameters of each module in real time. These parameters include the acquisition frequency of the environmental perception module, the calculation time of the AI ​​geological prediction module, the sensor signal strength of the pile condition monitoring module, and the output current of the hydraulic dynamic control module. When the operating status parameters of a module exceed the preset normal range, the central control unit determines that the module is faulty, outputs the faulty module identifier and fault type, and switches to a backup module or initiates an emergency control program. The emergency control program uses a preset fixed flow output value to drive the hydraulic system, ensuring the stable stopping of the pile.

[0020] Technical effects:

[0021] The inventive technical point of this invention lies in the fact that, through the collaboration of environmental perception, AI-based geological prediction, and pile monitoring modules, it integrates indoor wall lateral stress, hydraulic oil temperature and humidity coupling, and geological abrupt changes into a unified control system. It employs multi-formula progressive calculations to achieve dynamic hydraulic adjustment, forming a complete logic from parameter acquisition, AI prediction, flow control to closed-loop optimization. This technical point precisely solves the problems of hydraulic overload, response lag, and pile cracking caused by the coupling of multiple indoor factors in the prior art, ensuring the construction quality and system safety of large-scale static pressure piles indoors. Attached Figure Description

[0022] Figure 1 This is a flowchart of the hydraulic control method for large-scale indoor static pressure piles based on artificial intelligence, as described in this application. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0024] Traditional technical solutions have the following technical problems: In the construction of large-scale static pressure piles indoors, the existing hydraulic control scheme does not take into account the influence of lateral stress of the wall, hydraulic oil temperature and humidity and sudden changes in the stratum, which makes the hydraulic system prone to overload, the actuator response lags, and the pile body is prone to cracking when the stratum changes, which seriously affects the construction quality.

[0025] Based on this, please refer to Figure 1 This embodiment provides an artificial intelligence-based indoor large-scale static pressure pile hydraulic control method, including the steps of collecting indoor environmental basic data, pile parameters and hydraulic system parameters and inputting them into the control unit, and training an LSTM network with historical pile driving data; it also includes a real-time control stage and a closed-loop adjustment stage;

[0026] S1: In the real-time control phase, the environmental perception module collects indoor temperature difference, relative humidity, and wall lateral stress according to a preset cycle. The pile condition monitoring module collects pile strain, pile end resistance, and real-time pressure of the hydraulic system. The AI ​​stratum prediction module outputs the probability of stratum mutation and the real-time change of pile end resistance.

[0027] S2: The central control unit first calculates the target pressure of the hydraulic system based on the collected data, then calculates the target pressure correction coefficient based on the output of the AI ​​formation prediction module, and finally calculates the real-time output flow of the hydraulic pump based on the target pressure, correction coefficient and real-time collected hydraulic system pressure, and sends the flow command to the hydraulic dynamic control module to drive the oil cylinder to press the pile.

[0028] S3: During the closed-loop adjustment phase, the actual strain of the pile body is compared with the design threshold according to the preset cycle. When the threshold is exceeded, the flow calculation parameters are adjusted. When the probability of a sudden change in the formation output by the AI ​​formation prediction module exceeds the preset value, the correction coefficient calculation parameters are temporarily adjusted.

[0029] The core of this technical solution is the construction of a multi-module collaborative closed-loop control system. The preprocessing stage requires the input of basic parameters and training of the LSTM model. These basic parameters include environmental data such as indoor wall stiffness and initial temperature and humidity, pile material and side surface area, and hydraulic system data such as cylinder area and pipeline loss coefficient. These parameters provide the foundation for subsequent calculations. The LSTM model training uses over 500 sets of historical pile driving data to ensure the accuracy of predicting ground changes. The real-time control stage is the core execution phase. The environmental sensing module collects data such as temperature and humidity differences and lateral stress every 0.1 seconds, while the pile monitoring module simultaneously collects strain, pile end resistance, and system pressure. The AI ​​module outputs the probability of ground change and the amount of resistance change. The central control unit sequentially calculates the target pressure, correction coefficient, and real-time flow rate, achieving seamless integration from data acquisition to command output. The closed-loop adjustment stage ensures dynamic optimization of the system when strain exceeds the threshold or when the risk of ground change is high, through periodic comparison and parameter fine-tuning, thus guaranteeing construction stability.

[0030] This solution addresses the shortcomings of existing technologies in single-control by using multi-module collaboration and multi-factor coupled calculation, avoiding hydraulic overload, response lag and pile cracking, and improving construction accuracy and system reliability.

[0031] Traditional technical solutions have the following technical problems: the existing solutions do not specify the amount of historical data to be collected, the accuracy of model prediction, and the collection cycle of each module, resulting in poor generalization ability of LSTM models, insufficient timeliness of data collection, and inability to accurately support hydraulic control decisions.

[0032] Based on this, the number of historical pile driving data collected is no less than 500 sets, and the prediction accuracy of ground change after the LSTM network training is no less than 92%; the collection cycle of the environmental perception module and the pile condition monitoring module is set to 0.1 seconds, the cycle of comparing the actual strain of the pile body with the design threshold in the closed-loop adjustment stage is set to 1 second, and the design threshold of pile body strain is set to 1500με.

[0033] This technical solution focuses on parameter quantification and cycle setting. A minimum of 500 sets of historical data are collected because insufficient data leads to inadequate LSTM model training, failing to cover pile driving scenarios in different geological formations and environments. More than 500 sets of data ensure the model learns enough feature patterns. The 92% prediction accuracy is set based on the response speed requirements of indoor construction to sudden geological changes; accuracy below this level would lead to deviations in correction coefficient calculations, increasing the risk of pile cracking. The 0.1-second acquisition cycle is determined by comprehensively considering the hydraulic system's response speed and data processing capabilities; a cycle that is too long would cause data lag, failing to reflect load changes in a timely manner. The 1-second strain comparison cycle balances the timeliness of adjustment with system stability, avoiding frequent adjustments that cause fluctuations in hydraulic output. The 1500με strain threshold is determined based on the tensile strength of common pile materials, ensuring that the pile body will not suffer permanent damage due to excessive strain. This solution improves the reliability of the LSTM model and the timeliness of data acquisition by clearly defining quantified parameters and cycles, providing data support for precise hydraulic control and reducing construction risks.

[0034] Traditional technical solutions have the following technical problems: existing solutions do not preprocess the input basic parameters, and the dimensions and numerical ranges of different parameters vary greatly, which can easily lead to deviations in subsequent calculation results and affect the accuracy of hydraulic control.

[0035] Based on this, the basic indoor environmental data includes wall stiffness and initial temperature and humidity; the pile parameters include pile material and side surface area; and the hydraulic system parameters include the effective working area of ​​the hydraulic cylinder and the pressure loss coefficient of the hydraulic pipeline. After the central control unit inputs the parameters, the parameters are normalized. The normalization process adopts the linear normalization method to map the parameter values ​​to the 0-1 range. The mapping formula is: parameter normalized value = (parameter actual value - parameter minimum value) / (parameter maximum value - parameter minimum value).

[0036] The key to this technical solution lies in the classification and normalization of basic parameters. The classification clarifies three core data categories: environment, pile body, and hydraulic system. Wall stiffness affects lateral stress calculation; initial temperature and humidity are related to hydraulic oil viscosity; pile side surface area is related to vertical pressure; and cylinder area and pipeline loss coefficient directly participate in flow rate calculation. Categorized input ensures the accuracy of parameter retrieval. The normalization process uses linear normalization, mapping parameters with different dimensions (e.g., wall stiffness in N / m, temperature and humidity in ℃ / %) to the 0-1 range, eliminating calculation interference caused by dimensional differences. For example, wall stiffness might be 10^6 N / m, while humidity might be 50%. Directly using these parameters in calculations without processing would mask the humidity's influence. Normalization makes the weighting of each parameter's impact on the calculation results more reasonable, improving the accuracy of target pressure and flow rate calculations. This solution, through parameter classification and normalization, eliminates dimensional interference, ensures calculation accuracy, and lays the foundation for the accuracy of subsequent hydraulic control commands.

[0037] Traditional technical solutions have the following technical problems: existing solutions do not clearly define the data partitioning method and stopping conditions for LSTM model training, which leads to the model being prone to overfitting or underfitting, poor generalization ability, and inability to accurately predict formation changes.

[0038] Based on this, when training the AI ​​formation prediction module, the input historical pile driving data includes driving depth, real-time pressure and pile strain. During the training process, cross-validation is used to divide the historical data into training set and validation set in a 7:3 ratio. The training set is used for iterative optimization of model parameters, and the validation set is used to evaluate the model's generalization ability. When the prediction error on the validation set is less than the preset error threshold for 5 consecutive training rounds, the model training is stopped.

[0039] This technical solution details the LSTM model training process. The input data includes indentation depth, real-time pressure, and pile strain because these three parameters directly reflect the interaction between the soil and the pile. Changes in indentation depth reflect differences in soil hardness, real-time pressure fluctuations are related to pile end resistance, and pile strain reflects load changes. The combination of these three parameters comprehensively characterizes the soil state, providing sufficient features for abrupt change prediction. Cross-validation is used, dividing the training and validation sets in a 7:3 ratio. The 70% training set ensures the model has sufficient data to learn feature patterns, while the 30% validation set effectively evaluates the model's adaptability to new data, avoiding overfitting (smaller error on the training set but larger error on the validation set).

[0040] The stopping condition is set to the validation set prediction error being less than a threshold for five consecutive epochs, rather than a single epoch. This avoids misjudgments caused by random factors and ensures stable model convergence. For example, stopping training only after the error reaches the threshold in a single epoch might lead to insufficient model optimization and inadequate prediction accuracy in subsequent epochs. Meeting the threshold for five consecutive epochs indicates that the model parameters have approached their optimality and that the generalization ability is reliable. This approach improves the generalization ability and prediction stability of the LSTM model by clearly defining the training data, partitioning method, and stopping condition, ensuring accurate prediction of geological mutations.

[0041] Traditional technical solutions have the following technical problems: existing solutions do not establish a target pressure calculation model that includes the lateral stress of the wall and the coupling of hydraulic oil temperature and humidity, resulting in a mismatch between the target pressure and the actual load, and the hydraulic system is prone to overload or insufficient output.

[0042] Based on this, the central control unit uses the following formula to calculate the target pressure of the hydraulic system:

[0043] ;

[0044] In the formula, The target pressure for the hydraulic system is expressed in MPa. This is the reference force required for the vertical driving of the pile, expressed in kN. The wall constraint influence coefficient is obtained by fitting the indoor wall stiffness with the pile spacing. The lateral stress generated by the compression of the pile by the wall is expressed in MPa. The lateral surface area of ​​the pile is expressed in units of... ; This refers to the dynamic viscosity of the hydraulic oil, expressed in Pa·s. The effective working area of ​​the hydraulic cylinder, in units of ; The temperature compensation coefficient is determined by the hydraulic oil type. This represents the difference between the real-time temperature and the initial temperature of the hydraulic oil, in °C. The humidity influence coefficient is obtained from an indoor humidity sensor. The value represents indoor relative humidity, expressed as %.

[0045] This technical solution constructs a target pressure calculation model using a formula, which consists of two parts, the first part... Used to calculate the foundation pressure required to overcome the vertical load of the piles and the lateral load of the wall. It is the reference force for vertical pile driving. Then quantify the additional load generated by the lateral stress of the wall. By fitting the wall stiffness to the pile spacing, the accuracy of lateral load calculations is ensured. This demonstrates the impact of hydraulic oil viscosity on pressure transmission; higher viscosity results in greater pressure loss. The effective area of ​​the hydraulic cylinder determines the relationship between pressure and thrust. Part Two Used to compensate for the effect of temperature and humidity coupling on pressure. This reflects temperature changes; as temperature increases, the viscosity of the hydraulic oil decreases, improving pressure transmission efficiency, necessitating adjustment of the compensation amount. and Together, these parameters reflect the impact of humidity on oil properties. Increased humidity may lead to oil emulsification and viscosity changes. The exponential function relating humidity to humidity better reflects actual effects. All parameters have clearly defined dimensions, ensuring the accuracy of the physical meaning of the calculated results. For kN, MPa (i.e.) ), for (Right now The product of these two units is kN. Consistency is ensured, so that the numerator and denominator are consistent. and ( The combination of these elements ultimately results in the first part being in MPa, which is consistent with... Consistent.

[0046] The core function of this formula is to calculate the target pressure that the hydraulic system needs to output during the construction of large-scale static pressure piles indoors. Its design logic closely revolves around the pain points of the coupling between the wall constraints and hydraulic oil characteristics unique to indoor scenarios, making up for the shortcomings of existing technologies that only consider vertical pile driving force and ignore the synergistic influence of multiple factors.

[0047] The first part of the formula is This part essentially involves a comprehensive calculation of the total load on the pile and the transmission efficiency of the hydraulic system. Among these, The reference force required for the vertical driving of the pile only reflects the basic requirement for the pile to overcome the vertical resistance of the strata. However, in indoor scenarios, the force generated by the pile being squeezed by the surrounding walls... Lateral stress can create additional loads. If these are not included in the calculation, they can cause the hydraulic system to output pressure lower than the actual demand, leading to pile driving stagnation or equipment overload. The wall constraint influence coefficient is introduced to quantify the degree of influence of different wall stiffness and pile spacing on lateral stress—for example, the greater the wall stiffness and the smaller the distance between the piles and the wall, the better. The closer the value is to 1.2, the higher the contribution of lateral stress to the total load, thus ensuring the accuracy of the additional load calculation; The lateral surface area of ​​the pile is a key parameter for converting lateral stress into lateral force. A larger lateral surface area results in a wider range of compression from the wall onto the pile, leading to a greater lateral force. Therefore, it needs to be considered in conjunction with... Multiply to obtain the complete lateral additional force, then combine with The sum of these amounts constitutes the total load required to drive the pile in.

[0048] denominator This focuses on the pressure transmission characteristics of the hydraulic system. The dynamic viscosity of hydraulic oil directly affects the internal leakage and pressure loss of the hydraulic system—when poor indoor ventilation causes the oil temperature to rise... It will decrease, and the leakage within the system will increase. If not corrected, the actual pressure acting on the cylinder will be lower than the calculated value. The effective working area of ​​a hydraulic cylinder is the core of the conversion between pressure and thrust. According to the principles of mechanics, the cylinder thrust is equal to the product of the pressure and the working area; therefore, it needs to be converted through... The total load is converted into the corresponding pressure demand. Multiplying the two can correct the influence of fluid characteristics and mechanical structure on pressure transmission, ensuring that the calculated pressure can actually drive the cylinder to overcome the total load.

[0049] The second part of the formula is This section addresses the dynamic compensation for the coupled effects of temperature and humidity on hydraulic oil performance. Existing technologies often employ single temperature or humidity compensation, which cannot cope with the fluctuations in oil performance caused by the combined changes in indoor temperature and humidity. The real-time temperature difference between the hydraulic oil and its initial temperature reflects the effect of temperature on oil viscosity; an increase in temperature will cause... The reduction leads to a decrease in pressure transmission efficiency, therefore it is necessary to... Temperature compensation coefficient adjustment range The value is determined by the type of hydraulic oil. For example, hydraulic oils with higher viscosity grades are more sensitive to temperature changes. The value is close to 0.1. And... The introduction of indoor relative humidity is because increased humidity can lead to hydraulic oil emulsification. The viscosity change of emulsified oil deviates from the linear relationship under the influence of temperature alone; the higher the humidity, the weaker the effect of temperature on viscosity. Therefore, an exponential function is used. To fit this nonlinear relationship, The higher the humidity influence coefficient, the stronger the weakening effect of humidity on temperature compensation. For example, when... When it rises from 50% to 80%, The value will decrease significantly, thereby reducing the amount of temperature compensation and avoiding excessive compensation that could lead to high pressure.

[0050] This solution calculates the target pressure using a multi-factor coupling formula to ensure that the pressure matches the actual load and avoids hydraulic system overload or insufficient output.

[0051] Traditional technical solutions have the following technical problems: existing solutions do not consider the impact of sudden changes in strata on target pressure, and cannot adjust the pressure in advance. Sudden changes in strata can easily lead to pile cracking or hydraulic impact.

[0052] Based on this, the central control unit uses the following formula when calculating the target pressure correction factor:

[0053] ;

[0054] In the formula, This is a correction factor for the target pressure, and has no unit. The weight coefficients are predicted by the AI ​​and determined by the training accuracy of the LSTM model. The output value of the LSTM network represents the probability of abrupt changes in the formation, and its value ranges from 0 to 1. This includes the past 5 sets of historical pile driving data, including driving depth, real-time pressure, and pile strain. This refers to the density of the soil layer, in units of... ; This refers to the pile driving speed, expressed in m / min. This is the sensitivity coefficient for sudden changes in pile end resistance; This represents the real-time change in pile end resistance, expressed in kN.

[0055] This technical solution achieves predictive compensation for abrupt formation changes through a correction coefficient formula. With 1 as the baseline, when there is no risk of mutation, the correction coefficient is close to 1, and the target pressure does not need to be adjusted significantly; when there is a risk of mutation, the correction coefficient is increased or decreased to achieve early optimization of pressure. Some adjustments were made based on AI prediction results. Output the probability of abrupt changes in the formation. The higher the probability, the larger this value, and the greater the adjustment range of the correction factor. Determined by model accuracy; the higher the accuracy, The value can be increased appropriately to allow the AI ​​prediction results to have a greater impact on the correction coefficient. Partially adjusted based on real-time resistance changes. This reflects the instantaneous change in pile end resistance. The greater the change, the more likely a sudden change in the formation may have occurred, and the larger this value will be. The sensitivity coefficient is determined based on the pile material and stratum type to ensure appropriate sensitivity in response to sudden changes in resistance. , , As input to the LSTM model Provides historical trend data. Reflecting the basic characteristics of the strata, The influence of pile driving speed on resistance is considered, and the combination of these three factors ensures that the LSTM model can accurately predict the probability of abrupt changes. The parameters in the formula work together to achieve dual prediction of abrupt changes in the formation (AI prediction + real-time resistance change), making the correction coefficient more consistent with actual working conditions.

[0056] The core purpose of this formula is to calculate the coefficient used to correct the hydraulic target pressure through a dual mechanism of AI prediction and real-time feedback, so as to cope with the risk of sudden changes when large-scale indoor piles pass through different strata. It solves the problems of pile cracking and hydraulic impact caused by the inability to predict sudden changes in strata in existing technologies. Its design logic directly supports the technical feature of the collaboration between the AI ​​strata prediction module and the central control unit in the above embodiment.

[0057] The formula uses "1" as the baseline value for the correction coefficient. The technical significance of this design is that, when there is no risk of abrupt changes in formation, A value close to 1 means there is no need to... Additional adjustments are made to the target pressure to avoid unnecessary pressure fluctuations affecting pile driving stability; when there is a risk of sudden change, the subsequent two measures are superimposed to ensure stability. By deviating by 1, pressure can be corrected in advance or adjusted in real time, ensuring that the hydraulic system can quickly adapt to changes in the formation.

[0058] First item It is a forward-looking correction term based on AI prediction. Its core is to utilize the temporal feature learning ability of the Long Short-Term Memory (LSTM) network to identify abrupt changes in geological formations in advance. The selection of input parameters for the LSTM model follows a clear technical logic: The past five sets of historical pile driving data include driving depth, real-time pressure, and pile strain. These data can reflect the changing trend of ground resistance—for example, when the driving depth increases, the real-time pressure suddenly slows down, which may indicate that it is about to enter a loose stratum. Soil density is a core indicator of soil physical properties. Higher soil density indicates a more compact formation, resulting in greater variations in pile end resistance during abrupt changes. LSTM models can effectively address this. Predict the range of resistance fluctuations that may be caused by sudden changes; The pile driving speed affects the interaction time between the pile tip and the stratum. The faster the speed, the shorter the response time of the pile tip to abrupt changes in the stratum. The LSTM model needs to incorporate the speed parameter to adjust its prediction sensitivity. The LSTM model outputs the probability of abrupt changes in the stratum in the range of 0-1. The closer the probability is to 1, the higher the risk of abrupt change. In this case, further adjustments are needed. (AI prediction weighting coefficients) Adjust this part for Contribution level - The value is determined by the model's training accuracy; for example, when the model's prediction accuracy for abrupt changes in formation reaches 95%, A value close to 0.3 ensures that the AI ​​prediction results fully guide the correction direction; if the model accuracy is low, then... The value was reduced to 0.1 to avoid overcorrection due to prediction errors.

[0059] Second item It is an instant feedback correction term based on real-time data. Its design purpose is to make up for the prediction lag problem that may exist in the LSTM model, forming a dual guarantee from prediction to feedback. The real-time change in pile tip resistance is the most direct and immediate reflection of the soil condition—when a sudden change occurs in the soil strata, the pile tip resistance will increase or decrease sharply in a short period of time, for example, when moving from backfill soil into the original soil layer. The current may increase from 10kN to 50kN. By monitoring this change, we can quickly capture sudden situations that the LSTM model failed to predict in time. The role of the (pile end resistance change sensitivity coefficient) is to quantify... The degree of influence of the correction factor needs to be determined in conjunction with the pile material and stratum type: for example, high-strength piles can withstand greater sudden changes in resistance. A value close to 0.08 is chosen to avoid triggering significant corrections due to slight changes in resistance; however, low-strength piles are more sensitive to sudden changes in resistance. The value was increased to 0.15 to ensure timely correction and protect the pile body.

[0060] The collaborative logic of the two correction terms is reflected in the fact that when the LSTM model predicts a high mutation probability, The item will make Initially increase the target pressure to prepare for the impending surge in resistance; if at this time... It also showed a significant increase. The item will further increase This enables "real-time reinforcement and correction based on prediction"; conversely, if the LSTM model fails to predict abrupt changes (e.g., due to historical data not covering the stratigraphic type), but... A sudden increase Each item can trigger corrections independently, avoiding missed detections. This collaborative mechanism ensures the reliability and timeliness of the correction coefficients, preventing insufficient or excessive corrections caused by relying solely on AI predictions or real-time data.

[0061] This scheme uses a correction coefficient formula to compensate for the effects of sudden changes in strata in advance, thus avoiding pile cracking and hydraulic impact.

[0062] Traditional technical solutions have the following technical problems: existing solutions do not combine target pressure, correction coefficient and real-time pressure to calculate flow rate, resulting in a disconnect between flow output and actual demand, and lag in response of hydraulic actuators.

[0063] Based on this, the central control unit uses the following formula to calculate the real-time output flow of the hydraulic pump:

[0064] ;

[0065] In the formula, This refers to the real-time output flow rate of the hydraulic pump, expressed in L / min. This refers to the real-time pressure of the hydraulic system, expressed in MPa. The volume of the hydraulic cylinder is expressed in liters (L). The control period is measured in seconds (s). The density of hydraulic oil is expressed in units of... ; This refers to the pressure loss coefficient of the hydraulic pipeline. This is the pile strain compensation coefficient; The value represents the real-time strain of the pile body, expressed in με.

[0066] This technical solution achieves precise conversion from pressure to flow through a flow formula. The first part of the formula... Used to calculate the base flow rate required to compensate for the pressure difference. It is the revised target pressure, minus The pressure difference is obtained; the larger the pressure difference, the greater the required flow rate, in order to quickly adjust the system pressure to the target value. The volume of the hydraulic cylinder determines the relationship between the flow rate and the extension / retraction speed of the hydraulic cylinder. To control the cycle and ensure that the flow calculation and control rhythm are synchronized; and Consider oil density and pipeline pressure loss separately to avoid insufficient actual flow due to these losses. Part Two Used to fine-tune the flow rate based on the strain of the pile. This reflects the stress state of the pile body. The greater the strain, the closer the load on the pile body is to its limit. Appropriately reducing the flow rate in this part can prevent the strain from increasing further. The compensation coefficient is determined based on the pile strength to ensure appropriate fine-tuning. The dimensions of each parameter are matched, for example... and The units are all in MPa ( ), difference multiplied by ( To obtain the energy unit (N·m), divide by (s) to obtain the power unit (W), then divide by ( )and Finally, the unit conversion is combined to obtain L / min, ensuring the accuracy of the flow rate calculation result.

[0067] This formula is the core bridge connecting pressure demand and execution output. Its function is to convert the target pressure after geological mutation correction into the real-time flow rate that the hydraulic pump needs to output, ensuring that the action speed of the hydraulic actuator is accurately matched with the pressure demand. This solves the execution lag problem caused by the disconnect between flow output and pressure demand in the existing technology. Its design logic directly supports the technical feature of the hydraulic dynamic control module receiving flow command to drive the oil cylinder mentioned above.

[0068] The first part of the formula is The core of this part is calculating the basic flow rate required to compensate for the pressure gap. Its technical logic is based on the pressure, flow rate, and volume relationship of the hydraulic system. (Molecular part) Reflecting the pressure gap in the hydraulic system: It is the final target pressure after correction for abrupt changes in formation. This is the real-time pressure of the hydraulic system. If the difference between the two is positive, it indicates that the current pressure is insufficient to meet the pile driving requirements, and the pressure needs to be supplemented by increasing the flow rate; if it is negative, the flow rate needs to be reduced to avoid excessive pressure. The hydraulic cylinder volume is introduced because the flow rate directly determines the extension and retraction speed of the cylinder. The larger the cylinder volume, the more oil volume is required to push the pile. Therefore, the pressure gap needs to be multiplied by the cylinder volume to obtain the increase in oil volume required to make up for the pressure gap, ensuring that the flow rate calculation matches the actual needs of the cylinder.

[0069] denominator This is a correction to the timing and oil transfer losses. The control cycle is typically set to 0.1 seconds. This value is consistent with the data acquisition cycle of the environmental sensing module and the pile body monitoring module. The purpose is to synchronize the flow calculation with the data acquisition and pressure adjustment, avoiding a misalignment between flow output and pressure demand in time—for example, if... If the time is too long, such as 1 second, the flow rate adjustment will lag behind the pressure change, causing pressure fluctuations; if it is too short, such as 0.01 seconds, it will increase the system's computational burden. The function of hydraulic oil density is to convert mass flow rate into volumetric flow rate. While hydraulic systems actually control volumetric flow rate (unit: L / min), oil density affects its compressibility—higher density results in lower compressibility and more direct pressure transmission. Therefore, it is necessary to consider the specific characteristics of hydraulic oil density. The volumetric flow rate calculation was revised to ensure that the actual output oil quality meets the pressure replenishment requirements.

[0070] The (hydraulic pipeline pressure loss coefficient) is used to correct for flow loss caused by frictional and local losses along the pipeline. In indoor construction, hydraulic pipelines may have bends and numerous joints due to space constraints. These factors can cause some flow to fail to reach the cylinder during transmission due to pressure loss. The value typically ranges from 0.9 to 0.98; the more complex the piping, the higher the value. The smaller the value, the larger the flow needs to be calculated to compensate for the loss. For example, when When the value is 0.9, the calculated flow rate will be lower than that when there is no loss ( This increases by approximately 11%, ensuring that the actual flow reaching the cylinder meets the pressure requirements.

[0071] The second part of the formula is This section is a flow fine-tuning item for pile safety. Its design purpose is to protect the pile body through pile strain feedback while ensuring that the basic flow meets the pressure requirements, so as to avoid pile cracking due to excessive pressure. (Real-time strain of the pile body) directly reflects the stress state of the pile body. The greater the strain, the closer the stress on the pile body is to its material strength limit. At this time, the flow rate should be appropriately reduced to reduce the thrust of the hydraulic cylinder and avoid further increase in strain. The value of the (pile strain compensation coefficient) needs to be determined in conjunction with the tensile strength of the pile material, for example, for high-strength concrete piles. A value close to 0.03 indicates that slight strain changes only cause minor adjustments in flow rate, thus avoiding impact on pile driving efficiency; while low-strength piles... The value is increased to 0.06 to ensure rapid reduction of flow rate and protection of the pile body when strain exceeds the limit. For example, when... When increasing from 1000με to 1400με (close to the design threshold of 1500με), The item will make Reduced by approximately 24% (in terms of (Calculation), thereby reducing the cylinder thrust and controlling strain growth.

[0072] This solution uses a flow formula to match flow rate with actual pressure and response requirements, eliminating lag in the response of the actuator.

[0073] Traditional technical solutions have the following technical problems: existing solutions do not verify flow commands, nor do they clarify the relationship between flow and motor speed, which leads to abnormal flow output or unstable motor control, affecting the operation of the hydraulic system.

[0074] Based on this, after receiving the flow command, the hydraulic dynamic control module first verifies the validity of the command. The verification includes whether the flow command value is within the rated flow range of the hydraulic pump. If it exceeds the range, an alarm signal is output and the upper limit of the rated flow is adopted. After the verification is passed, the hydraulic dynamic control module drives the hydraulic pump motor through the PWM signal and adjusts the motor speed to achieve real-time flow output. The motor speed and flow output have a linear relationship, and the relationship expression is motor speed = (flow command / hydraulic pump displacement) × 60, with the unit being r / min.

[0075] This technical solution focuses on flow command processing and motor drive control. Validity verification is a crucial preliminary step. The rated flow rate of the hydraulic pump represents the upper limit for safe operation of the equipment. If the flow command exceeds this range, it will lead to motor overload or excessive hydraulic system pressure. An alarm signal will be output to promptly alert the operator. Using the upper limit of the rated flow rate ensures safety while meeting construction needs as much as possible, avoiding construction interruptions due to abnormal commands. After successful verification, the motor is driven by a PWM signal. The PWM signal can precisely control the motor speed by adjusting the duty cycle; a larger duty cycle results in a higher motor speed and a larger flow output, achieving continuous and adjustable flow. The linear relationship between motor speed and flow rate is derived based on the working principle of a hydraulic pump. The hydraulic pump displacement is the volume of oil output per pump revolution (usually in mL / r). The flow rate command (L / min) is converted to mL / min and then divided by the displacement (mL / r) to obtain the motor speed per minute (r / min). Multiplying by 60 is to unify the unit to r / min and ensure accurate speed calculation. For example, if the flow rate command is 10L / min (i.e., 10000mL / min) and the pump displacement is 100mL / r, then the motor speed = (10000 / 100) = 100r / min, which conforms to the actual operating law.

[0076] This solution ensures safe and stable flow output and improves the reliability of the hydraulic system through command verification and linear speed control.

[0077] Traditional technical solutions have the following technical problems: existing solutions do not specify the specific adjustment methods when strain exceeds the threshold and the probability of formation mutation exceeds the preset value, resulting in closed-loop control being unable to be effectively executed and the system being difficult to adapt to changes in operating conditions.

[0078] Based on this, during the closed-loop adjustment phase, when the actual strain of the pile exceeds the design threshold, the pile strain compensation coefficient in the flow calculation formula is increased by 0.01, and this coefficient is maintained until the next comparison cycle after each adjustment; when the probability of a sudden change in the formation output by the AI ​​formation prediction module exceeds 80%, the AI ​​prediction weight coefficient in the target pressure correction coefficient calculation formula is increased by 0.05, and when the probability of a sudden change in the formation drops below 50%, the initial value of the AI ​​prediction weight coefficient is restored.

[0079] This technical solution details the specific operation of closed-loop adjustment. When the pile strain exceeds the threshold, the strain compensation coefficient is adjusted; increasing it by 0.01 will adjust the flow rate formula. Increasing some values ​​reduces flow output, alleviates pile load, and prevents further strain increase; maintaining the coefficient until the next comparison cycle (1 second) avoids frequent adjustments that could cause flow fluctuations, ensuring system stability. When the probability of a geological mutation exceeds 80%, increasing the AI ​​prediction weight coefficient by 0.05 can improve the correction coefficient formula. Increased weights and more aggressive adjustment of correction coefficients allow for a significant advance adjustment of target pressure to address impending geological upheavals. When the probability drops below 50%, it indicates a reduced risk of upheaval, and restoring the initial value prevents over-correction that could lead to pressure anomalies. The adjustment ranges (0.01, 0.05) and thresholds (80%, 50%) were determined based on extensive indoor pile driving experiments, ensuring effective adjustments while avoiding over-adjustment. For example, excessively large increases in coefficients could cause a sudden drop in flow, affecting pile driving efficiency, while insufficient increases would fail to mitigate risks in a timely manner.

[0080] This solution ensures effective closed-loop control by clearly defining adjustment methods and parameters, thereby improving the system's adaptability to changes in operating conditions.

[0081] Traditional technical solutions have the following technical problems: existing solutions lack a fault diagnosis mechanism, and module failures cannot be detected and handled in a timely manner, which can easily lead to loss of control of the hydraulic system and cause construction accidents.

[0082] Based on this, a fault diagnosis phase is also included. During the fault diagnosis phase, the central control unit monitors the working status parameters of each module in real time. These parameters include the acquisition frequency of the environmental perception module, the calculation time of the AI ​​stratum prediction module, the sensor signal strength of the pile body status monitoring module, and the output current of the hydraulic dynamic control module. When the working status parameters of a certain module exceed the preset normal range, the central control unit determines that the module is faulty, outputs the faulty module identifier and fault type, and switches to the backup module or starts the emergency control program. The emergency control program uses a preset fixed flow output value to drive the hydraulic system to ensure that the pile body is stably stationary.

[0083] This technical solution constructs a complete fault diagnosis and emergency handling system. The selection of operating parameters is targeted: the acquisition frequency reflects the operating rhythm of the environmental sensing module; abnormal frequencies may indicate sensor malfunctions. Calculation time reflects the processing capacity of the AI ​​module; excessive time may lead to prediction lag. Signal strength reflects the sensor connection status of the pile monitoring module; low strength may indicate a circuit fault. Output current reflects the load status of the hydraulic dynamic control module; abnormal current may indicate a motor or drive circuit fault. The central control unit monitors these parameters in real time, promptly identifying faults when they exceed normal ranges. Output labels and types help operators quickly locate problems. Backup module switching or emergency program activation are key safeguards. Backup modules ensure continued system operation, while emergency programs use a fixed flow rate to stabilize the pile, preventing uncontrolled hydraulic output from causing pile tilting or equipment damage during malfunctions. The fixed flow rate value is determined based on the pile weight and indoor environment to ensure the pile does not sink or shift during stabilization.

[0084] This solution enhances the system's fault tolerance through fault diagnosis and emergency handling, preventing construction accidents caused by faults.

[0085] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A hydraulic control method for large-scale indoor static pressure piles based on artificial intelligence, comprising the steps of collecting indoor environmental foundation data, pile parameters, and hydraulic system parameters and inputting them into the control unit, and training an LSTM network using historical pile driving data; characterized in that, It also includes a real-time control phase and a closed-loop adjustment phase. In the real-time control phase, the environmental sensing module collects indoor temperature difference, relative humidity, and wall lateral stress at preset intervals, the pile condition monitoring module collects pile strain, pile end resistance, and real-time hydraulic system pressure, and the AI ​​stratum prediction module outputs the probability of stratum mutation and the real-time change in pile end resistance. The central control unit first calculates the target pressure of the hydraulic system based on the collected data, then calculates the target pressure correction coefficient based on the output of the AI ​​stratum prediction module, and finally calculates the real-time output flow of the hydraulic pump based on the target pressure, correction coefficient, and real-time collected hydraulic system pressure, and sends the flow command to the hydraulic dynamic control module to drive the cylinder to press the pile. In the closed-loop adjustment phase, the actual strain of the pile is compared with the design threshold at preset intervals. When the threshold is exceeded, the flow calculation parameters are adjusted. When the probability of stratum mutation output by the AI ​​stratum prediction module exceeds the preset value, the correction coefficient calculation parameters are temporarily adjusted. The central control unit uses the following formula to calculate the target pressure of the hydraulic system: ； In the formula, The target pressure for the hydraulic system is expressed in MPa. This is the reference force required for the vertical driving of the pile, expressed in kN. The wall constraint influence coefficient is obtained by fitting the indoor wall stiffness and pile spacing. The lateral stress generated by the compression of the pile by the wall is expressed in MPa. The lateral surface area of ​​the pile is expressed in units of... ; This refers to the dynamic viscosity of the hydraulic oil, expressed in Pa·s. The effective working area of ​​the hydraulic cylinder, in units of ; The temperature compensation coefficient is determined by the hydraulic oil type. This represents the difference between the real-time temperature and the initial temperature of the hydraulic oil, in °C. The humidity influence coefficient is obtained from an indoor humidity sensor. This refers to indoor relative humidity, expressed as % . The central control unit uses the following formula to calculate the target pressure correction factor: ； In the formula, This is a correction factor for the target pressure, and has no unit. The weight coefficients are predicted by the AI ​​and determined by the training accuracy of the LSTM model. The output value of the LSTM network represents the probability of abrupt changes in the formation, and its value ranges from 0 to 1. This includes the past 5 sets of historical pile driving data, including driving depth, real-time pressure, and pile strain. This refers to the density of the soil layer, in units of... ; This refers to the pile driving speed, expressed in m / min. This is the sensitivity coefficient for sudden changes in pile end resistance; This represents the real-time change in pile end resistance, in kN. The central control unit uses the following formula to calculate the real-time output flow of the hydraulic pump: ； In the formula, This refers to the real-time output flow rate of the hydraulic pump, expressed in L / min. This refers to the real-time pressure of the hydraulic system, expressed in MPa. The volume of the hydraulic cylinder is expressed in liters (L). The control period is measured in seconds (s). The density of hydraulic oil is expressed in units of... ; This refers to the pressure loss coefficient of the hydraulic pipeline. This is the pile strain compensation coefficient; The value represents the real-time strain of the pile body, expressed in με.

2. The hydraulic control method for large-scale indoor static pressure piles based on artificial intelligence according to claim 1, characterized in that, The number of historical pile driving data collected shall not be less than 500 sets, and the prediction accuracy of ground change after the LSTM network training is not less than 92%; the acquisition cycle of the environmental perception module and the pile condition monitoring module shall be set to 0.1 seconds, the cycle of comparing the actual strain of the pile body with the design threshold in the closed-loop adjustment stage shall be set to 1 second, and the design threshold of pile body strain shall be set to 1500με.

3. The hydraulic control method for large-scale indoor static pressure piles based on artificial intelligence according to claim 1, characterized in that, The basic data of the indoor environment include wall stiffness and initial temperature and humidity; the pile parameters include pile material and side surface area; the hydraulic system parameters include the effective working area of ​​the hydraulic cylinder and the pressure loss coefficient of the hydraulic pipeline; after the parameters are entered into the central control unit, the parameters are normalized. The normalization process adopts the linear normalization method to map the parameter values ​​to the 0-1 range. The mapping formula is: parameter normalized value = (parameter actual value - parameter minimum value) / (parameter maximum value - parameter minimum value).

4. The hydraulic control method for large-scale indoor static pressure piles based on artificial intelligence according to claim 1, characterized in that, During the training of the AI ​​geological formation prediction module, the input historical pile driving data includes driving depth, real-time pressure, and pile strain. The training process uses cross-validation, dividing the historical data into training and validation sets in a 7:3 ratio. The training set is used for iterative optimization of model parameters, and the validation set is used to evaluate the model's generalization ability. When the prediction error on the validation set is less than the preset error threshold for 5 consecutive training rounds, the model training is stopped.

5. The hydraulic control method for large-scale indoor static pressure piles based on artificial intelligence according to claim 1, characterized in that, After receiving the flow command, the hydraulic dynamic control module first verifies the validity of the command. The verification includes whether the flow command value is within the rated flow range of the hydraulic pump. If it exceeds the range, an alarm signal is output and the upper limit of the rated flow is adopted. After the verification is successful, the hydraulic dynamic control module drives the hydraulic pump motor through the PWM signal and adjusts the motor speed to achieve real-time flow output. The motor speed and flow output have a linear relationship, and the relationship expression is motor speed = (flow command / hydraulic pump displacement) × 60, with the unit being r / min.

6. The hydraulic control method for large-scale indoor static pressure piles based on artificial intelligence according to claim 1, characterized in that, During the closed-loop adjustment phase, when the actual strain of the pile exceeds the design threshold, the pile strain compensation coefficient in the flow calculation formula is increased by 0.01, and this coefficient is maintained until the next comparison cycle after each adjustment. When the probability of a sudden change in the formation output by the AI ​​formation prediction module exceeds 80%, the AI ​​prediction weight coefficient in the target pressure correction coefficient calculation formula is increased by 0.

05. When the probability of a sudden change in the formation drops below 50%, the initial value of the AI ​​prediction weight coefficient is restored.

7. The hydraulic control method for large-scale indoor static pressure piles based on artificial intelligence according to claim 1, characterized in that, It also includes a fault diagnosis phase. During the fault diagnosis phase, the central control unit monitors the working status parameters of each module in real time. The working status parameters include the acquisition frequency of the environmental perception module, the calculation time of the AI ​​stratum prediction module, the sensor signal strength of the pile body status monitoring module, and the output current of the hydraulic dynamic control module. When the working status parameters of a certain module exceed the preset normal range, the central control unit determines that the module is faulty, outputs the faulty module identifier and fault type, and switches to the backup module or starts the emergency control program. The emergency control program uses a preset fixed flow output value to drive the hydraulic system to ensure that the pile body is stably stationary.

Images