Multi-pump cooperative pressure control method for mining remote liquid supply pipeline

Abstract

The invention relates to the technical field of liquid supply pipelines, in particular to a mining remote liquid supply pipeline multi-pump cooperative pressure control method which comprises the following steps: firstly, acquiring multi-pump operation data through intelligent sensing equipment, performing characteristic analysis in combination with target process parameters, and extracting liquid supply pressure characteristics; and then, performing control interaction on multiple pumps by utilizing the target parameters and the pressure characteristics, and analyzing real-time pressure demand information. And performing cooperative calculation based on the demand traversal pressure characteristics, and formulating a multi-pump cooperative control strategy. A simulation execution mechanism is innovatively introduced, real-time monitoring is conducted on the strategy before the strategy is issued, a simulation result is generated, the control strategy is dynamically optimized and updated according to the result, and finally intelligent cooperative pressure control over the multiple pumps is achieved. According to the method, the problems of uneven multi-pump pressure, response lag, high energy consumption and the like in traditional control are effectively solved, and the stability, the safety and the energy efficiency level of the system under complex working conditions are improved through simulation, monitoring and closed loop correction.

Description

A method for multi-pump coordinated pressure control in a remote liquid supply pipeline for mining Technical Field

[0001] This invention relates to the field of liquid supply pipeline technology, and in particular to a multi-pump coordinated pressure control method for remote liquid supply pipelines in mines. Background Technology

[0002] In modern coal mine production, hydraulic supports, as the core support equipment for fully mechanized mining faces, rely on an efficient and reliable emulsion supply system for stable operation. Mine-use long-distance emulsion supply pipeline systems typically employ multiple supply pumps operating in parallel to meet the demands of long-distance, high-flow-rate, and high-pressure supply. However, due to complex underground working conditions, frequent load fluctuations, and large variations in pipeline resistance, problems such as uneven pressure distribution, unbalanced flow distribution, and start-up / shutdown shocks can easily occur among the supply pumps. This leads to increased system energy consumption, accelerated equipment wear, and even affects the normal operation of hydraulic supports, threatening safe production.

[0003] Currently, traditional emulsion pump stations mostly employ single-pump constant pressure control or simple alternating start-stop control methods, lacking the ability to coordinate and regulate multiple pumps, making it difficult to achieve precise and stable pressure control and dynamic response. Although some systems have introduced variable frequency speed regulation or pressure feedback control, they are still limited to single-machine closed-loop regulation, failing to optimize and coordinate at the overall system level, and cannot effectively cope with dynamic pressure demand changes under complex operating conditions. In addition, existing control methods have insufficient capabilities in acquiring and analyzing operating data, and lack adaptive collaborative strategy generation mechanisms based on operating condition characteristics, resulting in system lag and low response accuracy.

[0004] The above content is only used to help understand the technical solution of the present invention and does not represent an admission that the above content is prior art. Summary of the Invention

[0005] The main objective of this invention is to provide a multi-pump collaborative pressure control method for remote liquid supply pipelines in mines. This method aims to solve the technical problems of existing emulsion supply systems in mines, which lack a multi-pump collaborative control mechanism and an adaptive strategy based on operating conditions. As a result, these systems are unable to achieve precise and stable pressure control and dynamic response under complex downhole conditions, which can easily lead to uneven pressure, increased energy consumption, and equipment wear.

[0006] To achieve the above objectives, this invention provides a multi-pump collaborative pressure control method for a remote emulsion supply pipeline in a mine. The method includes: acquiring data from multiple supply pumps on the remote emulsion supply pipeline using intelligent sensing devices to obtain multiple pump operation datasets; determining target supply pressure process parameters based on the multiple emulsion supply pumps; performing supply characteristic analysis based on the multiple pump operation datasets and the target supply pressure process parameters to obtain multiple supply pressure characteristics; controlling the multiple emulsion supply pumps interactively according to the target supply pressure process parameters and the multiple supply pressure characteristics; determining mine emulsion supply pressure demand information based on the interaction parameters; performing collaborative calculations based on the mine emulsion supply pressure demand information, traversing the multiple supply pressure characteristics, and formulating a multi-pump collaborative pressure control strategy; simulating the execution of the multi-pump collaborative pressure control strategy on the multiple emulsion supply pumps and monitoring in real time to generate simulated supply monitoring results; and dynamically updating the multi-pump collaborative pressure control strategy based on the simulated supply monitoring results to perform intelligent collaborative pressure control on the multiple emulsion supply pumps.

[0007] Optionally, the step of determining the target supply pressure process parameters based on multiple emulsion supply pumps, and performing supply characteristic analysis based on the multiple pump operation datasets and the target supply pressure process parameters to obtain multiple supply pressure features includes: extracting multiple initial supply pressure parameters based on multiple emulsion supply pumps; performing multi-objective optimization on multiple emulsion supply pumps by traversing the multiple initial supply pressure parameters and setting supply pressure constraints; filtering the multiple initial supply pressure parameters according to the supply pressure constraints to determine the target supply pressure process parameters; matching the multiple pump operation datasets with the target supply pressure process parameters according to a time series to obtain multiple supply pressure datasets, wherein the multiple supply pressure datasets include pipeline inlet pressure datasets and pump outlet flow datasets; and performing supply characteristic calculations based on the pipeline inlet pressure datasets and the pump outlet flow datasets to obtain multiple supply pressure features.

[0008] Optionally, the step of controlling and interacting with multiple emulsion supply pumps according to the target supply pressure process parameters and the multiple supply pressure characteristics, and determining the mining emulsion supply pressure demand information based on the interaction parameters, includes: mapping the target supply pressure process parameters to multiple emulsion supply pumps according to the pipeline inlet pressure dataset to determine a first pressure mapping coefficient; mapping the target supply pressure process parameters to multiple emulsion supply pumps according to the pump outlet flow dataset to determine a second flow mapping coefficient; associating the first pressure mapping coefficient and the second flow mapping coefficient with the multiple emulsion supply pumps to generate associated interaction parameters; sending the associated interaction parameters to the multiple emulsion supply pumps for supply pressure matching calculation to generate multiple pressure matching coefficients; and using the multiple pressure matching coefficients as indexes to retrieve the target supply pressure process parameters to determine the mining emulsion supply pressure demand information.

[0009] Optionally, the step of performing collaborative calculations based on the mining emulsion supply pressure demand information to traverse the multiple supply pressure features and formulate a multi-pump collaborative pressure control strategy includes: performing data dependency analysis based on the multiple supply pressure features to construct association rules; mining the mining emulsion supply pressure demand information according to the association rules to obtain data mining results; performing fuzzy evaluation on the multiple supply pressure features based on the data mining results to generate multiple feature fuzzy scores; classifying the multiple supply pressure features according to the multiple feature fuzzy scores to determine a priority sequence; performing collaborative learning on the multiple supply pressure features according to the priority sequence to generate collaborative learning results; and performing collaborative verification based on the collaborative learning results to formulate the multi-pump collaborative pressure control strategy.

[0010] Optionally, the step of collaboratively learning the multiple liquid supply pressure features according to the priority sequence to generate collaborative learning results includes: assigning weights to the multiple liquid supply pressure features based on the priority sequence to generate multiple weight coefficients; integrating the multiple liquid supply pressure features according to the multiple weight coefficients to construct multiple liquid supply control spaces, the multiple liquid supply control spaces including pump action adjustment space and liquid supply pressure state space; introducing a reward function to perform reinforcement learning on the pump action adjustment space and the liquid supply pressure state space to generate reinforcement learning results; and performing continuous collaborative interactive feedback on multiple emulsion supply pumps based on the reinforcement learning results to generate the collaborative learning results.

[0011] Optionally, the step of simulating the execution of the multi-pump collaborative pressure control strategy on multiple emulsion supply pumps and monitoring it in real time to generate simulated supply monitoring results includes: retrieving historical downhole supply environment information of multiple emulsion supply pumps and constructing a simulation execution environment based on the historical downhole supply environment information; loading the multi-pump collaborative pressure control strategy onto the multiple emulsion supply pumps based on the simulation execution environment to simulate the execution of multiple emulsion supply pumps and generate multiple simulated supply parameters; combining the multiple simulated supply parameters according to the target supply pressure process parameters to determine multiple simulated supply groups; setting a desired supply pressure threshold, iterating through the multiple simulated supply groups and calculating the deviation from the desired supply pressure threshold to generate simulated supply deviation values; and adding the simulated supply deviation values ​​to the simulated supply monitoring results.

[0012] Optionally, adding the simulated liquid supply deviation value to the simulated liquid supply monitoring result includes: performing a liquid supply deviation analysis based on the desired liquid supply pressure threshold, setting a preset deviation distance, and determining whether the simulated liquid supply deviation value is greater than or equal to the preset deviation distance; if the simulated liquid supply deviation value is less than the preset deviation distance, generating a positive feedback parameter, activating a liquid supply monitoring command through the positive feedback parameter, continuously simulating and monitoring multiple emulsion liquid supply pumps through the liquid supply monitoring command, and generating the simulated liquid supply monitoring result; if the simulated liquid supply deviation value is greater than or equal to the preset deviation distance, generating a negative feedback parameter, activating a liquid supply anomaly command through the negative feedback parameter, tracing the anomaly through the liquid supply anomaly command, determining multiple simulated liquid supply anomaly points, and adding the multiple simulated liquid supply anomaly points to the simulated liquid supply monitoring result.

[0013] Optionally, after determining multiple simulated abnormal points in the fluid supply, the process includes: extracting the historical pressure loss coefficient of the corresponding pipeline section based on the multiple simulated abnormal points in the fluid supply, correcting the pressure loss coefficient by combining the real-time downhole ambient temperature and the current emulsion viscosity parameter, and obtaining a target corrected loss value; pre-compensating the original target fluid supply pressure process parameters based on the target corrected loss value, and synchronously updating the compensated target fluid supply pressure process parameters to the multi-pump collaborative pressure control strategy to achieve pre-adjustment of the collaborative control strategy.

[0014] Furthermore, to achieve the above objectives, the present invention also provides a multi-pump coordinated pressure control device for a remote liquid supply pipeline in a mine. The device includes: a memory, a processor, and a multi-pump coordinated pressure control program for a remote liquid supply pipeline in a mine stored in the memory and executable on the processor. The multi-pump coordinated pressure control program for a remote liquid supply pipeline in a mine is configured to implement the steps of the multi-pump coordinated pressure control method for a remote liquid supply pipeline in a mine as described above.

[0015] In addition, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing a multi-pump coordinated pressure control program for a remote liquid supply pipeline in a mine. When the multi-pump coordinated pressure control program for a remote liquid supply pipeline in a mine is executed by a processor, it implements the steps of the multi-pump coordinated pressure control method for a remote liquid supply pipeline in a mine as described above.

[0016] This invention provides a multi-pump collaborative pressure control method for remote liquid supply pipelines in mines. The method collects operational data from multiple liquid supply pumps and performs feature analysis based on target process parameters, enabling a comprehensive understanding of the pumps' operational status at the system level. Based on this, a collaborative strategy dynamically allocates the load of each pump, effectively avoiding common issues such as pressure fluctuations, uneven flow distribution, and mutual interference during parallel operation of multiple pumps, ensuring stable pressure output at the end of the remote liquid supply pipeline. For the complex operating conditions of coal mines, characterized by large load variations and frequent pipeline resistance fluctuations, this method introduces a control interaction and demand information determination mechanism. The system no longer passively responds to pressure deviations but actively analyzes the current liquid supply pressure demand based on real-time interactive parameters. Combining fuzzy evaluation and collaborative calculation, the control strategy can be adjusted in real-time according to changes in operating conditions, significantly improving the system's dynamic response speed and adjustment accuracy to sudden high loads or pipeline leaks. This method innovatively incorporates a simulation execution and real-time monitoring stage before strategy execution. Before actually issuing control commands, the effectiveness of the collaborative strategy is verified in a constructed simulation environment, and monitoring results are generated. If the simulation results show deviations from the expected threshold, the system automatically triggers anomaly tracing and dynamically updates the strategy. This simulation-before-execution mechanism significantly reduces the risk of equipment damage or pipeline rupture caused by erroneous control commands, thus improving the inherent safety level of the system. By constructing association rules, performing data mining, and utilizing reinforcement learning, including the collaborative learning and weight allocation implicit in the claims, the system can identify the optimal pump group combination and operating parameters. This not only avoids unnecessary frequent start-ups and shutdowns and ineffective work, reducing overall system energy consumption, but also balances the working time and wear of each pump, thereby extending the service life of the emulsion pump station and pipeline system and reducing operation and maintenance costs. Attached Figure Description

[0017] Figure 1 is a flowchart illustrating one embodiment of the multi-pump coordinated pressure control method for remote liquid supply pipelines in mines according to the present invention; Figure 2 is a flowchart illustrating another embodiment of the multi-pump coordinated pressure control method for remote liquid supply pipelines in mines according to the present invention.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] Referring to Figure 1, which is a flowchart illustrating an embodiment of the multi-pump coordinated pressure control method for remote liquid supply pipelines in mines according to the present invention, an embodiment of the multi-pump coordinated pressure control method for remote liquid supply pipelines in mines according to the present invention is presented.

[0021] In one embodiment, the multi-pump coordinated pressure control method for the remote emulsion supply pipeline in the mine includes: step S100, collecting data from multiple supply pumps on the remote emulsion supply pipeline in the mine using intelligent sensing devices to obtain multiple pump operation datasets.

[0022] The intelligent sensing device can be a collection of sensing devices deployed in the liquid supply system to collect real-time operating status parameters of the pump group and pipeline. It can provide the system with multi-dimensional, highly timely raw pump operation data to support subsequent feature analysis and collaborative decision-making. In this embodiment, the intelligent sensing device can continuously monitor the liquid supply pump and its connecting pipelines through physical quantity sensors such as pressure, flow rate, vibration, and temperature, and convert analog signals into digital data streams. For example, the intelligent sensing device can include, but is not limited to, one or more of pressure transmitters, electromagnetic flowmeters, and vibration acceleration sensors. The mine-use remote emulsion supply pipeline can be a long-distance, high-pressure transmission pipeline system connecting the emulsion pump station and the hydraulic support of the fully mechanized mining face. It can be used to achieve stable delivery of emulsion from the pump station to the working face, carrying the system's pressure and flow transmission functions. Multiple liquid supply pumps can be multiple hydraulic power devices arranged in parallel within the pump station, jointly outputting emulsion to the same liquid supply pipeline. They can be used to provide a power source for emulsion that meets the high pressure and high flow rate requirements of the fully mechanized mining face. Furthermore, the multiple liquid supply pumps may include, but are not limited to, a main pump, a standby pump, and a speed-regulating pump. The multiple pump operation dataset can be a collection of multi-dimensional time-series data generated by each liquid supply pump during operation, collected by intelligent sensing devices. This dataset can be used as the basic input for liquid supply characteristic analysis, reflecting the current operating conditions and performance status of each pump. In a specific embodiment, the multiple pump operation dataset can be obtained by synchronously recording parameters such as pressure, flow rate, speed, current, and start / stop status of each pump according to timestamps.

[0023] Data is collected from multiple pumps on a remote emulsion supply pipeline in a mine using intelligent sensing devices. This can be achieved by sensors deployed on the pump body and key nodes of the pipeline continuously collecting physical quantities and converting them into digital signals, which are then uploaded to the control unit. Furthermore, this operation can be accomplished through high-reliability data transmission using wired industrial buses (such as CAN or Profibus) or by using wireless mesh networks to achieve sensor self-organization and data aggregation, thereby enabling comprehensive and synchronous perception of the operating status of multiple pumps. The resulting datasets of multiple pump operations can be obtained by structuring and storing the raw sensor data according to pump number and time sequence, thus forming standardized data inputs that can be used for analysis.

[0024] Step S200: Determine the target supply pressure process parameters based on multiple emulsion supply pumps, and perform supply characteristic analysis based on multiple pump operation datasets and the target supply pressure process parameters to obtain multiple supply pressure characteristics.

[0025] The target fluid supply pressure process parameter can be the desired end-point fluid supply pressure value or pressure range set according to the hydraulic support operation requirements and pipeline design specifications. It can serve as a target benchmark for system pressure control, guiding the formulation of collaborative strategies. In an exemplary embodiment, the target fluid supply pressure process parameter can be preset through the fully mechanized mining process procedure or dynamically obtained from the upper-level control system. Fluid supply characteristic analysis can be a process of correlating pump operating data with the target pressure parameter to extract the fluid supply capacity characteristics of the pump group. It can be used to reveal the response capability, efficiency boundary, and stability performance of each pump under current operating conditions. Multiple fluid supply pressure characteristics can be structured feature vectors representing the fluid supply capacity and state of each pump extracted after fluid supply characteristic analysis. They can be used to quantify attributes such as the load potential, response delay, and pressure fluctuation sensitivity of each pump for collaborative calculation. For example, multiple fluid supply pressure characteristics can include, but are not limited to, dynamic response characteristics, steady-state output characteristics, and disturbance suppression characteristics.

[0026] Determining the target supply pressure process parameters based on multiple emulsion supply pumps can be achieved by calculating the required end pressure based on the hydraulic support action sequence of the fully mechanized mining face, pipeline length, and resistance model. Furthermore, this operation can be implemented by dynamically adjusting the target pressure according to the coal mining machine position via a central controller, or by pre-setting multi-level pressure process parameter templates based on historical operating condition clustering results, thereby establishing a pressure control target that matches the actual operating conditions. Analyzing the supply characteristics based on multiple pump operation datasets combined with the target supply pressure process parameters can be done by performing time-domain / frequency-domain correlation modeling between the operating data and the target pressure to extract the supply capacity characteristics of each pump. In a specific embodiment, this operation can be achieved by using principal component analysis (PCA) for dimensionality reduction and clustering to identify pump performance patterns, or by constructing an LSTM neural network to predict the response curves of each pump under the target pressure, thereby quantifying the effective supply capacity and stability boundary of each pump under the current operating conditions. Obtaining multiple supply pressure characteristics can be achieved by outputting structured feature vectors generated by the supply characteristic analysis module, thus providing standardized input for collaborative computing.

[0027] Step S300: Control multiple emulsion supply pumps according to the target supply pressure process parameters and multiple supply pressure characteristics, and determine the supply pressure demand information of mining emulsion based on the interaction parameters.

[0028] In this context, control interaction can refer to the system's behavior of applying tentative or regulatory commands to multiple pumps based on target parameters and supply characteristics to obtain feedback responses. This can be used to actively detect the system's actual pressure demand under current operating conditions, rather than relying solely on end-point deviation feedback. Furthermore, control interaction can coordinate with multiple supply pumps: observing overall system pressure changes by fine-tuning the output of a specific pump; and coordinate with interaction parameters: generating input data for demand analysis. Interaction parameters can be the system's response data to regulatory commands collected during control interaction, which can be used as a basis for analyzing the current actual supply pressure demand, reflecting implicit operating condition information such as pipeline resistance and load fluctuations. The supply pressure demand information for mining emulsions can be the actual supply pressure level required by the system at the current moment, obtained after analyzing the interaction parameters. This can be used to replace traditional passive deviation signals as the driving input for feedforward collaborative control.

[0029] Controlling multiple emulsion supply pumps by combining target supply pressure process parameters with multiple supply pressure characteristics can be achieved by sending small adjustment commands (such as ±5% speed changes) to some pumps and observing the overall system response. For example, this operation can be implemented by using a step disturbance method to test the system pressure recovery time or by using pseudo-random signal excitation to identify pipeline impedance characteristics, thereby actively stimulating system dynamics to obtain real-world operating information. Determining the mining emulsion supply pressure demand information based on the interaction parameters can be achieved by analyzing the rate of change and steady-state values ​​of pressure and flow during the interaction process to infer the current actual demand. In a specific embodiment, this operation can be achieved by fitting the interaction response curve based on a system identification model to estimate the required pressure, or by using a fuzzy inference engine to map the interaction parameters to pressure demand levels, thereby enabling the identification of the system's required pressure from a feedforward perspective, rather than relying on hysteresis.

[0030] Step S400: Based on the information on the supply pressure demand of mining emulsion, perform collaborative calculations on multiple supply pressure characteristics to formulate a multi-pump collaborative pressure control strategy.

[0031] The collaborative calculation can be an algorithmic process that iterates through the supply pressure characteristics of each pump based on the supply pressure demand information to solve for the optimal load allocation scheme. This can be used to generate a combination of control commands that meet multiple objective constraints such as pressure stability, minimum energy consumption, and wear equalization. The multi-pump collaborative pressure control strategy can be a comprehensive control scheme that includes execution parameters such as the start / stop status of each pump, speed setting, and load ratio. This can be used to guide the collaborative operation of multiple pumps to achieve stable pressure output and optimized resource allocation. Furthermore, the multi-pump collaborative pressure control strategy can include, but is not limited to, load balancing strategies, rapid response strategies, and energy-saving priority strategies. Collaborative calculation based on the supply pressure demand information of mining emulsion, iterating through multiple supply pressure characteristics, can solve a multi-objective optimization problem by iterating through feasible pump combinations and parameter configurations while meeting the demand pressure. In an exemplary embodiment, this operation can be achieved by using a genetic algorithm to search for the optimal pump group combination and operating parameters, or by constructing a mixed integer linear programming (MILP) model to solve for the minimum energy consumption configuration, thereby generating a candidate set of control strategies that considers stability, energy consumption, and lifespan. Developing a multi-pump collaborative pressure control strategy can involve selecting the optimal solution that satisfies the constraints from the collaborative calculation results as the final strategy, thereby outputting a set of executable pump control commands.

[0032] Step S500: Simulate the execution of a multi-pump collaborative pressure control strategy for multiple emulsion supply pumps and monitor it in real time to generate simulated supply monitoring results. Based on the simulated supply monitoring results, dynamically update the multi-pump collaborative pressure control strategy to perform intelligent collaborative pressure control on multiple emulsion supply pumps.

[0033] The simulation execution process involves digitally extrapolating a proposed multi-pump coordinated pressure control strategy in a virtual environment. This can be used to predict system behavior after strategy execution, avoiding the direct issuance of high-risk commands. Real-time monitoring involves continuously observing and recording data on the virtual system's state during simulation execution. This generates simulated fluid supply monitoring results to evaluate the strategy's effectiveness. These results can be key indicators such as virtual system pressure, flow rate, and pump status output during simulation execution, serving as a basis for determining whether the control strategy meets safety and performance thresholds. The dynamic update mechanism is a feedback process that automatically corrects the control strategy based on the deviation between the simulated fluid supply monitoring results and the expected thresholds, ensuring the effectiveness and safety of the final executed strategy.

[0034] Simulating the execution of a multi-pump collaborative pressure control strategy for multiple emulsion supply pumps can be achieved by loading the control strategy into a digital twin model and simulating the system's dynamic response. For example, this operation can be implemented by building a high-fidelity hydraulic system model based on AMESim or MATLAB / Simulink, or by using a simplified state-space model for rapid online simulation, thereby pre-verifying the safety and effectiveness of the strategy. Real-time monitoring and generating simulated supply monitoring results can be performed by recording key variables (such as end pressure, pump outlet pressure, and flow balance) during the simulation, providing a quantitative basis for strategy evaluation. The multi-pump collaborative pressure control strategy can be dynamically updated based on the simulated supply monitoring results. This can be achieved by triggering strategy recalculation or parameter fine-tuning if the monitoring results exceed a preset threshold. In one embodiment, this operation can be achieved by using a sliding window mechanism to perform gradient correction of strategy parameters, or by reverting to the previous effective strategy and initiating an anomaly tracing process, thereby ensuring the safety and reliability of the final executed strategy. Intelligent collaborative pressure control of multiple emulsion supply pumps can be achieved by distributing the verified control strategy to each pump execution unit for implementation, thereby achieving stable supply under multi-pump collaboration.

[0035] Taking a sudden high-load operation at a fully mechanized mining face as an example, the multi-pump coordinated pressure control method for remote hydraulic fluid supply pipelines in this embodiment can be as follows: when the coal mining machine cuts hard rock, causing a surge in the instantaneous hydraulic fluid demand of the hydraulic support, the system detects a rapid drop in end pressure through intelligent sensing devices; the control interaction module then applies a small speed-up command to the two main pumps, and the interaction parameters show that the pressure recovery is slow, thus analyzing that the current hydraulic fluid supply pressure demand is significantly higher than the original set value; the collaborative calculation module traverses the hydraulic fluid supply pressure characteristics of each pump and generates a coordinated strategy to activate a third standby pump and increase the speed of the main pump; this strategy is simulated and executed in a digital twin model, and the monitoring results show that the end pressure can recover to the target range within 2 seconds without the risk of overpressure; after the strategy is confirmed to be effective, it is issued for execution, and the three pumps operate smoothly in coordination to meet the high-load demand. At the same time, the successful response pattern is recorded through reinforcement learning for rapid response to similar future operating conditions.

[0036] In one embodiment, a target supply pressure process parameter is determined based on multiple emulsion supply pumps. Supply characteristics are analyzed based on multiple pump operation datasets combined with the target supply pressure process parameter to obtain multiple supply pressure features. This includes: extracting multiple initial supply pressure parameters from multiple emulsion supply pumps; performing multi-objective optimization on multiple emulsion supply pumps by traversing multiple initial supply pressure parameters and setting supply pressure constraints; filtering multiple initial supply pressure parameters according to the supply pressure constraints to determine the target supply pressure process parameter; matching the multiple pump operation datasets with the target supply pressure process parameter based on a time series to obtain multiple supply pressure datasets, which include pipeline inlet pressure datasets and pump outlet flow datasets; and calculating supply characteristics based on the pipeline inlet pressure dataset and the pump outlet flow dataset to obtain multiple supply pressure features.

[0037] The initial supply pressure parameter can be a set of candidate supply pressure setpoints initially extracted from the operating capacity or historical operating conditions of multiple emulsion supply pumps. This set can be used as input variables for multi-objective optimization to generate target supply pressure process parameters that meet system constraints. In an exemplary embodiment, the initial supply pressure parameter can be combined with the pump's rated parameters, historical operating records, or pipeline design specifications to generate a set of candidate pressure setpoints. Further, the initial supply pressure parameter can be one or more of the following: pump rated pressure parameters, historical peak pressure parameters, pipeline pressure resistance limits, etc. Multi-objective optimization can be a mathematical programming process to solve for the optimal supply pressure parameter while satisfying multiple mutually constraining objectives (such as pressure stability, minimum energy consumption, and wear equalization). It can be used to coordinate the conflicts between system performance, safety, and efficiency, and to select the comprehensively optimal pressure setting scheme. In a specific embodiment, multi-objective optimization can use each initial parameter as input to evaluate the overall performance of the pump group under the condition of satisfying a multi-objective function. For example, multi-objective optimization can employ Pareto front analysis to preserve the non-dominated solution set, or construct a weighted sum objective function to transform the multi-objective problem into a single-objective optimization problem, thereby screening out candidate pressure solutions that balance system efficiency and safety. The fluid supply pressure constraint can be a boundary condition of the feasible region of the fluid supply pressure set based on equipment capacity, pipeline strength, and process requirements. This can be used to ensure that the selected target fluid supply pressure process parameters are within a physically achievable and safe operating range. In this embodiment, the fluid supply pressure constraint can be set with upper and lower pressure limits based on equipment safety specifications, pipeline material strength, and hydraulic support operation requirements. Furthermore, the fluid supply pressure constraint can be an upper limit for pipeline burst pressure, a minimum operating pressure of the hydraulic support, or a maximum continuous output pressure of the pump unit, etc.

[0038] Multiple initial supply pressure parameters are extracted from multiple emulsion supply pumps. This can be achieved by generating a set of candidate pressure setpoints based on the pump's rated parameters, historical operating records, or pipeline design specifications. Furthermore, this operation can be implemented by extracting the rated operating pressure from the pump nameplate parameters as the initial value, or by back-calculating the required inlet pressure based on the measured terminal pressure under historical high-load conditions, thus providing an initial boundary for the feasible solution space for multi-objective optimization. Multi-objective optimization of multiple emulsion supply pumps is then performed by iterating through multiple initial supply pressure parameters. This can be achieved by using each initial parameter as input to evaluate the overall performance of the pump group under multiple objective functions (such as energy consumption, stability, and lifespan). Further, this operation can be achieved by using Pareto front analysis to preserve the non-dominated solution set, or by constructing a weighted sum objective function to transform the multi-objective problem into a single-objective optimization problem, thereby screening out candidate pressure solutions that balance system efficiency and safety. Supply pressure constraints are set, which can be based on equipment safety specifications, pipeline material strength, and hydraulic support operation requirements, setting upper and lower pressure limits. Furthermore, this operation can be achieved by setting an upper limit threshold based on the pipeline test pressure specified by API or national standards, or by setting the operating range in conjunction with the continuous operating pressure range recommended by the pump manufacturer, thereby eliminating infeasible or dangerous pressure setting schemes.

[0039] Screening multiple initial supply pressure parameters according to the supply pressure constraints can eliminate pressure parameters exceeding the constraint boundaries, retaining only the valid candidate set, thus ensuring the engineering feasibility of the target supply pressure process parameters. Determining the target supply pressure process parameters can involve selecting the optimal solution from the screened candidate set as the target pressure setting under the current operating conditions, thereby establishing a system-level, dynamically adaptable pressure control benchmark. The time series can be a synchronized timestamp structure of pump operation data and target pressure parameters arranged in chronological order, providing a temporal alignment basis for supply matching and ensuring data correlation within the same operating window. Supply matching can be the process of aligning and associating multiple pump operation datasets with the target supply pressure process parameters by timestamp, constructing a physically meaningful and temporally consistent supply pressure dataset to support subsequent characteristic calculations. In a specific embodiment, supply matching can align pump operation data with the corresponding target pressure values ​​using a unified timestamp, forming associated data pairs. For example, liquid supply matching can be achieved by using a sliding time window to segment and match the data, or by using interpolation to time-align asynchronous sampled data, thereby ensuring that subsequent characteristic calculations are based on the real system state within the same operating window.

[0040] The supply pressure dataset can be a structured time-series data set containing key variables such as pipeline inlet pressure and pump outlet flow rate, formed after supply matching. It can be used as a direct input for supply characteristic calculations, reflecting the dynamic behavior of the pump-pipeline coupled system. Furthermore, the supply pressure dataset can be one or more of the following: steady-state supply dataset, transient response dataset, abnormal operating condition dataset, etc. The pipeline inlet pressure dataset can be a data sequence recording the pressure change over time at the inlet of a remote supply pipeline. It can be used to characterize the overall pressure level at the pump set output and to evaluate the inter-pump synergy. The pump outlet flow rate dataset can be a time-series data set of measured flow rates at the outlet of each supply pump. It can be used to reflect the actual work capacity and load distribution status of each pump and to identify flow imbalance phenomena.

[0041] Matching multiple pump operation datasets with target supply pressure process parameters based on time series can be achieved by aligning pump operation data with corresponding target pressure values ​​using a unified timestamp, forming correlated data pairs. Furthermore, this operation can be implemented by segmenting the data using a sliding time window or by time-aligning asynchronously sampled data using interpolation, ensuring that subsequent characteristic calculations are based on the actual system state within the same operating window. Obtaining multiple supply pressure datasets can result in a structured dataset output after supply matching, containing fields such as pipeline inlet pressure and pump outlet flow rate, thus providing an analytical foundation with both physical correlation and temporal consistency.

[0042] Liquid supply characteristic calculation can be based on the joint analysis of pipeline inlet pressure and pump outlet flow rate, quantifying the liquid supply capacity of each pump and the system response characteristics. This process can generate distinctive liquid supply pressure characteristics to support precise coordinated control. In one embodiment, liquid supply characteristic calculation can calculate indicators such as pressure-flow relationship curves, response delay, and fluctuation sensitivity. Furthermore, liquid supply characteristic calculation can use system identification methods to fit the pump-pipeline transfer function, or calculate the local pressure-flow slope as a dynamic efficiency feature through a sliding window, thereby accurately characterizing the actual liquid supply capacity of each pump in a dynamically coupled system. Liquid supply characteristic calculation based on pipeline inlet pressure datasets combined with pump outlet flow rate datasets can calculate indicators such as pressure-flow relationship curves, response delay, and fluctuation sensitivity. Further, this operation can be achieved by using system identification methods to fit the pump-pipeline transfer function, or by calculating the local pressure-flow slope as a dynamic efficiency feature through a sliding window, thereby accurately characterizing the output capacity and response characteristics of each pump under dynamic operating conditions. Obtaining multiple liquid supply pressure characteristics can be achieved by outputting the feature vectors of each pump generated by the liquid supply characteristic calculation module, thereby providing a high-dimensional, differentiated, and distinguishable state representation for collaborative computing.

[0043] Taking the optimization of pressure settings in long-distance liquid supply pipelines as an example, the multi-pump collaborative pressure control method for remote liquid supply pipelines in mines in this embodiment can be implemented in a 3000-meter remote liquid supply system. Initial liquid supply pressure parameters include 25 MPa (pump rated value), 28 MPa (historical peak value), and 31.5 MPa (upper limit of pipeline pressure resistance). The system performs multi-objective optimization, aiming to minimize total energy consumption, maximize pressure stability, and balance pump wear. Liquid supply pressure constraints are set to be no less than 24 MPa (lower limit of support operation) and no more than 30 MPa (safety margin). Parameters within the 25–29 MPa range are retained after screening. After optimization, 27 MPa is selected as the target liquid supply pressure process parameter. Subsequently, the outlet flow rate of each pump and the pipeline inlet pressure data over the past hour are matched with second-level timestamps to form a liquid supply pressure dataset. Calculations show that the standard deviation of flow rate fluctuation for pump 1 at 27 MPa is 1.2 L / min, while that for pump 2 is 2.8 L / min, indicating a more stable response. This difference is encoded as a supply pressure characteristic, which is used in subsequent collaborative strategies to prioritize the allocation of the main load to pump 1.

[0044] In one embodiment, referring to Figure 2, multiple emulsion supply pumps are controlled and interacted according to the target supply pressure process parameters combined with multiple supply pressure characteristics. The mining emulsion supply pressure demand information is determined based on the interaction parameters, including: Step S301: Mapping the target supply pressure process parameters to multiple emulsion supply pumps according to the pipeline inlet pressure dataset to determine a first pressure mapping coefficient. The first pressure mapping coefficient can be a normalized weighted coefficient characterizing the contribution of each pump to the system inlet pressure after associating the target supply pressure process parameters with the pipeline inlet pressure dataset. It can be used to quantify the relative ability of each pump to maintain the pipeline inlet pressure target under the current operating conditions and to construct multi-dimensional interaction parameters. In an exemplary embodiment, the first pressure mapping coefficient can be obtained by regressing, interpolating, or matching the target pressure value with historical or real-time pipeline inlet pressure data, and assigning the contribution weight of each pump to the inlet pressure, in conjunction with the context. For example, the first pressure mapping coefficient can include, but is not limited to, one or more of the following: static pressure mapping coefficient, dynamic response mapping coefficient, and disturbance suppression mapping coefficient.

[0045] Step S302: Based on the target supply pressure process parameters, map the pump outlet flow rate dataset to multiple emulsion supply pumps to determine the second flow rate mapping coefficient. The second flow rate mapping coefficient can be a normalized coefficient characterizing the actual flow output capacity of each pump under the target pressure after associating the target supply pressure process parameters with the pump outlet flow rate dataset. It can be used to reflect the effective work capacity of each pump under the set pressure, compensating for the control blind spot caused by relying solely on pressure information. Furthermore, the second flow rate mapping coefficient can be obtained by analyzing the outlet flow rate data of each pump under the target pressure constraint, evaluating its effective flow output capacity at that pressure, and then normalizing it. In a specific embodiment, the second flow rate mapping coefficient may include, but is not limited to, the rated flow rate mapping coefficient, the instantaneous flow rate mapping coefficient, and the efficiency correction mapping coefficient.

[0046] Step S303: Associate the first pressure mapping coefficient and the second flow mapping coefficient with multiple emulsion supply pumps to generate associated interaction parameters. These associated interaction parameters can be multi-dimensional coupled parameter vectors generated by fusing the first pressure mapping coefficient and the second flow mapping coefficient on an individual pump basis. They can be used to comprehensively characterize the system role and collaborative potential of each pump in the pressure-flow dual-dimensional framework, serving as input for supply pressure matching calculation. In this embodiment, the associated interaction parameters can collaborate with multiple emulsion supply pumps, with each pump corresponding to a set of associated interaction parameters; they can also collaborate with supply pressure matching calculation, serving as its driving input. Further, associating the first pressure mapping coefficient and the second flow mapping coefficient with multiple emulsion supply pumps to generate associated interaction parameters can be achieved by combining the corresponding pressure mapping coefficient and flow mapping coefficient of each pump into a two-dimensional or high-dimensional vector. For example, this operation can be achieved by vector concatenation to form interaction parameters in the form of [pressure coefficient, flow coefficient], or by weighted fusion to generate a single comprehensive coefficient with weights dynamically adjusted by the operating condition type. This allows for the formation of a multi-dimensional pump state representation that integrates pressure and flow information, supporting refined collaboration.

[0047] Step S304: Send the associated interaction parameters to multiple emulsion supply pumps for supply pressure matching calculation to generate multiple pressure matching coefficients. The supply pressure matching calculation can be a process where each supply pump calculates its degree of compatibility with the target pressure under the current operating conditions based on the received associated interaction parameters. This can be used to generate individualized pressure matching coefficients, reflecting the rationality and efficiency of the pump performing the current supply task. The multiple pressure matching coefficients can be numerical indicators output by each pump after the supply pressure matching calculation, characterizing its compatibility with the target supply pressure process parameters. They can be used as indexes to retrieve the current actual supply pressure requirement from the target parameter space, achieving requirement inversion. In a specific embodiment, the multiple pressure matching coefficients may include, but are not limited to, high-fit matching coefficients, critical matching coefficients, and inefficient matching coefficients. Furthermore, sending the associated interaction parameters to multiple emulsion supply pumps for supply pressure matching calculation to generate multiple pressure matching coefficients can be a process where each pump receives its own associated interaction parameters and combines them with its local status, such as speed and current, to calculate its degree of compatibility with the target pressure under the current configuration. For example, this operation can output a matching score between 0 and 1 through the pump controller's built-in matching function, or predict whether the target pressure can be stably maintained under the current parameters through a local simulation model and output a Boolean or continuous matching coefficient. This enables distributed and individualized fit assessment, providing a basis for demand inversion.

[0048] Step S305: Use multiple pressure matching coefficients as indexes to search for the target liquid supply pressure process parameters and determine the liquid supply pressure requirement information for mining emulsion.

[0049] Furthermore, multiple pressure matching coefficients can be used as indexes to retrieve the target supply pressure process parameters to determine the supply pressure demand information for mining emulsions. This can be achieved by using the pressure matching coefficients as query keys to retrieve the most likely actual demand pressure from a pre-built target pressure-operating condition mapping table or model. In an exemplary embodiment, this operation can be implemented by constructing a hash index table to map typical combinations of matching coefficients to corresponding demand pressure values, or by training a neural network model with matching coefficients as input and continuous demand pressure estimates as output. This allows for the inference of actual demand from the system response, completing feedforward demand analysis.

[0050] Taking a sudden increase in resistance due to partial blockage in the pipeline as an example, the multi-pump coordinated pressure control method for the remote liquid supply pipeline in this embodiment can be as follows: When a partial blockage occurs in the middle section of the remote liquid supply pipeline, the system's target liquid supply pressure process parameter remains 27 MPa. The control module calculates the first pressure mapping coefficients for pump 1 and pump 2 based on the pipeline inlet pressure dataset, which are 0.6 and 0.4, respectively; simultaneously, based on the pump outlet flow dataset, due to the aging and flow attenuation of pump 2, its second flow mapping coefficient is only 0.3, while that of pump 1 is 0.7. After association, the associated interaction parameters for the two pumps are generated as [0.6, 0.7] and [0.4, 0.3], respectively. After being distributed to each pump, the pressure matching coefficient calculated for pump 1 is 0.92, while that for pump 2 is only 0.45. The system uses this coefficient combination as an index to retrieve the actual required pressure corresponding to a similar matching pattern in the historical operating condition database, which is 29.5 MPa (due to increased resistance, a higher inlet pressure is required to maintain 27 MPa at the end). Based on this, update the information on the required supply pressure of the mining emulsion and initiate a collaborative strategy to increase the total output pressure, so as to prevent the support from becoming unstable due to insufficient pressure.

[0051] In one embodiment, a multi-pump collaborative pressure control strategy is formulated by traversing multiple supply pressure features based on the supply pressure demand information of mining emulsion, including: performing data dependency analysis based on multiple supply pressure features to construct association rules; mining the supply pressure demand information of mining emulsion according to the association rules to obtain data mining results; performing fuzzy evaluation on multiple supply pressure features based on the data mining results to generate multiple feature fuzzy scores; classifying multiple supply pressure features according to the multiple feature fuzzy scores to determine a priority sequence; performing collaborative learning on multiple supply pressure features based on the priority sequence to generate collaborative learning results; and performing collaborative verification based on the collaborative learning results to formulate a multi-pump collaborative pressure control strategy.

[0052] Data dependency analysis can be a process of identifying and modeling statistical or logical correlations among multiple fluid supply pressure features. It can be used to reveal the coupling relationships between various pump characteristic variables, providing a foundation for constructing high-confidence association rules. In this embodiment, the operational principle of data dependency analysis can be explained in context, i.e., by calculating the correlation coefficients, mutual information, or Granger causality between features to identify strong dependencies. Furthermore, data dependency analysis can be a preliminary step in the process, supporting the reliability of association rules. For example, data dependency analysis can include, but is not limited to, one or more of time-series dependency analysis, causal dependency analysis, and covariance dependency analysis. Data dependency analysis based on multiple fluid supply pressure features can involve calculating the correlation coefficients, mutual information, or Granger causality between features to identify strong dependencies. Further, this operation can be achieved by using Bayesian networks to model conditional dependencies between features, or by using the sliding window mutual information method to detect dynamic dependency changes, thereby discovering implicit operating condition coupling patterns and improving the accuracy of subsequent rule construction.

[0053] Although association rules are not listed as separate objects, their construction depends on the results of data dependency analysis. In this context, they exist as an intermediate product, serving to formalize the identified strong dependencies into a matchable rule structure for subsequent data mining. Mining the supply pressure demand information of mining emulsions according to association rules can be achieved by treating the demand information as transaction items and matching it against the constructed association rules to extract matching patterns. Furthermore, this operation can be achieved by using the FP-Growth algorithm to efficiently match high-frequency rules, or by combining semantic similarity to expand the rule matching range, thereby quickly locating operational scenarios similar to the current demand from historical experience.

[0054] The data mining results can be structured knowledge output after pattern extraction of the supply pressure demand information of mining emulsion based on association rules. This structured knowledge is used to transform the original demand information into a set of contextual features that can be used for fuzzy evaluation. In an exemplary embodiment, the data mining results can be obtained by describing the acquisition method in conjunction with the context, i.e., outputting structured knowledge fragments after rule matching or pattern recognition. For example, the data mining results can include, but are not limited to, one or more of frequent pattern sets, abnormal operating condition patterns, and steady-state operating patterns. Obtaining the data mining results can be achieved by outputting structured knowledge fragments after rule matching or pattern recognition. Furthermore, this operation provides contextual basis for fuzzy evaluation. Fuzzy evaluation is performed by traversing multiple supply pressure features based on the data mining results. This can be achieved by using the data mining results as input and calling a preset fuzzy rule library to score each feature. Furthermore, this operation can be achieved by using a Mamdani-type fuzzy inference system for multi-input single-output scoring, or by constructing an adaptive membership function to dynamically adjust the scoring scale according to the operating conditions, thereby achieving robust evaluation of the applicability of pump features under uncertain operating conditions.

[0055] Feature fuzzy scores can be quantified membership values ​​given by fuzzy evaluation methods to assess the applicability or superiority of each fluid supply pressure feature under current operating conditions. They can be used to perform a comparable soft ranking of pump features in uncertain environments, avoiding misjudgments caused by hard thresholds. In one specific embodiment, the acquisition method of feature fuzzy scores can be described in context, i.e., formed by the membership degree values ​​output by the fuzzy inference system. For example, feature fuzzy scores can include, but are not limited to, one or more of pressure response fuzzy scores, energy efficiency fuzzy scores, and wear risk fuzzy scores. Generating multiple feature fuzzy scores can be achieved by outputting the fuzzy membership degree values ​​corresponding to each fluid supply pressure feature. Further, this operation forms a ranking-based soft evaluation index.

[0056] Assigning grades to multiple fluid supply pressure features based on fuzzy feature scores can be achieved by setting a threshold for the scoring interval, mapping continuous scores to discrete grade labels. Furthermore, this operation can be implemented by using equal-width binning to divide grade intervals, or by automatically determining grade boundaries based on cluster centers, thereby simplifying the complexity of subsequent priority decisions. The grade label can be a qualitative classification label assigned to the fluid supply pressure features based on the fuzzy feature scores, which can be used to map continuous scores to discrete grades, facilitating the generation of subsequent priority sequences. In this embodiment, the operating principle of the grade label can be explained in context, i.e., the conversion from quantification to qualitative analysis is completed through threshold mapping. For example, the grade label can include, but is not limited to, one or more of high-fitting grades, medium-fitting grades, and low-fitting grades.

[0057] Determining the priority sequence can be achieved by sorting by level and score within the same level to generate a feature execution order list. Furthermore, this operation provides guidance for collaborative learning in exploring priorities. The priority sequence can be an execution priority list formed by sorting multiple fluid supply pressure features by level identifier, which can be used to guide the directionality of resource allocation and strategy exploration during collaborative learning. In an exemplary embodiment, the priority sequence can be described in context as being obtained by sorting based on a combination of level and score. For example, the priority sequence can include, but is not limited to, one or more of the following: master priority sequence, backup activation sequence, energy-saving maintenance sequence, etc. Collaborative learning of multiple fluid supply pressure features based on the priority sequence can be performed by prioritizing the sampling of pump combinations corresponding to high-priority features during reinforcement learning or parameter optimization. Furthermore, this operation can be achieved by assigning higher initial Q values ​​to high-priority actions in Q-learning, or by increasing the selection probability of high-priority individuals in a genetic algorithm, thereby accelerating policy convergence and focusing on high-potential configuration spaces.

[0058] Generating collaborative learning results can be the output of optimized control parameters or policy functions obtained after priority-guided learning. Furthermore, this operation reflects the experience sharing and joint optimization results among pump groups. Although the collaborative learning results are not listed as independent objects, as the output of collaborative learning, they are used as the object of verification in subsequent collaborative verification, and their content includes the optimized pump combination control logic or parameter set. Collaborative verification can be a process of verifying the consistency and feasibility of collaborative learning results under the constraints of multi-pump joint operation, and can be used to ensure that the learned strategy meets the overall stability and safety requirements of the system. In a specific embodiment, collaborative verification can explain the operating principle in context, that is, testing the joint execution effect of the learning results in a digital twin model or multi-agent simulation environment. Furthermore, collaborative verification can be the final verification step in the process. For example, collaborative verification can include simulating the pressure superposition effect when multiple pumps are loaded simultaneously, or verifying the robustness of the strategy under abnormal operating conditions such as pipeline leakage, or one or more of these. Collaborative verification based on collaborative learning results can be conducted by testing the joint execution effect of the learning results in a digital twin model or multi-agent simulation environment. Furthermore, this operation can be performed by simulating the pressure superposition effect when multiple pumps are simultaneously loaded, or by verifying the robustness of the strategy under abnormal operating conditions such as pipeline leaks. This allows for the elimination of strategies that are individually optimal but conflict with the overall system, ensuring overall feasibility. Developing a multi-pump collaborative pressure control strategy can involve transforming the learning results verified through collaborative testing into an executable set of pump start / stop, speed, and loading commands. Ultimately, this operation outputs a collaborative scheme for practical control.

[0059] Taking the case of a sudden increase in local resistance in the pipeline as an example, the multi-pump collaborative pressure control method for the remote liquid supply pipeline in this embodiment can be as follows: When a section of the remote liquid supply pipeline experiences a sudden increase in local resistance due to sediment accumulation, the system collects data showing an expansion in the pressure difference at the outlets of each pump; the liquid supply pressure characteristics show that the flow rate of some pumps has significantly decreased. Data dependency analysis reveals that this phenomenon is highly correlated with the historical "pipeline blockage precursor" pattern, and the association rule triggers the corresponding data mining results; the fuzzy evaluation module scores the characteristics of each pump accordingly, identifies two high-wear pumps that have a sluggish response under the current resistance, and assigns them low fuzzy scores; the level label classifies them as "low fit level", and the priority sequence places them at the bottom; collaborative learning focuses on the combination of the remaining three pumps in good condition, and quickly converges to a high-response configuration of two main pumps and one backup pump in reinforcement learning; collaborative verification confirms that this combination can maintain stable terminal pressure without overpressure; the final strategy activates this configuration, avoiding the surge in energy consumption and pipeline impact caused by the forced loading of inefficient pumps.

[0060] In one embodiment, multiple liquid supply pressure features are collaboratively learned according to a priority sequence to generate collaborative learning results. This includes: assigning weights to multiple liquid supply pressure features based on the priority sequence to generate multiple weight coefficients; integrating multiple liquid supply pressure features according to the multiple weight coefficients to construct multiple liquid supply control spaces, which include a pump action adjustment space and a liquid supply pressure state space; introducing a reward function to perform reinforcement learning on the pump action adjustment space and the liquid supply pressure state space to generate reinforcement learning results; and continuously coordinating and interacting with multiple emulsion supply pumps based on the reinforcement learning results to generate collaborative learning results.

[0061] The weight allocation can be a process of assigning different importance coefficients to each fluid supply pressure feature based on a priority sequence. This can be used to highlight the regulatory role of high-priority pumps and suppress the interference of low-priority features on the strategy. In this embodiment, the weight allocation can be explained in conjunction with the context, i.e., mapping the priority sequence to a nonlinear weight function (such as exponential decay) to generate differentiated weight coefficients. For example, the weight allocation can use a softmax function to transform the priority sequence into a probability distribution of weights, or set a threshold truncation mechanism to assign non-zero weights only to the top N features. In an exemplary embodiment, the weight allocation can include, but is not limited to, one or more of static weight allocation, dynamic decay weight allocation, and context-aware weight allocation. The weight coefficient can be a numerical parameter characterizing the relative importance of each fluid supply pressure feature under the current operating conditions, and can be used as a weighting basis for feature integration and control space construction. Furthermore, the weight coefficient can be obtained by outputting numerical weight parameters that correspond one-to-one with each fluid supply pressure feature. In a specific embodiment, the weight coefficient can include, but is not limited to, one or more of stability weight, energy efficiency weight, and response speed weight.

[0062] The liquid supply control space can be a multi-dimensional decision-response joint space composed of the pump action adjustment space and the liquid supply pressure state space, which can be used to provide a structured exploration and optimization environment for reinforcement learning. In this embodiment, the liquid supply control space can be described in context, that is, the integrated features are mapped to the pump action adjustment space and the liquid supply pressure state space respectively to form a joint decision framework. For example, the liquid supply control space can model the action-state mapping through a state transition matrix, or construct the pump interaction topology space using a graph neural network. In an exemplary embodiment, the liquid supply control space can include, but is not limited to, one or more of the following: steady-state control subspace, transient response subspace, anomaly recovery subspace, etc. The pump action adjustment space can be a combination set of all liquid supply pumps' executable operations (such as start-stop, frequency conversion, load ratio), which can be used to define the range of control actions that the system can apply. In a specific embodiment, the pump action adjustment space can include, but is not limited to, one or more of the following: discrete action subspace, continuous speed regulation subspace, hybrid action subspace, etc. The liquid supply pressure state space can be a set of pressure, flow rate, and stability indicators presented by the system under different pump combinations and actions, which can be used to characterize the system response results caused by the control actions. In one exemplary embodiment, the supply pressure state space may include, but is not limited to, one or more of the following: end pressure state, pipeline differential pressure state, pump outlet fluctuation state, etc.

[0063] The reward function can be an evaluation function used to quantify the merits of an action-state pair in reinforcement learning, guiding the policy towards optimization towards stable stress, low energy consumption, and balanced wear. In this embodiment, the reward function can be explained in context, i.e., a composite reward expression is designed that includes terms such as stress bias, energy consumption, number of start-stop cycles, and wear accumulation. Furthermore, the reward function can be implemented by linearly weighting multiple objective terms or by introducing a piecewise function to penalize over-limit states. For example, the reward function can include, but is not limited to, one or more of the following: multi-objective weighted reward, piecewise threshold reward, and risk-penalized reward.

[0064] Weighting multiple supply pressure features based on a priority sequence can be achieved by mapping them to a non-linear weighting function (such as exponential decay) according to their priority, generating differentiated weight coefficients. Furthermore, this operation can be implemented by using a softmax function to transform the priority sequence into a probability distribution of weights, or by setting a threshold truncation mechanism to assign non-zero weights only to the top N features, thus allowing high-priority pumps to dominate strategy generation and improving control efficiency. Generating multiple weight coefficients can be achieved by outputting numerical weight parameters that correspond one-to-one with each supply pressure feature. Furthermore, this operation can be achieved by directly outputting the aforementioned weighting process, thus providing a weighting basis for subsequent feature integration. Integrating multiple supply pressure features based on multiple weight coefficients can be achieved by weighted fusion of feature vectors to form a compressed or enhanced joint representation. Furthermore, this operation can be achieved by using a weighted average to integrate the supply capacity features of each pump, or by constructing an attention mechanism to dynamically fuse features from multiple pumps, thereby highlighting the feature contributions of key pumps and reducing noise interference.

[0065] Constructing multiple liquid supply control spaces can be achieved by mapping the integrated features to the pump action adjustment space and the liquid supply pressure state space, respectively, forming a joint decision-making framework. Furthermore, this operation can be implemented by modeling the action-state mapping using a state transition matrix, or by constructing an inter-pump interaction topology space using a graph neural network, thus allowing a structured expression of the relationship between control inputs and system responses. Introducing a reward function can be achieved by designing a composite reward expression that includes pressure deviation, energy consumption, start-stop frequency, wear accumulation, etc. Furthermore, this operation can be implemented by linearly weighting and combining multiple objective terms, or by introducing a piecewise function to penalize out-of-limit states, thereby driving reinforcement learning to converge towards multi-objective optimization.

[0066] Reinforcement learning is applied to the pump action adjustment space and the supply pressure state space. This can be achieved by executing actions in the control space, observing state changes, calculating rewards, and updating the policy network. Furthermore, this operation can be implemented using a Deep Q-Network (DQN) to handle the discrete action space, or by employing a Proximal Policy Optimization (PPO) algorithm to handle continuous speed regulation actions. This establishes a mapping from actions to performance, enabling adaptive policy generation. The reinforcement learning results can be output as a trained and optimized policy network or action value function. Further, this operation can be achieved by extracting the optimal policy after the aforementioned reinforcement learning process converges, thus providing an intelligent decision-making model to guide the coordinated control of the pump group. Continuous coordinated interactive feedback based on the reinforcement learning results for multiple emulsion supply pumps can be implemented by applying the policy to the actual pump group, collecting new operating data, and feeding it back to the learning system for policy fine-tuning. Further, this operation can be achieved by using an experience replay buffer to store interactive data for batch updates, or by implementing an online fine-tuning mechanism to locally update policy parameters after each round of interaction. This forms an online learning closed loop, improving the policy's adaptability to real-world operating conditions. Generating collaborative learning results can involve integrating reinforcement learning outcomes with interactive feedback information to output collaborative knowledge that can ultimately be used for policy formulation. Furthermore, this operation can be achieved by fusing historical interaction experience with the current policy output, thereby providing data-driven decision-making support for multi-pump collaborative pressure control strategies.

[0067] Taking the emergency response to a sudden pipeline leak as an example, the multi-pump collaborative pressure control method for remote liquid supply pipelines in mines in this embodiment can be as follows: When a minor leak occurs in the remote pipeline, causing a slow drop in pressure, the system analyzes the need for a slight increase in pressure; the priority sequence shows that pumps B and D are listed as high priority due to their recent good maintenance; the weight allocation module assigns them weight coefficients of 0.6 and 0.35, respectively; the integrated features construct a pump action adjustment space including the frequency conversion actions of pumps B and D and the corresponding pressure recovery state space; the reward function emphasizes the balance between pressure recovery speed and energy consumption; reinforcement learning explores and discovers in this space that increasing the frequency of pump B by 8% while maintaining the original speed of pump D can stabilize the pressure the fastest and with the lowest energy consumption; after the strategy is issued and executed, the system collects new data and feeds it back to the learning module; the continuous collaborative interactive feedback mechanism records the successful response pattern and fine-tunes the strategy network to make the response to similar leak scenarios in the future more accurate and efficient.

[0068] In one embodiment, a multi-pump coordinated pressure control strategy is simulated and monitored in real time for multiple emulsion supply pumps, generating simulated supply monitoring results. This includes retrieving historical downhole supply environment information from multiple emulsion supply pumps and constructing a simulated execution environment based on this information. The historical downhole supply environment information can be a data set recording the operating conditions and external disturbances experienced by the mining emulsion supply system in its past operations. This data can provide a realistic basis for constructing a high-fidelity simulated execution environment, improving the representativeness and reliability of strategy verification. In this embodiment, the historical downhole supply environment information can be extracted from the system's historical database, including structured or unstructured data such as pipeline resistance changes, load abrupt events, leakage records, temperature fluctuations, and pump start-up and shutdown logs. Furthermore, the historical downhole supply environment information may include, but is not limited to, one or more of the following: historical pipeline resistance sequences, load disturbance event logs, and environmental temperature and humidity records. The simulation execution environment can be a virtual system model built based on historical downhole fluid supply environment information, used for digital simulation of the execution effect of multi-pump collaborative control strategies. It can be used to pre-simulate control strategy behavior under conditions close to real downhole operation, avoiding the risks associated with directly affecting physical equipment. In an exemplary embodiment, the simulation execution environment can use historical environmental information as boundary conditions or disturbance inputs, integrating hydraulic dynamics, fluid transport delay, and pump characteristic curves to construct a dynamic simulation model. For example, the simulation execution environment can employ steady-state operating condition simulation environments, abrupt load simulation environments, and pipeline leakage simulation environments.

[0069] Retrieving historical downhole fluid supply environment information from multiple emulsion supply pumps can be achieved by retrieving historical operating records similar to the current operating conditions from a central database or edge storage nodes. Furthermore, this operation can be implemented by sliding-matching typical operating condition data from the most recent 72 hours based on a time window, or by using operating condition clustering tags (such as "normal coal cutting," "initial support," and "pipeline maintenance") to selectively retrieve corresponding historical segments, thereby providing a realistic and diverse operating condition background for the simulation execution environment. Constructing the simulation execution environment based on historical downhole fluid supply environment information can be achieved by loading historical environmental information as input perturbations or initial conditions into a digital twin model. In a specific embodiment, this operation can be achieved by configuring a hydraulic network model containing time-varying resistance and random leakage in the simulation modeling tool, or by injecting historical pressure-flow time-series data into the simulation solver as an external stimulus using a real-time database interface, thereby improving the simulation environment's fidelity to the real-world downhole complexity.

[0070] Based on a simulated execution environment, a multi-pump collaborative pressure control strategy is applied to simulate the execution of multiple emulsion supply pumps, generating multiple simulated supply parameters. These simulated supply parameters can be simulated values ​​of key performance indicators of each pump and pipeline output after running the multi-pump collaborative pressure control strategy in the simulated execution environment. They can be used to characterize the expected system response of the strategy under specific virtual operating conditions, serving as the basis for subsequent combination and evaluation. For example, simulated supply parameters can include, but are not limited to, one or more of simulated terminal pressure, simulated pump outlet flow rate, and simulated pump load current. Simulating the execution of the multi-pump collaborative pressure control strategy on multiple emulsion supply pumps based on the simulated execution environment can involve applying the control strategy to be verified as an input command to the virtual pump model to drive the simulation. Furthermore, this operation can be achieved by using discrete event simulation to execute start / stop and speed adjustment commands step-by-step, or by simulating the dynamic coupling of the pump-pipeline using a continuous-time differential equation solver, thereby generating predictive outputs reflecting the behavior of the strategy under specific operating conditions. Multiple simulated liquid supply parameters can be generated, which can be time-series outputs of variables such as pressure, flow rate, and power at key nodes during the simulation process, thereby forming a structured simulation dataset that can be used for subsequent analysis.

[0071] Multiple simulated fluid supply parameters are combined according to the target fluid supply pressure process parameters to determine multiple simulated fluid supply groups. These simulated fluid supply groups can be evaluation units formed by structurally aggregating multiple simulated fluid supply parameters according to the logical relationship of the target fluid supply pressure process parameters. This facilitates grouping and comparing strategy outputs according to process target dimensions, supporting consistency verification across multiple scenarios. In a specific embodiment, the simulated fluid supply groups may include, but are not limited to, one or more of the following: high-pressure demand group, low-flow maintenance group, and rapid response transition group. Combining multiple simulated fluid supply parameters according to the target fluid supply pressure process parameters can be achieved by logically grouping the simulated parameters based on the target pressure range or action stage. Furthermore, this operation can be achieved by dividing parameter subsets according to the target pressure level (e.g., 30MPa, 35MPa) or by associating parameter fragments according to the hydraulic support action type (lifting column, pushing conveyor, lowering support), thereby enabling the organization of simulation results according to process semantics, facilitating targeted evaluation. Determining multiple simulated fluid supply groups can output a structured set of combined parameters, thus providing standardized input units for deviation calculations.

[0072] Set the desired liquid supply pressure threshold, iterate through multiple simulated liquid supply groups and calculate the deviation from the desired liquid supply pressure threshold to generate simulated liquid supply deviation values; add the simulated liquid supply deviation values ​​to the simulated liquid supply monitoring results.

[0073] The desired supply pressure threshold can be the upper or lower limit of the allowable deviation range of the simulated supply pressure, set according to process safety and equipment protection requirements. It can serve as a benchmark for judging whether the simulated supply pressure deviation is acceptable, supporting the determination of strategy effectiveness. In this embodiment, the desired supply pressure threshold can be predefined by system safety specifications or operating procedures and can be dynamically adjusted according to the target supply pressure process parameters. The simulated supply deviation value can be the quantitative difference between various pressure indicators in the simulated supply group and the desired supply pressure threshold. It can be used to constitute the core criterion for the simulated supply monitoring results, triggering strategy correction or decision-making. For example, the simulated supply deviation value can include, but is not limited to, one or more of the following: steady-state pressure overshoot, dynamic recovery time deviation, and inter-group pressure imbalance. Setting the desired supply pressure threshold can be done by setting the allowable pressure fluctuation range according to system safety margin and process accuracy requirements. Furthermore, this operation can be achieved by setting a fixed bandwidth threshold (e.g., ±1.5MPa) or a relative threshold (e.g., ±5%) based on the dynamic proportion of the target pressure, thereby establishing a quantitative standard for determining strategy effectiveness.

[0074] The deviation calculation involves iterating through multiple simulated fluid supply groups and comparing them to the desired fluid supply pressure threshold. This can be achieved by calculating the algebraic or absolute deviation of the pressure index in each simulated fluid supply group from the threshold boundary. In an exemplary embodiment, this operation can be implemented by calculating the difference between the maximum instantaneous overpressure value and the upper limit threshold, or the mean square error between the integral pressure-time curve and the ideal trajectory, thereby quantifying the gap between the strategy execution result and the desired target. Generating simulated fluid supply deviation values ​​can be achieved by outputting the numerical results obtained from the deviation calculation, thus forming an interpretable strategy risk indicator. Adding simulated fluid supply deviation values ​​to the simulated fluid supply monitoring results can be achieved by embedding the deviation values ​​as key fields into the data structure of the simulated fluid supply monitoring results, thereby improving the evaluation dimensions of the monitoring results and supporting subsequent dynamic update mechanisms.

[0075] Taking the strategy pre-verification under the condition of sudden pipeline leakage as an example, the multi-pump coordinated pressure control method for remote fluid supply pipelines in mines in this embodiment can be as follows: The system detects a recent slow leakage event caused by a loose pipeline joint, which is recorded in the historical downhole fluid supply environment information. When a newly formulated multi-pump coordinated pressure control strategy needs to be issued, the system retrieves the leakage condition segment and constructs a simulated execution environment that includes local flow loss. After the strategy is simulated and executed in this environment, simulated fluid supply parameters are generated, showing that the terminal pressure drops by 2.8 MPa within 30 seconds. These parameters are combined into a "leakage response group" according to the current target pressure of 35 MPa and compared with the expected threshold (34–36 MPa) to calculate the simulated fluid supply deviation value of -1.8 MPa. This deviation value is written into the simulated fluid supply monitoring results. Because it exceeds the allowable range (-1.0 MPa), the system automatically triggers strategy optimization, adds the logic of early intervention of the backup pump, and re-simulates until the deviation value reaches the target before allowing actual execution.

[0076] In one embodiment, adding the simulated liquid supply deviation value to the simulated liquid supply monitoring result includes: performing a liquid supply deviation analysis based on the desired liquid supply pressure threshold, setting a preset deviation distance, and determining whether the simulated liquid supply deviation value is greater than or equal to the preset deviation distance; if the simulated liquid supply deviation value is less than the preset deviation distance, generating a positive feedback parameter, activating a liquid supply monitoring command through the positive feedback parameter, continuously simulating and monitoring multiple emulsion liquid supply pumps through the liquid supply monitoring command, and generating a simulated liquid supply monitoring result; if the simulated liquid supply deviation value is greater than or equal to the preset deviation distance, generating a negative feedback parameter, activating a liquid supply anomaly command through the negative feedback parameter, tracing the anomaly through the liquid supply anomaly command, determining multiple simulated liquid supply anomaly points, and adding the multiple simulated liquid supply anomaly points to the simulated liquid supply monitoring result.

[0077] The fluid supply deviation analysis can be a logical judgment process that evaluates and classifies simulated fluid supply deviation values ​​based on the expected fluid supply pressure threshold. It can provide a decision-making basis for subsequent feedback mechanisms, distinguishing whether the strategy execution result is within an acceptable range. In this embodiment, the fluid supply deviation analysis completes a preliminary qualitative judgment of the strategy execution effect by logically comparing the simulated fluid supply deviation value with the allowable deviation range of the expected fluid supply pressure threshold. The preset deviation distance can be a pre-set tolerance threshold for determining whether the simulated fluid supply deviation value is acceptable. It can be used as a criterion boundary for triggering positive / negative feedback, realizing a graded response for strategy verification. For example, the preset deviation distance can be configured according to system safety margin, equipment pressure resistance, and process accuracy requirements, and can be a fixed value or dynamically adjusted with the target pressure. Furthermore, setting the preset deviation distance can be based on system safety specifications and historical operating experience to configure deviation tolerance values, thereby establishing a quantitative boundary for judging the effectiveness of the strategy. In an exemplary embodiment, the preset deviation distance can be set as a fixed deviation distance using static engineering experience values, or dynamically adjusted based on historical rewards from reinforcement learning. For example, the preset deviation distance may include, but is not limited to, one or more of the following: steady-state tolerance distance, transient overshoot tolerance, and inter-group imbalance tolerance.

[0078] Determining whether the simulated fluid supply deviation is greater than or equal to a preset deviation distance can be achieved by performing a numerical comparison operation and outputting a Boolean judgment result. Furthermore, this judgment operation can be implemented using a standard comparison logic unit, thereby determining whether to execute a positive or negative feedback path subsequently. If the simulated fluid supply deviation is less than the preset deviation distance, a positive feedback parameter is generated. The positive feedback parameter can be a confirmatory control signal generated when the simulated fluid supply deviation is less than the preset deviation distance, which can be used to activate the continuous simulation monitoring process and maintain the verification state of the current strategy. In a specific embodiment, the positive feedback parameter can be one or more of the following, including but not limited to a steady-state confirmation signal, a low-risk pass flag, and a strategy hold command. Generating the positive feedback parameter can be achieved by creating a confirmation signal indicating the validity of the strategy when the judgment result is true, thereby initiating the strategy hold and continuous monitoring process.

[0079] The liquid supply monitoring command is activated by positive feedback parameters. This command, triggered by positive feedback parameters, is a control command used to maintain continuous simulation of a multi-pump system. It ensures continuous tracking of system behavior within the effective range of the strategy, preventing state drift. Furthermore, the liquid supply monitoring command can coordinate with multiple emulsion supply pumps: driving them to maintain their current operating state in the simulation environment; and coordinating with simulated liquid supply monitoring results: continuously updating the monitoring data stream. Activating the liquid supply monitoring command via positive feedback parameters can use these parameters as an enable signal to trigger continuous simulation monitoring tasks, thereby maintaining the online verification state of the current strategy. Continuous simulation monitoring of multiple emulsion supply pumps is performed using the liquid supply monitoring command. This operation can extend the simulation time step in the simulation execution environment, continuously acquiring system states to verify the stability of the strategy in the time dimension. Further, this operation can be achieved by continuously adding new operating condition disturbances in a rolling window for robustness testing, or by extending the simulation period in the steady-state phase to observe long-term drift trends, enabling the system to confirm the continued effectiveness of the strategy under dynamic disturbances.

[0080] Generating simulated fluid supply monitoring results can integrate parameters and status data from continuous monitoring to form a structured output, thereby completing the final verification loop of the effective strategy. If the simulated fluid supply deviation is greater than or equal to a preset deviation distance, a negative feedback parameter is generated. The negative feedback parameter can be an abnormal warning signal generated when the simulated fluid supply deviation is greater than or equal to the preset deviation distance, which can be used to trigger an abnormal handling process and prevent the current strategy from being executed directly. In an exemplary embodiment, the negative feedback parameter can be one or more of the following, including but not limited to overpressure warning signals, flow imbalance alarms, and response lag indicators. Generating the negative feedback parameter can also be done by creating a warning signal indicating that the strategy is at risk when the judgment result is false, thereby blocking the direct execution of the strategy and entering the diagnostic process.

[0081] The fluid supply anomaly command is activated via negative feedback parameters. This command, activated by negative feedback parameters, is a control command used to initiate anomaly tracing analysis, guiding the system to pinpoint the specific cause of pressure deviation. Furthermore, the fluid supply anomaly command can coordinate with simulated anomaly points in the fluid supply: driving anomaly point identification algorithms; and with simulated fluid supply monitoring results: injecting diagnostic information. Activating the fluid supply anomaly command via negative feedback parameters can use these parameters as trigger conditions to initiate an anomaly tracing subroutine, thereby guiding the system into fault location mode. Anomaly tracing via the fluid supply anomaly command can analyze the state trajectory during simulation execution, identifying key variables or components causing pressure deviation, thus pinpointing the root cause of the problem rather than merely identifying the phenomenon. Further, this operation can be achieved by using sensitivity analysis to calculate the contribution of each pump output to the terminal pressure, or by using graph neural networks to reverse-track the anomaly propagation path in the pipeline-pump topology, enabling the system to accurately identify the cause of the anomaly.

[0082] Identifying multiple anomaly points in the simulated fluid supply can provide interpretable optimization input; these anomaly points can be specific locations or behavioral characteristics identified by the anomaly tracing module. These anomaly points can be specific faults or performance degradation locations identified during simulation that cause the fluid supply pressure to deviate from the expected threshold. They can provide precise optimization directions for strategy correction, avoiding blind adjustments. For example, simulated anomaly points can include, but are not limited to, one or more of the following: pump outlet response delay points, sudden changes in local pipeline resistance, and nodes of imbalance in multi-pump flow distribution. Adding multiple simulated anomaly points to the simulated fluid supply monitoring results allows the anomaly points to be embedded as diagnostic fields into the monitoring results' data structure, thereby enhancing the information density of the monitoring results and supporting subsequent dynamic strategy updates.

[0083] Taking pressure fluctuations caused by asynchronous responses of multiple pumps as an example, the multi-pump collaborative pressure control method for remote liquid supply pipelines in mines in this embodiment can be as follows: When simulating the execution of a certain collaborative strategy, the system calculates that the simulated liquid supply deviation value is 1.6 MPa, while the preset deviation distance is 1.2 MPa. Because the deviation value exceeds the tolerance, the system generates negative feedback parameters and activates the liquid supply anomaly command; the anomaly tracing module analyzes the pressure-time curves of each pump and finds that pump No. 3 has a response delay of 800 ms after the command is issued, resulting in insufficient initial pressure, followed by overshoot due to overcompensation; this response lag behavior is identified as the "pump outlet response delay point" and listed as the liquid supply simulation anomaly point; this point information is written into the simulated liquid supply monitoring results, triggering the strategy update mechanism, and the new strategy adds a preload command to pump No. 3. If the deviation value drops to 0.9 MPa (less than 1.2 MPa) in the next simulation, a positive feedback parameter is generated, the liquid supply monitoring command is activated, and the system continues to simulate the strategy for 30 seconds. Only after confirming that there is no drift is actual execution allowed.

[0084] In one embodiment, after identifying multiple simulated abnormal points in the fluid supply, the process includes: extracting the historical pressure loss coefficient of the corresponding pipeline segment based on the multiple simulated abnormal points in the fluid supply, correcting the pressure loss coefficient by combining the real-time ambient temperature downhole and the current emulsion viscosity parameter, and obtaining the target corrected loss value; pre-compensating the original target fluid supply pressure process parameters based on the target corrected loss value, and synchronously updating the compensated target fluid supply pressure process parameters to the multi-pump collaborative pressure control strategy to achieve the pre-adjustment of the collaborative control strategy.

[0085] The corresponding pipeline segment can be a section of the emulsion delivery pipeline directly associated with the simulated abnormal point in the fluid supply topology. It can be used as a spatial positioning unit for pressure loss analysis and correction, supporting accurate pressure loss modeling. In an exemplary embodiment, the corresponding pipeline segment can achieve automatic matching of abnormal points and pipeline segments through a GIS pipeline digital twin model, or by tracing upstream high-resistance pipeline segments based on the pressure gradient direction of the abnormal point. The historical pressure loss coefficient can be an empirical parameter recorded in the system database, characterizing the pressure loss characteristics of a specific pipeline segment under unit flow rate in past operations. It can be used to provide a benchmark reference for current pressure loss estimation. Furthermore, the historical pressure loss coefficient can be obtained through inversion of historical steady-state operating data or offline calibration, and is usually stored in the form of a pressure loss to flow rate square ratio. For example, the historical pressure loss coefficient can include, but is not limited to, one or more of the following: straight pipe section friction loss coefficient, elbow local resistance coefficient, valve throttling loss coefficient, etc. The real-time underground ambient temperature can be the measured ambient temperature of the mine operating area at the current moment, and can be used as a key input variable for dynamic correction of emulsion viscosity. In one specific embodiment, the real-time ambient temperature downhole can be collected in real time by temperature sensors deployed at pump stations or along pipelines.

[0086] The current emulsion viscosity parameter can be a fluid property index reflecting the current flow resistance characteristics of the emulsion, and can directly affect the accuracy of pipeline pressure loss calculation. Furthermore, the current emulsion viscosity parameter can be directly measured by an online viscometer, or calculated from tables or models based on the emulsion ratio, usage time, and real-time downhole ambient temperature. For example, the current emulsion viscosity parameter can include, but is not limited to, one or more of dynamic viscosity, kinematic viscosity, and apparent viscosity. The pressure loss coefficient can be a comprehensive parameter describing the resistance generated by the pipeline system to the emulsion flow, related to fluid viscosity, pipe diameter, roughness, and flow velocity, and can be used to quantify the expected pressure decay from the pump outlet to the end. The target correction loss value can be an updated pressure loss parameter obtained by dynamically correcting the historical pressure loss coefficient with the current operating conditions (temperature, viscosity), and can be used as the basis for pre-compensation calculations to improve the accuracy of target pressure setting. In a specific embodiment, the target correction loss value can include, but is not limited to, one or more of temperature-corrected pressure loss values, viscosity-compensated pressure loss values, and comprehensive dynamic pressure loss values.

[0087] The original target supply pressure process parameter can be the expected terminal pressure value set in the initial stage of strategy formulation, without considering the current pipeline pressure loss changes, and can be used as a baseline input for pre-compensation. The compensated target supply pressure process parameter can be a new pressure setting value obtained by superimposing the target correction loss value on the original target, used to offset the predicted pressure drop, and can be used to enable the multi-pump collaborative strategy output to more accurately meet the actual terminal pressure demand. For example, the compensated target supply pressure process parameter can adopt feedforward compensation pressure setting, dynamic increase of target pressure, viscosity-temperature adaptive target pressure, etc. Extracting the historical pressure loss coefficient of the corresponding pipeline segment based on multiple supply simulation anomaly points can be achieved by mapping the anomaly points to the pipeline topology map and querying the historical pressure loss parameters stored in the segment. Furthermore, this operation can be achieved by automatically matching anomaly points and pipeline segments through a GIS pipeline digital twin model, or by tracing upstream high-resistance pipeline segments in reverse based on the pressure gradient direction of the anomaly point, thereby establishing the correlation between the anomaly and the physical pressure loss model.

[0088] By combining real-time downhole ambient temperature and current emulsion viscosity parameters to correct the pressure loss coefficient, the viscosity change can be mapped to a pressure loss coefficient adjustment using fluid dynamics relationships. Furthermore, this operation can be achieved through interpolation correction using a three-dimensional temperature-viscosity-pressure loss mapping table, or by constructing a lightweight neural network model to predict pressure loss changes under viscosity-temperature coupling online, thus enabling the transformation from static historical parameters to dynamic operating condition adaptation. Obtaining the target corrected loss value can be achieved by outputting the pressure loss coefficient value after operating condition correction, thereby forming a quantitative pressure drop estimate that can be used for pre-compensation. Pre-compensation of the original target supply pressure process parameters based on the target corrected loss value can be achieved by adding the target corrected loss value to the original target supply pressure process parameters to generate a new pressure setpoint. Further, this operation can be achieved by directly compensating the total pressure loss using a linear superposition method, or by compensating in segments according to pipe sections and superimposing the total compensation at the end, thereby offsetting the predicted pressure drop in advance and improving the accuracy of end-point pressure control.

[0089] The compensated target supply pressure process parameters are synchronously updated to the multi-pump collaborative pressure control strategy. This can be achieved by using the new target pressure value as input to re-trigger the collaborative calculation module or by directly replacing the target field in the strategy. Furthermore, this operation can either update only the target pressure parameters in the strategy without replanning the pump combination, or recalculate the optimal pump group and operating parameters using the new target pressure as input, thereby achieving feedforward pre-adjustment of the control strategy. Pre-adjustment of the collaborative control strategy can be achieved by completing the strategy parameter update and preparing for execution, thus enabling the system to have compensation capabilities before actual disturbances occur.

[0090] Taking the increase in emulsion viscosity due to low winter temperatures leading to insufficient pressure supply as an example, the multi-pump coordinated pressure control method for remote mine fluid supply pipelines in this embodiment can be as follows: During simulation, the system identifies a persistently low terminal pressure and locates the abnormal fluid supply point in the main pipeline section 1200 meters from the pump station. The system extracts the historical pressure loss coefficient of this pipeline section as 0.8 MPa / (L / min). 2 Simultaneously, the real-time downhole ambient temperature was read as 5°C. Based on the emulsion's usage cycle, the current viscosity was estimated to be 40% higher than at room temperature. Therefore, the system corrected the pressure loss coefficient to 1.12 MPa / (L / min). 2 The target corrected loss value was obtained; the original target liquid supply pressure process parameter was 35MPa, which was increased to 36.2MPa after pre-compensation; this new target was synchronized to the multi-pump collaborative pressure control strategy, and the load of the three pumps was redistributed according to the new target. After actual execution, the terminal pressure stabilized at 35.1MPa, meeting the support operation requirements and avoiding the risk of insufficient initial support force due to increased viscosity.

[0091] Furthermore, to achieve the above objectives, the present invention also provides a multi-pump coordinated pressure control device for a remote liquid supply pipeline in a mine. The device includes: a memory, a processor, and a multi-pump coordinated pressure control program for a remote liquid supply pipeline in a mine stored in the memory and executable on the processor. The multi-pump coordinated pressure control program for a remote liquid supply pipeline in a mine is configured to implement the steps of the multi-pump coordinated pressure control method for a remote liquid supply pipeline in a mine as described above.

[0092] In addition, to achieve the above objectives, the present invention also provides a computer-readable storage medium storing a multi-pump coordinated pressure control program for a remote liquid supply pipeline in a mine. When the multi-pump coordinated pressure control program for a remote liquid supply pipeline in a mine is executed by a processor, it implements the steps of the multi-pump coordinated pressure control method for a remote liquid supply pipeline in a mine as described above.

[0093] Other embodiments or specific implementations of the multi-pump coordinated pressure control device for remote liquid supply pipelines in mines described in this invention can be referred to the above-mentioned method embodiments, and will not be repeated here.

[0094] The above are merely preferred embodiments of the present invention and do not limit the scope of the patent. Any equivalent structural or procedural transformations made based on the description and drawings of the present invention, or direct or indirect applications in other related technical fields, are similarly included within the scope of patent protection of the present invention.

Claims

1. A method for multi-pump coordinated pressure control in a remote liquid supply pipeline for mining, characterized in that, The method includes: collecting data from multiple supply pumps on a remote emulsion supply pipeline for mining using intelligent sensing devices to obtain multiple pump operation datasets; determining target supply pressure process parameters based on the multiple emulsion supply pumps; performing supply characteristic analysis based on the multiple pump operation datasets and the target supply pressure process parameters to obtain multiple supply pressure characteristics; controlling the multiple emulsion supply pumps interactively according to the target supply pressure process parameters and the multiple supply pressure characteristics; determining mining emulsion supply pressure demand information based on the interaction parameters; performing collaborative calculations by traversing the multiple supply pressure characteristics based on the mining emulsion supply pressure demand information to formulate a multi-pump collaborative pressure control strategy; simulating the execution of the multi-pump collaborative pressure control strategy on the multiple emulsion supply pumps and monitoring in real time to generate simulated supply monitoring results; and dynamically updating the multi-pump collaborative pressure control strategy based on the simulated supply monitoring results to perform intelligent collaborative pressure control on the multiple emulsion supply pumps.

2. The multi-pump coordinated pressure control method for remote liquid supply pipelines in mines as described in claim 1, characterized in that, The process of determining the target supply pressure process parameters based on multiple emulsion supply pumps, and analyzing supply characteristics based on the multiple pump operation datasets and the target supply pressure process parameters to obtain multiple supply pressure features, includes: extracting multiple initial supply pressure parameters based on multiple emulsion supply pumps; performing multi-objective optimization on multiple emulsion supply pumps by traversing the multiple initial supply pressure parameters and setting supply pressure constraints; filtering the multiple initial supply pressure parameters according to the supply pressure constraints to determine the target supply pressure process parameters; matching the multiple pump operation datasets with the target supply pressure process parameters according to a time series to obtain multiple supply pressure datasets, which include pipeline inlet pressure datasets and pump outlet flow datasets; and calculating supply characteristics based on the pipeline inlet pressure datasets and the pump outlet flow datasets to obtain multiple supply pressure features.

3. The multi-pump coordinated pressure control method for remote liquid supply pipelines in mines as described in claim 2, characterized in that, The step of controlling and interacting with multiple emulsion supply pumps according to the target supply pressure process parameters and the multiple supply pressure characteristics, and determining the mining emulsion supply pressure demand information based on the interaction parameters, includes: mapping the target supply pressure process parameters to multiple emulsion supply pumps according to the pipeline inlet pressure dataset to determine a first pressure mapping coefficient; mapping the target supply pressure process parameters to multiple emulsion supply pumps according to the pump outlet flow dataset to determine a second flow mapping coefficient; associating the first pressure mapping coefficient and the second flow mapping coefficient with the multiple emulsion supply pumps to generate association interaction parameters; sending the association interaction parameters to the multiple emulsion supply pumps for supply pressure matching calculation to generate multiple pressure matching coefficients; and using the multiple pressure matching coefficients as indexes to retrieve the target supply pressure process parameters to determine the mining emulsion supply pressure demand information.

4. The multi-pump coordinated pressure control method for remote liquid supply pipelines in mines as described in claim 1, characterized in that, The step of collaboratively calculating and formulating a multi-pump collaborative pressure control strategy by traversing multiple supply pressure features based on the mining emulsion supply pressure demand information includes: performing data dependency analysis based on the multiple supply pressure features to construct association rules; mining the mining emulsion supply pressure demand information according to the association rules to obtain data mining results; performing fuzzy evaluation on the multiple supply pressure features based on the data mining results to generate multiple feature fuzzy scores; classifying the multiple supply pressure features according to the multiple feature fuzzy scores to determine a priority sequence; performing collaborative learning on the multiple supply pressure features based on the priority sequence to generate collaborative learning results; and performing collaborative verification based on the collaborative learning results to formulate the multi-pump collaborative pressure control strategy.

5. The multi-pump coordinated pressure control method for remote liquid supply pipelines in mines as described in claim 4, characterized in that, The step of collaboratively learning the multiple liquid supply pressure features according to the priority sequence to generate collaborative learning results includes: assigning weights to the multiple liquid supply pressure features based on the priority sequence to generate multiple weight coefficients; integrating the multiple liquid supply pressure features according to the multiple weight coefficients to construct multiple liquid supply control spaces, the multiple liquid supply control spaces including pump action adjustment space and liquid supply pressure state space; introducing a reward function to perform reinforcement learning on the pump action adjustment space and the liquid supply pressure state space to generate reinforcement learning results; and performing continuous collaborative interactive feedback on multiple emulsion liquid supply pumps based on the reinforcement learning results to generate the collaborative learning results.

6. The multi-pump coordinated pressure control method for remote liquid supply pipelines in mines as described in claim 1, characterized in that, The step of simulating the execution of the multi-pump collaborative pressure control strategy on multiple emulsion supply pumps and generating simulated supply monitoring results includes: retrieving historical downhole supply environment information of multiple emulsion supply pumps and constructing a simulated execution environment based on the historical downhole supply environment information; loading the multi-pump collaborative pressure control strategy onto the multiple emulsion supply pumps based on the simulated execution environment to simulate the execution of multiple emulsion supply pumps and generate multiple simulated supply parameters; combining the multiple simulated supply parameters according to the target supply pressure process parameters to determine multiple simulated supply groups; setting a desired supply pressure threshold, iterating through the multiple simulated supply groups and calculating the deviation from the desired supply pressure threshold to generate simulated supply deviation values; and adding the simulated supply deviation values ​​to the simulated supply monitoring results.

7. The multi-pump coordinated pressure control method for remote liquid supply pipelines in mines as described in claim 6, characterized in that, Adding the simulated liquid supply deviation value to the simulated liquid supply monitoring result includes: performing a liquid supply deviation analysis based on the desired liquid supply pressure threshold, setting a preset deviation distance, and determining whether the simulated liquid supply deviation value is greater than or equal to the preset deviation distance; if the simulated liquid supply deviation value is less than the preset deviation distance, generating a positive feedback parameter, activating a liquid supply monitoring command through the positive feedback parameter, continuously simulating and monitoring multiple emulsion liquid supply pumps through the liquid supply monitoring command, and generating the simulated liquid supply monitoring result; if the simulated liquid supply deviation value is greater than or equal to the preset deviation distance, generating a negative feedback parameter, activating a liquid supply anomaly command through the negative feedback parameter, tracing the anomaly through the liquid supply anomaly command, determining multiple simulated liquid supply anomaly points, and adding the multiple simulated liquid supply anomaly points to the simulated liquid supply monitoring result.

8. The multi-pump coordinated pressure control method for remote liquid supply pipelines in mines as described in claim 7, characterized in that, After determining multiple simulated abnormal points in the fluid supply, the process includes: extracting the historical pressure loss coefficient of the corresponding pipeline section based on the multiple simulated abnormal points in the fluid supply, correcting the pressure loss coefficient by combining the real-time ambient temperature and the current emulsion viscosity parameter in the well, and obtaining the target corrected loss value; performing pre-compensation on the original target fluid supply pressure process parameters based on the target corrected loss value, and synchronously updating the compensated target fluid supply pressure process parameters to the multi-pump collaborative pressure control strategy to achieve pre-adjustment of the collaborative control strategy.

9. A multi-pump coordinated pressure control device for a remote liquid supply pipeline in mining, characterized in that, The device includes: a memory, a processor, and a multi-pump coordinated pressure control program for a remote liquid supply pipeline in mining, stored in the memory and executable on the processor, the multi-pump coordinated pressure control program for a remote liquid supply pipeline in mining configured to implement the steps of the multi-pump coordinated pressure control method for a remote liquid supply pipeline in mining as described in any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a multi-pump coordinated pressure control program for a remote liquid supply pipeline in a mine. When the multi-pump coordinated pressure control program for a remote liquid supply pipeline in a mine is executed by a processor, it implements the steps of the multi-pump coordinated pressure control method for a remote liquid supply pipeline in a mine as described in any one of claims 1 to 8.

Images