Continuous intelligent refueling control method for fracturing unit

Abstract

The invention relates to a continuous intelligent refueling control method for a fracturing unit, and the method achieves the integration of the monitoring of the fuel oil liquid level of a diesel-driven fracturing unit and the control of a continuous refueling device through the construction of a data center, and achieves the prediction of the oil consumption of each device of the fracturing unit through the introduction of an LSTM oil consumption prediction model and a dynamic priority scheduling algorithm. Dynamic priority refueling scheduling ensures timely fuel oil supply of the whole unit, and fully ensures continuity in the construction process of the fracturing unit. The system effectively solves the problems that a current continuous refueling automatic control system only supplies fuel oil simply according to high and low liquid level domain values, cannot cope with the working condition that multiple large-scale fracturing devices need to be refueled at the same time, and needs manual control; according to the invention, the labor intensity of operators can be greatly reduced, and the continuity of the construction operation of the diesel-driven fracturing unit is improved.

Description

Technical Field

[0001] This invention relates to the field of diesel-driven fracturing units, and in particular to a continuous intelligent refueling control method for fracturing units. Background Technology

[0002] In oil and gas fracturing operations of diesel-driven fracturing units, a continuous and stable fuel supply is a core requirement for ensuring efficient equipment operation. However, while current centralized fuel supply systems achieve automation through PLC control, their fuel supply strategies are based on fixed threshold triggers and lack dynamic scheduling capabilities. When multiple fracturing trucks simultaneously run out of fuel, the system is prone to a sudden drop in fuel supply rate and pump overload due to concurrent refueling requests. This can easily lead to some diesel-driven fracturing trucks shutting down unexpectedly due to untimely refueling. Therefore, current continuous fuel delivery systems mainly rely on manual control based on experience, resulting in high labor intensity for operators and requiring high levels of personal skills. Consequently, current continuous refueling control systems cannot meet the continuous automatic refueling requirements of diesel-driven fracturing units during large-scale fracturing operations. Summary of the Invention

[0003] This invention provides a continuous intelligent refueling control method for fracturing units and a control method thereof. By constructing a data platform, it integrates fuel level monitoring and continuous refueling device control of diesel-driven fracturing units. It also introduces an LSTM fuel consumption prediction model and a dynamic priority scheduling algorithm to predict the fuel consumption of each piece of equipment in the fracturing unit and dynamically prioritize refueling scheduling, ensuring timely fuel supply for the entire unit and fully guaranteeing the continuity of fracturing unit operations.

[0004] The technical solution adopted in this invention is: a continuous intelligent oiling control method for fracturing units, characterized by the following steps: S1. Modify the diesel-driven fracturing unit so that each fracturing unit can independently collect its own fuel level signal and calculate the fuel consumption rate based on the fuel level and tank volume. S2. Construct a data platform to integrate signal acquisition from fracturing units and control of continuous refueling units; S3. Configure the pipelines corresponding to each fracturing unit in the data platform to determine the explosion-proof solenoid valves and flow meters corresponding to each fracturing unit, and configure the model and tank volume information of each fracturing unit. S4. The data platform writes the volume information into the corresponding fracturing unit, and the fracturing unit calculates the oil consumption rate itself. S5. Data acquisition and preprocessing: to obtain fracturing unit operation data, accumulate raw data, filter the raw data, and build training models and decision-making basis. S6. Deploy an LSTM oil consumption prediction model in the data platform to predict the oil consumption trend of each piece of equipment in the fracturing unit in the next 30 minutes, so as to provide a basis for dynamic scheduling. S7. Design a dynamic priority scheduling algorithm to enable intelligent allocation of refueling order based on the prediction results of each device, and limit the number of concurrent refueling devices based on the actual performance of the continuous refueling device. S8. Control command generation and execution, to transform scheduling decisions into control signals for the continuous refueling unit; S9. System verification and optimization to ensure system stability and prediction accuracy.

[0005] Preferably, in step S3, a configuration interface is built on the data platform to achieve a one-to-one correspondence between the fracturing unit and the oil pipeline, and to know the size of the oil tank of the corresponding fracturing unit. Then, when the fuel level of the fracturing unit changes, the system can identify and control the explosion-proof solenoid valve on the corresponding pipeline.

[0006] As a preferred embodiment, step S5 is implemented as follows: S51, Data Acquisition Collect real-time data on fracturing unit rotation speed, fuel level, real-time fuel consumption, fuel temperature, and operating mode; S52, Data Cleaning Outlier handling: For data where the rotational speed exceeds the rated value, a sliding window mean is used to fill the gaps; Normalization: The rotation speed and liquid level data are grouped and normalized according to the equipment model; S53, Feature Engineering Generate historical fuel consumption rates.

[0007] As a preferred embodiment, step S6 is implemented as follows: S61, Model Training Input data: 60-minute continuous speed sequence + ambient temperature + fuel consumption rate + fuel level + fuel tank volume + equipment model number, input in real time every second; Output data: Fuel availability time, dynamically updated every 30 seconds.

[0008] Network structure: Two-layer LSTM + fully connected layer, optimized using Huber Loss; S62, Model Lightweighting The model was converted to FP16 accuracy using TensorRT and then deployed to the field data platform. S63, Online Reasoning After receiving data in real time, a prediction is performed every 30 seconds, and the prediction result and 95% confidence interval are output.

[0009] As a preferred embodiment, step S7 is implemented as follows: S71, Priority Calculation in: To predict the remaining available time, This is the current equipment speed. This refers to the rated speed of the current equipment model; urgency Fuel remaining time (TTE) directly determines the risk of downtime and is the most critical indicator. Load intensity High-speed equipment consumes more fuel, so it is necessary to prioritize its continuous operation. S72, Concurrency Control If TTE > 60 minutes, and the current device is in refueling mode, then refueling should be stopped and the global counter N should be updated, where N is the number of devices currently being refueled. When N≥15, new requests enter the waiting queue; Preemptive scheduling is adopted: if a high-priority device requests refueling, the refueling task of the lowest-priority device is interrupted; S73, Time Slice Rotation For devices with similar TTE, they are started at intervals after being sorted by priority.

[0010] As a preferred embodiment, step S9 is implemented as follows: S91, Offline Testing Use historical data to validate model prediction errors and the effectiveness of concurrency control strategies; S92, On-site commissioning Deploy a prototype system at the oilfield development site to compare the number of downtimes and fuel utilization rates under manual and intelligent control. S93, Continuous Optimization By using a federated learning framework, data from various well sites is aggregated regularly to update the global model, improving cross-scenario adaptability.

[0011] The beneficial effects of this invention are as follows: This invention integrates fuel level monitoring and continuous refueling control of diesel-driven fracturing units by constructing a data platform. It also introduces an LSTM fuel consumption prediction model and a dynamic priority scheduling algorithm to predict the fuel consumption of each piece of equipment in the fracturing unit and dynamically prioritize refueling, ensuring timely fuel supply for the entire unit and fully guaranteeing the continuity of fracturing operations. This invention effectively solves the problem that current automatic continuous refueling control systems simply supply fuel based on high and low liquid level thresholds, which cannot handle the situation where many pieces of equipment need to be refueled simultaneously in large-scale fracturing operations, requiring manual control. This invention can greatly reduce the labor intensity of operators and improve the continuity of diesel-driven fracturing unit operations. Attached Figure Description

[0012] Figure 1 This is a control flowchart of the present invention; Figure 2 This is a schematic diagram of the control electrical system of the present invention; Figure 3 This is a schematic diagram illustrating the operation of the present invention. Detailed Implementation

[0013] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0014] like Figures 1-3 As shown, this invention provides a continuous intelligent refueling control method for fracturing units, characterized by the following steps: S1. Modify the diesel-driven fracturing unit so that each fracturing unit can independently collect its own fuel level signal and calculate the fuel consumption rate based on the fuel level and tank volume. S2. Construct a data platform to integrate signal acquisition from fracturing units and control of continuous refueling units; S3. Configure the pipelines corresponding to each fracturing unit in the data platform to determine the explosion-proof solenoid valves and flow meters corresponding to each fracturing unit, and configure the model and tank volume information of each fracturing unit. S4. The data platform writes the volume information into the corresponding fracturing unit, and the fracturing unit calculates the oil consumption rate itself. S5. Data acquisition and preprocessing: to obtain fracturing unit operation data, accumulate raw data, filter the raw data, and build training models and decision-making basis. S6. Deploy an LSTM oil consumption prediction model in the data platform to predict the oil consumption trend of each piece of equipment in the fracturing unit in the next 30 minutes, so as to provide a basis for dynamic scheduling. S7. Design a dynamic priority scheduling algorithm to enable intelligent allocation of refueling order based on the prediction results of each device, and limit the number of concurrent refueling devices based on the actual performance of the continuous refueling device. S8. Control command generation and execution, to transform scheduling decisions into control signals for the continuous refueling unit; S9. System verification and optimization to ensure system stability and prediction accuracy.

[0015] In this embodiment, in step S3, a configuration interface is built on the data platform to achieve a one-to-one correspondence between the fracturing device and the oil pipeline, and to know the size of the oil tank of the corresponding fracturing device. Then, when the fuel level of the fracturing device changes, the system can identify it and control the explosion-proof solenoid valve on the corresponding pipeline to achieve the control effect. In this embodiment, step S5 is implemented as follows: S51, Data Acquisition Collect real-time data such as fracturing unit rotation speed (RPM), fuel oil level (%), real-time fuel consumption (L / min), oil temperature (°C), and operating mode (fracturing / standby); S52, Data Cleaning Outlier handling: For data where the rotational speed exceeds the rated value, a sliding window mean is used to fill the gaps; Normalization: Normalize the rotation speed and liquid level data by equipment model (Min-Max Scaling); S53, Feature Engineering Generate historical fuel consumption rate (60-minute average of sliding window).

[0016] As a preferred embodiment, step S6 is implemented as follows: S61, Model Training Input data: 60-minute continuous speed sequence + ambient temperature + fuel consumption rate + fuel level + fuel tank volume + equipment model number (One-Hot), input every second in real time; Output data: Fuel availability time, i.e., the predicted fuel consumption value at 6 future time points (every 5 minutes), dynamically updated every 30 seconds.

[0017] Network structure: Two-layer LSTM (64 hidden neurons) + fully connected layer, optimized using Huber Loss; S62, Model Lightweighting The model was converted to FP16 accuracy using TensorRT and then deployed to the field data platform. S63, Online Reasoning After receiving data in real time, a prediction is performed every 30 seconds, and the prediction result and 95% confidence interval are output.

[0018] As a preferred embodiment, step S7 is implemented as follows: S71, Priority Calculation in: To predict the remaining available time, This is the current equipment speed. This refers to the rated speed of the current equipment model; urgency The remaining fuel time (TTE) directly determines the downtime risk and is the most critical indicator. For example, if the equipment has only enough fuel to last for 10 minutes, an immediate response is required, so it is given the highest weight. Load intensity High-speed equipment consumes more fuel and its continuous operation should be prioritized; for example, equipment with a speed of 80% or more of the rated value may be in a high-pressure operating state and requires accelerated fuel supply. Both 0.7 and 0.3 can be adjusted according to the actual application. S72, Concurrency Control If TTE > 60 minutes, and the current device is in refueling mode, then refueling should be stopped and the global counter N should be updated, where N is the number of devices currently being refueled. When N≥15, new requests enter the waiting queue; Preemptive scheduling is adopted: if a high-priority device (TTE < 30 minutes) requests refueling, the refueling task of the lowest-priority device is interrupted; S73, Time Slice Rotation For devices with similar TTE (e.g., difference < 3 minutes), start them at intervals according to priority (e.g., start one every 30 seconds).

[0019] As a preferred embodiment, step S9 is implemented as follows: S91, Offline Testing Use historical data to verify the model prediction error (MAE≤5L / 5 minutes) and the effectiveness of the concurrency control strategy (load fluctuation <10%). S92, On-site commissioning Deploy a prototype system at the oilfield development site to compare the number of downtimes and fuel utilization rates under manual and intelligent control. S93, Continuous Optimization By using a federated learning framework, data from various well sites is aggregated regularly to update the global model, improving cross-scenario adaptability.

[0020] It should be noted that the above description of the technical solutions is exemplary, and this specification may be embodied in different forms and should not be construed as limiting it to the technical solutions set forth herein. Rather, providing these descriptions will ensure that the disclosure of this invention is thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art. Furthermore, the technical solutions of this invention are defined only by the scope of the claims.

[0021] Finally, it should be noted that the above embodiments are merely representative examples of the present invention. Obviously, the present invention is not limited to the above embodiments and many variations are possible. Any simple modifications, equivalent changes, and alterations made to the above embodiments based on the technical essence of the present invention should be considered within the protection scope of the present invention.

Claims

1. A continuous intelligent oiling control method for a fracturing unit, characterized in that: Includes the following steps: S1. Modify the diesel-driven fracturing unit so that each fracturing unit can independently collect its own fuel level signal and calculate the fuel consumption rate based on the fuel level and tank volume. S2. Construct a data platform to integrate signal acquisition from fracturing units and control of continuous refueling units; S3. Configure the pipelines corresponding to each fracturing unit in the data platform to determine the explosion-proof solenoid valves and flow meters corresponding to each fracturing unit, and configure the model and tank volume information of each fracturing unit. S4. The data platform writes the volume information into the corresponding fracturing unit, and the fracturing unit calculates the oil consumption rate itself. S5. Data acquisition and preprocessing: to obtain fracturing unit operation data, accumulate raw data, filter the raw data, and build training models and decision-making basis. S6. Deploy an LSTM oil consumption prediction model in the data platform to predict the oil consumption trend of each piece of equipment in the fracturing unit in the next 30 minutes, so as to provide a basis for dynamic scheduling. S7. Design a dynamic priority scheduling algorithm to enable intelligent allocation of refueling order based on the prediction results of each device, and limit the number of concurrent refueling devices based on the actual performance of the continuous refueling device. S8. Control command generation and execution, to transform scheduling decisions into control signals for the continuous refueling unit; S9. System verification and optimization to ensure system stability and prediction accuracy.

2. The continuous intelligent oiling control method for fracturing units according to claim 1, characterized in that: In step S3, a configuration interface is built on the data platform to achieve a one-to-one correspondence between the fracturing unit and the oil pipeline, and to know the size of the oil tank of the corresponding fracturing unit. Then, when the fuel level of the fracturing unit changes, the system can identify and control the explosion-proof solenoid valve on the corresponding pipeline.

3. The continuous intelligent oiling control method for fracturing units according to claim 1, characterized in that: The specific implementation method of step S5 is as follows: S51, Data Acquisition Collect real-time data on fracturing unit rotation speed, fuel level, real-time fuel consumption, fuel temperature, and operating mode; S52, Data Cleaning Outlier handling: For data where the rotational speed exceeds the rated value, a sliding window mean is used to fill the gaps; Normalization: The rotation speed and liquid level data are grouped and normalized according to the equipment model; S53, Feature Engineering Generate historical fuel consumption rates.

4. The continuous intelligent oiling control method for fracturing units according to claim 1, characterized in that: The specific implementation method of step S6 is as follows: S61, Model Training Input data: 60-minute continuous speed sequence + ambient temperature + fuel consumption rate + fuel level + fuel tank volume + equipment model number, input in real time every second; Output data: Fuel availability time, dynamically updated every 30 seconds; Network structure: Two-layer LSTM + fully connected layer, optimized using Huber Loss; S62, Model Lightweighting The model was converted to FP16 accuracy using TensorRT and then deployed to the field data platform. S63, Online Reasoning After receiving data in real time, a prediction is performed every 30 seconds, and the prediction result and 95% confidence interval are output.

5. The continuous intelligent oiling control method for fracturing units according to claim 1, characterized in that: The specific implementation method of step S7 is as follows: S71, Priority Calculation in: To predict the remaining available time, This is the current equipment speed. This refers to the rated speed of the current equipment model; urgency Fuel remaining time (TTE) directly determines the risk of downtime and is the most critical indicator. Load intensity High-speed equipment consumes more fuel, so it is necessary to prioritize its continuous operation. S72, Concurrency Control If TTE > 60 minutes, and the current device is in refueling mode, then refueling should be stopped and the global counter N should be updated, where N is the number of devices currently being refueled. When N≥15, new requests enter the waiting queue; Preemptive scheduling is adopted: if a high-priority device requests refueling, the refueling task of the lowest-priority device is interrupted; S73, Time Slice Rotation For devices with similar TTE, they are started at intervals after being sorted by priority.

6. The continuous intelligent oiling control method for fracturing units according to claim 1, characterized in that: The specific implementation method of step S9 is as follows: S91, Offline Testing Use historical data to validate model prediction errors and the effectiveness of concurrency control strategies; S92, On-site commissioning Deploy a prototype system at the oilfield development site to compare the number of downtimes and fuel utilization rates under manual and intelligent control. S93, Continuous Optimization By using a federated learning framework, data from various well sites is aggregated regularly to update the global model, improving cross-scenario adaptability.

Images