Hydraulic pump control optimization method and system for rare earth magnetic material extraction

Abstract

The application discloses a hydraulic pump control optimization method and system for rare earth magnetic material extraction, and relates to the technical field of hydraulic pump control.The method comprises the following steps: positioning a plurality of process stages in a rare earth magnetic material extraction system where hydraulic pumps are served, and directionally associating a plurality of hydraulic association index groups; performing local control strategy fitting according to the plurality of hydraulic association index groups, and outputting a plurality of initial pump control strategies; performing cross-stage coupling conflict arbitration, and outputting a plurality of optimized pump control strategies; after receiving and adopting the plurality of optimized pump control strategies, performing time sequence collaborative control on the hydraulic pumps; and performing closed-loop dynamic updating on the control parameters of the plurality of service hydraulic pumps in the rare earth magnetic material extraction system.The application solves the technical problem of insufficient hydraulic pump control precision in the rare earth magnetic material extraction process in the prior art, which leads to low extraction stability and efficiency, and achieves the technical effect of realizing precise control of the hydraulic pumps in the rare earth magnetic material extraction, and improving the stability and efficiency of the extraction process.

Description

Technical Field

[0001] This invention relates to the field of hydraulic pump control technology, and specifically to a hydraulic pump control optimization method and system for rare earth magnetic material extraction. Background Technology

[0002] In the extraction of rare earth magnetic materials, hydraulic pumps, as key equipment in hydrometallurgical production lines, play a crucial role in power supply and material transportation. Their operating status directly affects extraction efficiency and product quality. Traditional hydraulic pump control methods often employ single-parameter adjustment or centralized control, which are difficult to adapt to the differentiated needs of different process stages. This can easily lead to problems such as cross-stage parameter interference, response lag, and excessive energy consumption, resulting in insufficient stability of the extraction process, resource waste, and fluctuations in product purity.

[0003] Existing technologies suffer from insufficient control precision of hydraulic pumps during the extraction of rare earth magnetic materials, resulting in low extraction stability and efficiency. Summary of the Invention

[0004] This application provides a hydraulic pump control optimization method and system for rare earth magnetic material extraction, which addresses the technical problem of insufficient hydraulic pump control precision in the rare earth magnetic material extraction process, resulting in low extraction stability and efficiency.

[0005] In view of the above problems, this application provides a hydraulic pump control optimization method and system for rare earth magnetic material extraction.

[0006] The first aspect of this application provides a hydraulic pump control optimization method for rare earth magnetic material extraction, the method comprising:

[0007] In the rare earth magnetic material extraction system, multiple process stages are located where hydraulic pumps are used. Based on the functional attributes of the hydraulic pumps in these stages, multiple sets of hydraulic correlation indicators are associated. Multiple edge computing nodes, independently deployed in each process stage, collect real-time process information based on these hydraulic correlation indicators and perform local control strategy fitting, outputting multiple initial pump control strategies. After receiving the initial pump control strategies uploaded by the edge computing nodes, the collaborative control center outputs multiple optimized pump control strategies through cross-stage coupling conflict arbitration. After receiving and adopting the optimized pump control strategies issued by the collaborative control center, the edge computing nodes implement time-series collaborative control of multiple service hydraulic pumps in the process stages. Based on the dynamic disturbance range of the hydraulic correlation indicator sets, the edge computing nodes perform closed-loop dynamic updates of the control parameters of multiple service hydraulic pumps in the rare earth magnetic material extraction system.

[0008] A second aspect of this application provides a hydraulic pump control optimization system for rare earth magnetic material extraction, the system comprising:

[0009] The system comprises the following modules: an index group association module, used to locate multiple process stages in the rare earth magnetic material extraction system where hydraulic pumps are used, and to associate multiple hydraulic-related index groups based on the functional attributes of the hydraulic pumps in each process stage; a pump control strategy output module, used by multiple edge computing nodes independently deployed in each process stage to perform local control strategy fitting based on the multiple hydraulic-related index groups and real-time process information collected in each process stage; a pump control strategy optimization module, used by the collaborative control center to receive the multiple initial pump control strategies uploaded by the multiple edge computing nodes and, through cross-stage coupling conflict arbitration, output multiple optimized pump control strategies; a collaborative control module, used by the multiple edge computing nodes to receive and adopt the multiple optimized pump control strategies issued by the collaborative control center, and to implement time-series collaborative control of multiple service hydraulic pumps in the multiple process stages; and a closed-loop dynamic update module, used by the multiple edge computing nodes to perform closed-loop dynamic updates of the control parameters of multiple service hydraulic pumps in the rare earth magnetic material extraction system based on the dynamic disturbance range of the multiple hydraulic-related index groups.

[0010] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0011] In a rare earth magnetic material extraction system, multiple process stages served by hydraulic pumps are located, and multiple sets of hydraulic correlation indicators are directionally associated. Multiple edge computing nodes, independently deployed in each process stage, collect real-time process information based on these hydraulic correlation indicators, perform local control strategy fitting, and output multiple initial pump control strategies. A collaborative control center receives these initial pump control strategies from the edge computing nodes and outputs multiple optimized pump control strategies. After receiving and adopting these optimized strategies from the collaborative control center, time-series collaborative control is implemented on multiple service hydraulic pumps in the process stages. Based on the dynamic disturbance range of the multiple hydraulic correlation indicator sets, closed-loop dynamic updates are performed on the control parameters of the multiple service hydraulic pumps in the rare earth magnetic material extraction system. This achieves precise control of the hydraulic pumps in rare earth magnetic material extraction, improving the stability and efficiency of the extraction process. Attached Figure Description

[0012] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0013] Figure 1 This is a schematic diagram of a hydraulic pump control optimization method for rare earth magnetic material extraction provided in an embodiment of this application.

[0014] Figure 2 A schematic diagram of the hydraulic pump control optimization system for rare earth magnetic material extraction provided in this application embodiment.

[0015] Figure labeling: Indicator group association module 10, pump control strategy output module 20, pump control strategy optimization module 30, collaborative control module 40, closed-loop dynamic update module 50. Detailed Implementation

[0016] This application provides a hydraulic pump control optimization method and system for rare earth magnetic material extraction, which addresses the technical problem of insufficient hydraulic pump control precision in the rare earth magnetic material extraction process, resulting in low extraction stability and efficiency.

[0017] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0018] Example 1, as Figure 1 As shown, this application provides a hydraulic pump control optimization method for rare earth magnetic material extraction, the method comprising:

[0019] Step S100: In the rare earth magnetic material extraction system, locate multiple process stages where hydraulic pumps are used, and based on the hydraulic pump functional attributes of the multiple process stages, orient and associate multiple hydraulic correlation index groups.

[0020] Specifically, in the rare earth magnetic material extraction system, which is a physical entity, the system includes multiple process stages such as slurry conveying, pressure filtration, solvent extraction, and back-extraction washing. The system identifies each of these stages where hydraulic pumps provide process support. Then, based on the functional attributes of the hydraulic pumps in each process stage—for example, in the slurry conveying stage, the hydraulic pump provides power for material transport; in the pressure filtration stage, it controls pressure to achieve solid-liquid separation; in the solvent extraction stage, it requires precise flow control to ensure extraction efficiency; and in the back-extraction washing stage, it requires stable pressure to ensure washing effectiveness—hydraulic correlation index groups are specifically associated with each functional attribute. This ensures that the hydraulic correlation index group for each process stage accurately reflects the core operating parameters and process requirements of the hydraulic pumps at that stage.

[0021] Step S200: Multiple edge computing nodes independently deployed in the multiple process stages collect multiple real-time process information in the multiple process stages according to the multiple hydraulic correlation index groups, perform local control strategy fitting, and output multiple initial pump control strategies.

[0022] Specifically, in multiple process stages such as slurry transportation, pressure filtration, solvent extraction, and back-extraction washing, each stage independently deploys corresponding edge computing nodes. Each edge computing node, based on the hydraulic correlation index group associated with that stage, collects real-time process information, including physical quantity layer information, chemical quantity layer information, and time series layer information, in real time through a pre-deployed dedicated sensing topology, such as a dedicated topology formed by the sensor network of a certain process stage through local binding of upload permissions. Then, based on this information, hydraulic process auxiliary deviation identification is performed, the information is parsed to obtain data of various dimensions, and the static deviation is calculated by comparing it with the preset multi-layer process set values. The confidence weight is then combined with dynamic trend feature analysis and sensing accuracy attributes to output a hydraulic auxiliary deviation vector. Then, a pump control correction model architecture containing a static deviation layer, a dynamic feature layer, and a fusion output layer is constructed based on the deviation vector. By processing the deviation and dynamic features separately and performing cross-coupling at each layer, the local control strategy fitting is completed, and finally, the initial pump control strategy corresponding to each stage is output.

[0023] Step S300: After receiving multiple initial pump control strategies uploaded by the multiple edge computing nodes, the collaborative control center outputs multiple optimized pump control strategies through cross-stage coupling conflict arbitration.

[0024] Specifically, after receiving multiple initial pump control strategies uploaded by independently deployed edge computing nodes in various process stages such as slurry transportation, pressure filtration, solvent extraction, and back-extraction washing, the collaborative control center first performs adjacent strategy conflict detection on these initial pump control strategies, identifies and outputs material balance conflict sets, including material accumulation or shortage caused by mismatched flow rates in different stages; energy consumption accumulation conflict sets, including excessive energy consumption caused by simultaneous high-pressure operation in multiple stages; and process interference conflict sets, including extraction effect deviations caused by mutual influence of pressure fluctuations between stages. Then, based on the material balance conflict sets, dynamic flow calibration is performed according to process priority to output a flow calibration strategy; time-sharing staggered pressurization is planned based on the energy consumption accumulation conflict sets to output a time sequence avoidance strategy; and chemical isolation strategies are output through transition buffers and instruction insertion for process interference conflict sets. Finally, cross-exclusivity resolution is performed on the flow calibration strategy, time sequence avoidance strategy, and chemical isolation strategy to eliminate contradictions and duplications between strategies, and finally, multiple optimized pump control strategies adapted to each process stage are output.

[0025] Step S400: After receiving and adopting multiple optimized pump control strategies issued by the collaborative control center, the multiple edge computing nodes implement time-series collaborative control of multiple service hydraulic pumps in the multiple process stages.

[0026] Specifically, edge computing nodes, independently deployed in multiple process stages such as slurry conveying, pressure filtration, solvent extraction, and back-extraction washing, receive optimized pump control strategies adapted to their respective stages from the collaborative control center. Based on these strategies, they implement time-series collaborative control of the service hydraulic pumps in each stage. By integrating reliable pressure control commands, fatigue adaptive flow modulation, and dead-time compression parameters in the optimized pump control strategies, the operating sequence of hydraulic pumps in different process stages is coordinated. For example, after the material conveying is completed in the slurry conveying stage, the hydraulic pumps in the pressure filtration stage start pressurizing according to a preset sequence. The hydraulic pumps in the solvent extraction and back-extraction washing stages also operate sequentially or collaboratively according to the time sequence plan, ensuring that the operating rhythm of the hydraulic pumps in each stage matches the process connection requirements, reducing cross-stage interference, and improving the overall extraction efficiency.

[0027] Step S500: The multiple edge computing nodes perform closed-loop dynamic updates on the control parameters of multiple service hydraulic pumps in the rare earth magnetic material extraction system based on the dynamic disturbance range of the multiple hydraulic correlation index groups.

[0028] Specifically, edge computing nodes in multiple process stages, such as slurry transportation, pressure filtration, solvent extraction, and back-extraction washing, continuously monitor their respective hydraulic correlation index groups and capture the dynamic disturbance range of these indicators during operation in real time. Based on this disturbance range and combined with the control requirements of hydraulic pumps in each stage, the control parameters of each service hydraulic pump in the rare earth magnetic material extraction system, such as pressure, flow rate, and response delay compensation, are dynamically updated in a closed loop. Through continuous feedback and adjustment, the pump control parameters are always adapted to the real-time process information, ensuring that the hydraulic pumps can still operate stably when the indicators fluctuate, maintaining efficient coordination among the process stages, and thus ensuring the continuity and reliability of the rare earth magnetic material extraction process.

[0029] In one possible implementation, step S100 further includes:

[0030] The multiple process stages include slurry transport, pressure filtration, solvent extraction, and back-extraction and washing.

[0031] Specifically, the multiple process stages include a slurry delivery stage, a pressure filtration separation stage, a solvent extraction stage, and a back-extraction and washing stage. In the slurry delivery stage, a hydraulic pump is used to transport the rare earth slurry from the storage equipment to the subsequent processing stage. Stable flow rate and pressure are crucial to prevent slurry sedimentation or pipeline blockage. In the pressure filtration separation stage, the hydraulic pump provides pressure to allow the slurry to pass through a filter medium for solid-liquid separation. Precise control of pressure and holding time is necessary to improve separation efficiency. The solvent extraction stage is a critical step in rare earth separation and purification. The hydraulic pump is responsible for delivering the aqueous slurry and organic solvent to the extraction equipment in a specific ratio. The accuracy of flow rate control directly affects the extraction balance and separation effect. In the back-extraction and washing stage, a hydraulic pump delivers the back-extraction agent and washing liquid to elute and purify the target rare earth elements in the loaded organic phase. Coordination of flow rate and pressure parameters is necessary to ensure the back-extraction rate and product purity.

[0032] In one possible implementation, step S300 further includes:

[0033] Step S310: Perform adjacent strategy conflict detection on the multiple initial pump control strategies, and output the material balance conflict set, energy consumption accumulation conflict set, and process disturbance conflict set.

[0034] Step S320: Perform dynamic flow calibration for process priority based on the material balance conflict set, and output the flow calibration strategy.

[0035] Step S330: Perform time-sharing peak-shifting pressurization timing planning based on the energy consumption accumulation conflict set, and output timing avoidance strategy.

[0036] Step S340: Perform transition buffer neutralization instruction insertion for the process interference conflict set, and output chemical isolation strategy.

[0037] Step S350: Perform cross-exclusivity resolution on the flow calibration strategy, timing avoidance strategy and chemical isolation strategy to output the multiple optimized pump control strategies.

[0038] Specifically, the collaborative control center constructs a conflict detection matrix to compare and analyze the parameters of adjacent process stages in multiple initial pump control strategies. For material balance conflicts, it extracts the flow setpoint and material transfer rate in each strategy, calculates the input-output difference between adjacent stages, and includes the material balance conflict set when the difference exceeds the preset balance threshold. For energy consumption accumulation conflicts, it statistically analyzes the pressurization power and duration of each strategy, superimposes the energy consumption data of adjacent stages, and includes the energy consumption accumulation conflict set if the total energy consumption exceeds the stage energy consumption limit. For process interference conflicts, it analyzes the pressure fluctuation range and chemical reagent delivery rhythm of adjacent strategies, and includes the process interference conflict set when pressure fluctuations cross influence or reagent delivery timing conflicts. Finally, it outputs three types of conflict sets.

[0039] Based on the material accumulation or shortage problem caused by flow mismatch between adjacent process stages in the material balance conflict concentration, the process priority is first determined according to the core role of each process stage in the rare earth magnetic material extraction process. For example, the solvent extraction stage, which directly affects the extraction purity, has a higher priority than the slurry transportation stage. Then, the flow parameters involved in the conflict concentration are dynamically calibrated based on the priority. By calculating the material demand difference between adjacent stages, the hydraulic pump flow rate of the low-priority stage is adaptively adjusted to match the material handling capacity of the high-priority stage. At the same time, the flow deviation is dynamically corrected by combining the material concentration, transmission rate and other data in the real-time process information. Finally, a flow calibration strategy that ensures the balance of material input and output in each stage is output.

[0040] Based on the problem of excessive energy consumption due to the simultaneous high-pressure operation of multiple adjacent process stages in the energy consumption accumulation conflict concentration, this paper first analyzes the hydraulic pump pressurization power, pressurization duration, and operation period corresponding to each initial pump control strategy in the conflict concentration. Then, combined with the pressurization necessity and energy consumption sensitivity of each process stage, a time-sharing and staggered pressurization sequence planning is carried out. By adjusting the execution time of high-energy-consuming pressurization operations, the high-pressure operation periods of different process stages are staggered. For example, the high-pressure conveying of the slurry conveying stage is arranged during the low-pressure holding period of the filter press separation stage to avoid concentrated energy consumption accumulation. At the same time, pressurization sequence interval thresholds are set according to the process requirements of each stage to ensure that the staggered arrangement does not affect the process continuity. Finally, a timing avoidance strategy that coordinates the pressurization sequence of each stage to reduce the overall energy consumption is output.

[0041] To address the issue of mutual interference between adjacent process stages caused by pressure fluctuations and sudden flow changes, this approach first identifies the process connection nodes involved in the conflict. Transition buffer zones are then established at these nodes, specifying parameters such as the pressure range, flow fluctuation threshold, and duration of the buffer zones. Subsequently, based on these buffer parameters, neutralization commands are inserted into the relevant initial pump control strategies. For example, during the transition from the slurry transport stage to the filter press stage, a pressure buffer adjustment command is inserted to smooth pressure fluctuations; at the junction of the solvent extraction stage and the back-extraction washing stage, a flow rate gradual change command is inserted to reduce reagent transport impact. Through the insertion of these neutralization commands, process interference between adjacent stages is isolated, preventing parameter fluctuations from adversely affecting the chemical reaction conditions and separation effects of subsequent stages. Ultimately, a chemical isolation strategy that eliminates process interference between stages is output.

[0042] The cross-exclusivity of flow calibration, timing avoidance, and chemical isolation strategies was resolved by comparing each strategy individually in terms of parameter settings, timing arrangements, and process integration requirements. This identified inconsistencies, such as the flow calibration strategy requiring increased flow during a specific period, while the timing avoidance strategy limiting pressurization power during that period prevented flow increase, or the transition buffer duration of the chemical isolation strategy conflicting with the dynamic adjustment cycle of flow calibration. To address these inconsistencies, priority was weighed and parameters adjusted based on the core requirements of each stage of rare earth magnetic material extraction, eliminating mutual exclusion between strategies and ensuring compatibility and conflict-free operation when the three strategies work synergistically. Finally, the resolved strategies were integrated to form multiple optimized pump control strategies adapted to each process stage, including slurry delivery, pressure filtration, solvent extraction, and back-extraction washing, and then output.

[0043] In one possible implementation, step S200 further includes:

[0044] Step S210: Based on the first hydraulic correlation index group, perform local binding of the upload permission of the first sensor network pre-deployed in the first process stage to obtain the first dedicated sensing topology.

[0045] Step S220: The first edge computing node receives the first real-time process information uploaded by the first dedicated sensing topology.

[0046] Step S230: Based on the first real-time process information, identify hydraulic process auxiliary deviations and output the first hydraulic auxiliary deviation vector.

[0047] Step S240: Perform pump control parameter correction modeling based on the first hydraulic auxiliary deviation vector, and output the first initial pump control strategy.

[0048] Specifically, for the first process stage, based on the first hydraulic correlation index group corresponding to this stage, which covers indicators related to the operation of hydraulic pumps such as pressure, flow rate, and chemical reagent concentration, the upload permissions of the pre-deployed first sensor network are locally bound, restricting the data upload range of the sensor network to the local area of ​​this process stage. This ensures that the information collected by the sensors is only transmitted to the edge computing node responsible for this stage, avoiding cross-stage data interference, and ultimately forming a first dedicated sensing topology for the first process stage.

[0049] The first edge computing node deployed in the first process stage receives the first real-time process information uploaded by the first dedicated sensing topology through a preset communication protocol. This information includes physical quantity layer information of this stage, such as the pressure and flow data of the hydraulic pump; chemical quantity layer information, such as the concentration and pH value of the extract; and time series layer information, such as the time series of changes of each parameter, to provide raw data support for the fitting of subsequent local control strategies.

[0050] When identifying hydraulic process auxiliary deviations based on the first real-time process information, the information is first parsed to extract the first physical quantity layer information, the first chemical quantity layer information, and the first time series layer information. Then, predefined multi-layer process setting values ​​for the first process stage are called, and the extracted layer information is mapped and compared with the corresponding setting values. The first pressure dimension deviation, the first chemical dimension deviation, and the first time series dimension deviation are obtained by calculating the difference values. At the same time, dynamic trend feature analysis is performed on the first physical quantity layer information to output the first dynamic trend feature group. Finally, combined with the sensing accuracy attribute of the first dedicated sensing topology, the first dynamic trend feature group, the first pressure dimension deviation, the first chemical dimension deviation, and the first time series dimension deviation are weighted by confidence to comprehensively generate and output the first hydraulic auxiliary deviation vector.

[0051] When modeling pump control parameter correction based on the first hydraulic auxiliary deviation vector, a pump control correction model architecture is first constructed, comprising a static deviation layer, a dynamic feature layer, and a fusion output layer. The static deviation layer and the dynamic feature layer are connected in parallel, with their outputs both connected to the fusion output layer. In the static deviation layer, the first pressure dimension deviation, the first chemical dimension deviation, and the first time-series dimension deviation are corrected and compensated in parallel, outputting the first basic pressure correction term, the first flow correction term, and the first response delay compensation amount, respectively. In the dynamic feature layer, the first dynamic trend feature group is modeled using multimodal compensation, outputting the first dynamic compensation instruction set, which includes the first pressure impact pre-compensation pulse, the first PID integral adjustment amount, the first reverse damping control term, and the first mechanical fatigue suppression coefficient. In the fusion output layer, the first dynamic compensation instruction set is cross-coupled with the first basic pressure correction term, the first flow correction term, and the first response delay compensation amount, and the first initial pump control strategy is output after confidence weighting.

[0052] In one possible implementation, step S230 further includes:

[0053] Step S231: Analyze the first real-time process information to obtain the first physical quantity layer information, the first chemical quantity layer information, and the first time sequence layer information.

[0054] Step S232: Predefine the multi-layer process settings for the first process stage.

[0055] Step S233: Using the multi-layer process setting value mapping comparison, the first physical quantity layer information, the first chemical quantity layer information, and the first time series layer information are used to perform static deviation calculation and output the first pressure dimension deviation, the first chemical dimension deviation, and the first time series dimension deviation.

[0056] Step S234: Perform dynamic trend feature analysis on the first physical quantity layer information and output the first dynamic trend feature group.

[0057] Step S235: Based on the sensing accuracy attributes of the first dedicated sensing topology, perform confidence weighting on the first dynamic trend feature group, the first pressure dimension deviation, the first chemical dimension deviation, and the first time series dimension deviation, and output the first hydraulic auxiliary deviation vector.

[0058] Specifically, the first real-time process information is analyzed to separate the first physical quantity layer information reflecting the physical parameters of the hydraulic pump operation, such as pressure, flow rate, and flow velocity, the first chemical quantity layer information reflecting the chemical characteristics of the extraction system, such as solution concentration, pH value, and ion content, and the first time series layer information recording the changes of each parameter over time, such as the pressure fluctuation sequence over time and the time nodes of flow rate changes.

[0059] Based on the process requirements of the first process stage and the hydraulic pump operation standards, multi-layer process setting values ​​are predefined. These setting values ​​cover physical quantity standards such as pressure threshold and flow range corresponding to the first physical quantity layer information, chemical quantity standards such as solution concentration range and pH value permissible range corresponding to the first chemical quantity layer information, and time sequence standards such as parameter change cycle and response time limit corresponding to the time sequence layer information.

[0060] Using predefined multi-layer process setpoints, mapping and comparing them with the parsed first physical quantity layer information, first chemical quantity layer information, and first time series layer information respectively. The physical quantity layer setpoints are compared one by one with the actual pressure and flow rate data in the first physical quantity layer information, and the difference between the actual value and the setpoint is calculated to obtain the first pressure dimension deviation. The chemical quantity layer setpoints are matched and compared with the actual concentration, pH value, and other data in the first chemical quantity layer information, and the first chemical dimension deviation is output by calculating the deviation. The time series layer setpoints are compared with the actual time series data in the first time series layer information, and the first time series dimension deviation is calculated based on the time difference or period deviation. Finally, these three types of static deviations are output.

[0061] When performing dynamic trend feature analysis on the information of the first physical quantity layer, the information is first divided into rolling time series segments based on a preset time window to obtain a continuous data block sequence. Then, the flow-pressure derivative feature term is extracted from the continuous data block sequence to obtain the first dynamic derivative feature set. Next, the over-limit fluctuation state is quantified based on the continuous data blocks to output the first over-limit fluctuation duration ratio and the first pressure oscillation spectrum intensity. Then, the continuous data blocks are traversed to obtain the first pressurization cycle accumulation number, and this number is associated with the first flow trend direction marker. Finally, the first dynamic derivative feature set, the first over-limit fluctuation duration ratio, the first pressure oscillation spectrum intensity, the first pressurization cycle accumulation number, and the first flow trend direction marker are encapsulated to output the first dynamic trend feature group.

[0062] Based on the sensing accuracy attributes of the first dedicated sensing topology, including the accuracy level, data acquisition error range, and signal stability parameters of each sensor in the topology, the confidence weights corresponding to the first dynamic trend feature group, the first pressure dimension deviation, the first chemical dimension deviation, and the first time-series dimension deviation are determined. For example, features or deviations corresponding to data acquired by sensors with higher accuracy and smaller errors are assigned higher weights. Subsequently, these features and deviations are multiplied by their respective weights, and the results are integrated to finally output the first hydraulic auxiliary deviation vector that comprehensively reflects the deviations and dynamic trends of each dimension.

[0063] In one possible implementation, step S240 further includes:

[0064] Step S241: Construct a pump control correction model architecture, wherein the pump control correction model architecture includes a static deviation layer, a dynamic feature layer and a fusion output layer, the static deviation layer and the dynamic feature layer are connected in parallel, and the output end is connected to the fusion output layer.

[0065] Step S242: Perform parallel correction and compensation of the first pressure dimension deviation, the first chemical dimension deviation, and the first time sequence dimension deviation in the static deviation layer, and output the first basic pressure correction term, the first flow rate correction term, and the first response delay compensation amount.

[0066] Step S243: Perform multimodal compensation modeling of the first dynamic trend feature group in the dynamic feature layer, and output the first dynamic compensation instruction set, wherein the first dynamic compensation instruction set includes a first pressure shock pre-compensation pulse, a first PID integral adjustment, a first reverse damping control term, and a first mechanical fatigue suppression coefficient.

[0067] Step S244: Execute the cross-coupling of the first dynamic compensation instruction set, the first basic pressure correction term, the first flow correction term, and the first response delay compensation amount in the fusion output layer, and output the first initial pump control strategy.

[0068] Specifically, a pump control correction model architecture suitable for hydraulic pump control in the first process stage is constructed. This architecture clearly includes three core layers: a static deviation layer, a dynamic feature layer, and a fusion output layer. The static deviation layer and the dynamic feature layer are configured in parallel, serving as parallel input processing layers for the model. They can independently process the input static deviation information and dynamic trend features, respectively. Simultaneously, the outputs of both the static deviation layer and the dynamic feature layer are connected to the fusion output layer, allowing the processing results from both layers to be aggregated and comprehensively calculated. This forms a model structure with two-level parallel processing and a single-level fusion output, providing architectural support for the subsequent accurate correction of pump control parameters.

[0069] In the static deviation layer, for the first pressure dimension deviation, a proportional-integral correction algorithm is used. The deviation value is multiplied by a preset pressure adjustment coefficient and then the integral compensation is added to calculate the first basic pressure correction term. For the first chemical dimension deviation, the chemical parameter-flow correlation mapping table is queried. This table stores the flow correction coefficients corresponding to different chemical deviations. The deviation value is substituted into the table to match the corresponding coefficient, and then multiplied by the baseline flow rate to output the first flow correction term. For the first time-series dimension deviation, the first response delay compensation amount is calculated based on the time-series synchronization error formula: error value = actual response time - set response time, combined with the delay compensation factor. The three correction operations are executed synchronously through a parallel computing module and the results are output.

[0070] Based on the pressure change rate in the first dynamic trend feature group, a first pressure shock pre-compensation pulse is determined and output to offset pressure sudden changes in advance by matching a preset pressure shock compensation pulse parameter library. Then, according to the first over-limit fluctuation duration ratio, the integral time constant of the flow PID controller is adjusted according to preset rules to output a first PID integral adjustment quantity that can suppress flow fluctuations. Then, based on the first pressure oscillation spectrum intensity, the reverse damping control parameter library is queried to match and output a first reverse damping control term that can weaken pressure oscillations. Finally, combined with the first pressurization cycle accumulation number and the first flow trend direction mark, a first mechanical fatigue suppression coefficient for reducing equipment wear is calculated and output through a fatigue strength attenuation model. These four instructions together constitute the first dynamic compensation instruction set.

[0071] In the fusion output layer, cross-coupling operations are performed on the first dynamic compensation instruction set, the first basic pressure correction term, the first flow correction term, and the first response delay compensation amount. After generating the initial disturbance rejection pressure command by linearly superimposing the first basic pressure correction term and the first pressure impact pre-compensation pulse, it is injected into the first reverse damping control term to output a reliable pressure control command. The first flow correction term and the first mechanical fatigue suppression coefficient are multiplied to obtain the reference fatigue adaptive flow rate. Then, the response rate of the flow rate is dynamically modulated by the first PID integral adjustment amount to output the modulated fatigue adaptive flow rate. The timing components of the first response delay compensation amount and the first PID integral adjustment amount are integrated to output the dead time compression parameter. Finally, the reliable pressure control command, the modulated fatigue adaptive flow rate, and the dead time compression parameter are weighted by confidence to comprehensively output the first initial pump control strategy.

[0072] In one possible implementation, step S234 further includes:

[0073] Step S2341: Perform rolling time-series segmentation on the first physical quantity layer information based on a preset time window to obtain a continuous data block sequence.

[0074] Step S2342: Extract the flow pressure derivative feature term from the continuous data block sequence to obtain the first dynamic derivative feature set.

[0075] Step S2343: Quantize the over-limit fluctuation state based on the continuous data block, and output the first over-limit fluctuation duration ratio and the first pressure oscillation spectrum intensity.

[0076] Step S2344: Traverse the continuous data blocks to obtain the first pressurization cycle cumulative number, wherein the first pressurization cycle cumulative number is associated with the first flow trend direction marker.

[0077] Step S2345: Encapsulate the first dynamic derivative feature set, the first over-limit fluctuation duration ratio, the first pressure oscillation spectrum intensity, the first pressurization cycle cumulative number and the first flow trend direction marker, and output the first dynamic trend feature set.

[0078] Specifically, rolling temporal segmentation is performed on the information of the first physical quantity layer based on a preset time window (e.g., 10 seconds / window). That is, starting from the initial moment, after extracting data segments according to the length of the time window, the window slides backward with a fixed step size (e.g., 5 seconds) to continuously extract until all information is covered, and finally a continuous and partially overlapping sequence of continuous data blocks is obtained, thereby realizing the temporal segmentation of physical quantity information.

[0079] For the continuous data block sequence obtained by rolling time-series segmentation of the first physical quantity layer information, the first and second derivatives of the flow and pressure parameters with time in each data block are calculated one by one. At the same time, feature terms such as the rate of change of the flow-pressure ratio and the partial derivative of pressure with respect to flow are extracted. These feature terms that reflect the dynamic change rate of the parameters are integrated and summarized to form the first dynamic derivative feature set that can reflect the dynamic response characteristics of the hydraulic system.

[0080] When quantifying over-limit fluctuations based on continuous data blocks, a normal fluctuation threshold range is first set for physical parameters such as pressure and flow rate in each continuous data block. This threshold is predefined based on the process requirements of the first process stage. Then, each data block is traversed, and the duration for which the physical parameters exceed the preset threshold is counted. The ratio of this duration to the total duration of the data block is used as the over-limit percentage of a single block. The average of the over-limit percentages of all data blocks is then taken to obtain the first over-limit fluctuation duration percentage. At the same time, spectral analysis, such as Fourier transform, is performed on the pressure fluctuation signal in each data block to extract the energy intensity corresponding to the main oscillation frequency in the spectrum. The analysis results of all data blocks are combined to output the first pressure oscillation spectrum intensity that can reflect the severity of pressure fluctuations.

[0081] The system iterates through each data block in the continuous data block sequence, identifies the pressurization cycle of the hydraulic pump within each data block, and determines that a pressurization cycle is completed when the pressure parameter in the data block rises from below a preset pressurization threshold to above the threshold and then falls back below the threshold. The number of cycles is accumulated to obtain the first pressurization cycle cumulative count. At the same time, when recording each pressurization cycle, the system combines the change direction of the flow parameter during the corresponding cycle. For example, if the flow increases with the pressure, it is marked as positive, and if it decreases with the pressure, it is marked as negative. This associates the first pressurization cycle cumulative count with the corresponding first flow trend direction mark, thus binding the cycle count with the flow change trend.

[0082] The extracted first dynamic derivative feature set, the first over-limit fluctuation duration ratio, the first pressure oscillation spectrum intensity, the first pressurization cycle accumulation number, and the associated first flow trend direction marker are structured and encapsulated. According to the preset data format, these feature information are integrated into a complete feature set in the form of feature matrix and label combination. Finally, the first dynamic trend feature group that can comprehensively reflect the dynamic change trend of the first physical quantity layer information is output.

[0083] In one possible implementation, step S243 further includes:

[0084] Step S2431: Output the first pressure shock pre-compensation pulse based on the pressure change rate matching of the first dynamic trend feature group.

[0085] Step S2432: Based on the integral time constant of the first over-limit fluctuation continuous proportion adjustment flow PID controller, output the first PID integral adjustment amount.

[0086] Step S2433: Output the first reverse damping control term based on the intensity matching of the first pressure oscillation spectrum.

[0087] Step S2434: Based on the first pressurization cycle cumulative number and the first flow trend direction mark, perform fatigue strength attenuation modeling and output the first mechanical fatigue inhibition coefficient.

[0088] Specifically, based on the pressure change rate contained in the first dynamic trend feature group, which originates from the first dynamic derivative feature set, it is matched with a pre-stored pressure shock pre-compensation pulse parameter library. The parameter library contains parameters such as pulse amplitude, width, and triggering timing corresponding to different pressure change rates. When the pressure change rate reaches or exceeds the set shock warning threshold, the corresponding pulse parameters are selected according to the matching result, and a first pressure shock pre-compensation pulse is generated and output to intervene in the pressure output of the hydraulic pump in advance and offset the impending pressure shock.

[0089] Based on the percentage of the first over-limit fluctuation, the integral time constant of the flow PID controller is dynamically adjusted according to the preset adjustment rules. If the percentage of the first over-limit fluctuation is high, it indicates that the flow fluctuation is frequent and severe. In this case, the integral time constant is reduced to speed up the response speed of the integral action and enhance the ability to correct flow deviation. If the percentage is low, it indicates that the flow is relatively stable. In this case, the integral time constant is increased to avoid system oscillation caused by excessive integration. Finally, the adjusted integral time constant is used as the first PID integral adjustment output.

[0090] Based on the intensity of the first pressure oscillation spectrum, a preset reverse damping control parameter library is consulted. This library stores parameters such as damping coefficients, suppression frequency ranges, and control strengths corresponding to different spectral intensity ranges. According to the specific value of the current first pressure oscillation spectrum intensity and its main oscillation frequency components, the most suitable reverse damping parameter combination is matched from the parameter library. This generates and outputs the first reverse damping control term used to suppress pressure oscillations, thereby reducing the amplitude and duration of pressure fluctuations.

[0091] Fatigue strength attenuation modeling is performed based on the cumulative number of the first pressurization cycle and the first flow trend direction mark. First, the cumulative number of pressurization cycles is taken as the core variable and substituted into the preset fatigue attenuation formula: fatigue strength = initial strength × e^(-k × number of cycles), where k is the attenuation coefficient. Then, the attenuation coefficient k is corrected by combining the first flow trend direction mark. When the flow is in the opposite direction, the value of k is increased to accelerate the attenuation correction. The compensation coefficient reflecting the current mechanical fatigue state of the hydraulic pump is calculated through this model, and finally the first mechanical fatigue suppression coefficient used to reduce equipment wear is output.

[0092] In one possible implementation, step S244 further includes:

[0093] Step S2441: After generating an initial anti-disturbance pressure command by linearly superimposing the first basic pressure correction term and the first pressure shock pre-compensation pulse, the initial anti-disturbance pressure command is injected into the first reverse damping control term, and a reliable pressure control command is output.

[0094] Step S2442: Perform a product operation on the first flow correction term and the first mechanical fatigue suppression coefficient to generate a reference fatigue adaptive flow. Then, apply the first PID integral adjustment to dynamically modulate the response rate of the reference fatigue adaptive flow and output the modulated fatigue adaptive flow.

[0095] Step S2443: By integrating the timing components of the first response delay compensation amount and the first PID integral adjustment amount, output the dead time compression parameter.

[0096] Step S2444: Perform confidence-weighted processing on the reliable pressure control command, modulated fatigue adaptive flow rate, and dead-time compression parameters, and output the first initial pump control strategy.

[0097] Specifically, the first basic pressure correction term and the first pressure shock pre-compensation pulse are first linearly superimposed, that is, by directly adding the values ​​of the two according to their corresponding dimensions, an initial anti-disturbance pressure command that can initially resist sudden pressure disturbances is generated; then, the initial anti-disturbance pressure command is input to the first reverse damping control term, and the pressure fluctuation is further weakened by the suppression effect of reverse damping, and finally a more stable and reliable pressure control command is output.

[0098] The first flow correction term and the first mechanical fatigue suppression coefficient are multiplied together to obtain a reference fatigue adaptive flow that takes into account both flow correction requirements and mechanical fatigue suppression factors. Then, the response rate of the reference fatigue adaptive flow is dynamically modulated by applying the first PID integral adjustment. The flow rate is adjusted in real time according to the magnitude of the integral adjustment to make the flow response more in line with the process timing requirements. Finally, the modulated fatigue adaptive flow after rate modulation is output.

[0099] Extract time-related time components from the first response delay compensation quantity, such as the duration of delay compensation and the trigger time, and time components from the first PID integral adjustment quantity, such as the start time of integral action and the adjustment period. Integrate these two components onto the same time axis using a time alignment algorithm to calculate and output the comprehensive correction parameter that can shorten the system response dead zone, i.e., the dead zone time compression parameter.

[0100] Based on the importance and influence weight of reliable pressure control commands, modulated fatigue adaptive flow, and dead time compression parameters in the rare earth magnetic material extraction process, corresponding confidence coefficients are assigned to these three parameters. For example, the confidence coefficient for reliable pressure control is set to 0.4, the confidence coefficient for modulated fatigue adaptive flow is set to 0.3, and the confidence coefficient for dead time is set to 0.3. Then, each parameter is multiplied by its respective confidence coefficient, and the product results are summed and integrated to form a comprehensive set of control strategy parameters. Finally, the first initial pump control strategy that can guide the operation of the hydraulic pump in the first process stage is output.

[0101] Example 2, based on the same inventive concept as the hydraulic pump control optimization method for rare earth magnetic material extraction in the foregoing examples, such as... Figure 2 As shown, this application provides a hydraulic pump control optimization system for rare earth magnetic material extraction. The system and method embodiments in this application are based on the same inventive concept. The system includes:

[0102] The indicator group association module 10 is used to locate multiple process stages in the rare earth magnetic material extraction system where hydraulic pumps are used, and to directionally associate multiple hydraulic association indicator groups based on the hydraulic pump functional attributes of the multiple process stages.

[0103] The pump control strategy output module 20 is used to perform local control strategy fitting by multiple edge computing nodes independently deployed in the multiple process stages based on the multiple hydraulic correlation index groups to collect multiple real-time process information in the multiple process stages, and output multiple initial pump control strategies.

[0104] The pump control strategy optimization module 30 is used to coordinate the control center to receive multiple initial pump control strategies uploaded by the multiple edge computing nodes, and output multiple optimized pump control strategies through cross-stage coupling conflict arbitration.

[0105] The collaborative control module 40 is used to implement time-series collaborative control of multiple service hydraulic pumps in the multiple process stages after the multiple edge computing nodes receive and adopt multiple optimized pump control strategies issued by the collaborative control center.

[0106] The closed-loop dynamic update module 50 is used by the multiple edge computing nodes to perform closed-loop dynamic updates on the control parameters of multiple service hydraulic pumps in the rare earth magnetic material extraction system based on the dynamic disturbance range of the multiple hydraulic correlation index groups.

[0107] Furthermore, the system is also used to implement the following functions:

[0108] Adjacent strategy conflict detection is performed on the multiple initial pump control strategies, outputting a material balance conflict set, an energy consumption accumulation conflict set, and a process disturbance conflict set; dynamic flow calibration for process priority is performed based on the material balance conflict set, outputting a flow calibration strategy; time-sharing peak-shifting pressurization timing planning is performed based on the energy consumption accumulation conflict set, outputting a timing avoidance strategy; transition buffer neutralization command insertion is performed on the process disturbance conflict set, outputting a chemical isolation strategy; cross-exclusivity resolution is performed on the flow calibration strategy, timing avoidance strategy, and chemical isolation strategy, outputting the multiple optimized pump control strategies.

[0109] Furthermore, the system is also used to implement the following functions:

[0110] Based on the first hydraulic correlation index group, the upload permission of the first sensor network pre-deployed in the first process stage is locally bound to obtain the first dedicated sensing topology; the first edge computing node receives the first real-time process information uploaded by the first dedicated sensing topology; based on the first real-time process information, hydraulic process auxiliary deviation identification is performed, and the first hydraulic auxiliary deviation vector is output; based on the first hydraulic auxiliary deviation vector, pump control parameter correction modeling is performed, and the first initial pump control strategy is output.

[0111] Furthermore, the system is also used to implement the following functions:

[0112] The first real-time process information is parsed to obtain the first physical quantity layer information, the first chemical quantity layer information, and the first time series layer information; multi-layer process setting values ​​for the first process stage are predefined; the first physical quantity layer information, the first chemical quantity layer information, and the first time series layer information are mapped and compared using the multi-layer process setting values, and static deviation calculation is performed to output the first pressure dimension deviation, the first chemical dimension deviation, and the first time series dimension deviation; dynamic trend feature analysis is performed on the first physical quantity layer information to output the first dynamic trend feature group; based on the sensing accuracy attribute of the first dedicated sensing topology, the confidence weights of the first dynamic trend feature group, the first pressure dimension deviation, the first chemical dimension deviation, and the first time series dimension deviation are calculated to output the first hydraulic auxiliary deviation vector.

[0113] Furthermore, the system is also used to implement the following functions:

[0114] A pump control correction model architecture is constructed, comprising a static deviation layer, a dynamic feature layer, and a fusion output layer. The static deviation layer and the dynamic feature layer are connected in parallel, and their outputs are connected to the fusion output layer. Parallel correction and compensation are performed on the static deviation layer for the first pressure dimension deviation, the first chemical dimension deviation, and the first time-series dimension deviation, outputting a first basic pressure correction term, a first flow correction term, and a first response delay compensation amount. Multimodal compensation modeling of the first dynamic trend feature group is performed on the dynamic feature layer, outputting a first dynamic compensation instruction set, which includes a first pressure impact pre-compensation pulse, a first PID integral adjustment term, a first reverse damping control term, and a first mechanical fatigue suppression coefficient. The cross-coupling of the first dynamic compensation instruction set, the first basic pressure correction term, the first flow correction term, and the first response delay compensation amount is executed on the fusion output layer, outputting the first initial pump control strategy.

[0115] Furthermore, the system is also used to implement the following functions:

[0116] Based on a preset time window, the information of the first physical quantity layer is subjected to rolling time-series segmentation to obtain a continuous data block sequence; the flow-pressure derivative feature term is extracted from the continuous data block sequence to obtain a first dynamic derivative feature set; the over-limit fluctuation state is quantified based on the continuous data blocks to output the first over-limit fluctuation duration ratio and the first pressure oscillation spectrum intensity; the continuous data blocks are traversed to obtain the first pressurization cycle accumulation number, wherein the first pressurization cycle accumulation number is associated with the first flow trend direction marker; the first dynamic derivative feature set, the first over-limit fluctuation duration ratio, the first pressure oscillation spectrum intensity, the first pressurization cycle accumulation number, and the first flow trend direction marker are encapsulated to output the first dynamic trend feature group.

[0117] Furthermore, the system is also used to implement the following functions:

[0118] The first pressure shock pre-compensation pulse is output based on the pressure change rate matching of the first dynamic trend feature group; the first PID integral adjustment amount is output based on the integral time constant of the first over-limit fluctuation continuous proportion adjustment flow PID controller; the first reverse damping control term is output based on the first pressure oscillation spectrum intensity matching; the first mechanical fatigue suppression coefficient is output based on the first pressurization cycle cumulative number and the first flow trend direction mark for fatigue strength attenuation modeling.

[0119] Furthermore, the system is also used to implement the following functions:

[0120] After generating an initial anti-disturbance pressure command by linearly superimposing the first basic pressure correction term and the first pressure shock pre-compensation pulse, the initial anti-disturbance pressure command is injected into the first reverse damping control term, and a reliable pressure control command is output. After performing a product operation on the first flow correction term and the first mechanical fatigue suppression coefficient to generate a reference fatigue adaptive flow, the response rate of the reference fatigue adaptive flow is dynamically modulated by the first PID integral adjustment, and the modulated fatigue adaptive flow is output. By integrating the timing components of the first response delay compensation and the first PID integral adjustment, a dead time compression parameter is output. The reliable pressure control command, the modulated fatigue adaptive flow, and the dead time compression parameter are subjected to confidence weighting processing, and the first initial pump control strategy is output.

[0121] Furthermore, the system is also used to implement the following functions:

[0122] The multiple process stages include slurry transport, pressure filtration, solvent extraction, and back-extraction washing.

[0123] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. Additionally, the processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0124] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0125] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and variations fall within the scope of this application and its equivalents, this application intends to include such modifications and variations.

Claims

1. A method for optimizing hydraulic pump control for rare earth magnetic material extraction, characterized in that, The method includes: In the rare earth magnetic material extraction system, multiple process stages are located where hydraulic pumps are used, and multiple hydraulic correlation index groups are directionally associated based on the functional attributes of the hydraulic pumps in the multiple process stages. Multiple edge computing nodes, independently deployed in the multiple process stages, collect multiple real-time process information in the multiple process stages based on the multiple hydraulic correlation index groups, perform local control strategy fitting, and output multiple initial pump control strategies. After receiving multiple initial pump control strategies uploaded by the multiple edge computing nodes, the collaborative control center outputs multiple optimized pump control strategies through cross-stage coupling conflict arbitration. After receiving and adopting multiple optimized pump control strategies issued by the collaborative control center, the multiple edge computing nodes implement time-series collaborative control of multiple service hydraulic pumps in the multiple process stages. The multiple edge computing nodes perform closed-loop dynamic updates on the control parameters of multiple service hydraulic pumps in the rare earth magnetic material extraction system based on the dynamic disturbance range of the multiple hydraulic correlation index groups. Among them, after receiving multiple initial pump control strategies uploaded by the multiple edge computing nodes, the collaborative control center outputs multiple optimized pump control strategies through cross-stage coupling conflict arbitration, including: The multiple initial pump control strategies are subjected to adjacent strategy conflict detection, and the material balance conflict set, energy consumption accumulation conflict set, and process disturbance conflict set are output. Based on the material balance conflict set, perform dynamic flow calibration for process priority and output a flow calibration strategy. Based on the energy consumption accumulation conflict set, time-sharing peak-shifting pressurization timing planning is performed, and timing avoidance strategy is output; For the aforementioned set of process interference conflicts, a transition buffer and instruction insertion are performed, and a chemical isolation strategy is output. The flow calibration strategy, timing avoidance strategy, and chemical isolation strategy are cross-exclusively resolved to output the multiple optimized pump control strategies.

2. The hydraulic pump control optimization method for rare earth magnetic material extraction as described in claim 1, characterized in that, The method involves multiple edge computing nodes independently deployed in multiple process stages collecting real-time process information based on multiple sets of hydraulic correlation indicators in each process stage, fitting local control strategies, and outputting multiple initial pump control strategies. Based on the first hydraulic correlation index group, the upload permission of the first sensor network pre-deployed in the first process stage is bound locally to obtain the first dedicated sensing topology; The first edge computing node receives the first real-time process information uploaded by the first dedicated sensing topology; Based on the first real-time process information, hydraulic process auxiliary deviation identification is performed, and a first hydraulic auxiliary deviation vector is output. Based on the first hydraulic auxiliary deviation vector, pump control parameters are corrected and modeled to output the first initial pump control strategy.

3. The hydraulic pump control optimization method for rare earth magnetic material extraction as described in claim 2, characterized in that, Based on the first real-time process information, hydraulic process auxiliary deviation identification is performed, and a first hydraulic auxiliary deviation vector is output. The method includes: The first real-time process information is analyzed to obtain the first physical quantity layer information, the first chemical quantity layer information, and the first time sequence layer information; Predefine multi-layer process settings for the first process stage; The first physical quantity layer information, the first chemical quantity layer information and the first time series layer information are mapped and compared using the multi-layer process set value to perform static deviation calculation and output the first pressure dimension deviation, the first chemical dimension deviation and the first time series dimension deviation. Perform dynamic trend feature analysis on the information of the first physical quantity layer, and output the first dynamic trend feature group; Based on the sensing accuracy attributes of the first dedicated sensing topology, the confidence weights of the first dynamic trend feature group, the first pressure dimension deviation, the first chemical dimension deviation, and the first time series dimension deviation are calculated, and the first hydraulic auxiliary deviation vector is output.

4. The hydraulic pump control optimization method for rare earth magnetic material extraction as described in claim 3, characterized in that, The method includes: modeling pump control parameter correction based on the first hydraulic auxiliary deviation vector, and outputting a first initial pump control strategy. A pump control correction model architecture is constructed, wherein the pump control correction model architecture includes a static deviation layer, a dynamic feature layer and a fusion output layer, the static deviation layer and the dynamic feature layer are connected in parallel, and the output end is connected to the fusion output layer; Parallel correction and compensation are performed on the first pressure dimension deviation, the first chemical dimension deviation, and the first time dimension deviation in the static deviation layer, and the first basic pressure correction term, the first flow rate correction term, and the first response delay compensation amount are output. Multimodal compensation modeling of the first dynamic trend feature group is performed in the dynamic feature layer, and a first dynamic compensation instruction set is output. The first dynamic compensation instruction set includes a first pressure shock pre-compensation pulse, a first PID integral adjustment, a first reverse damping control term, and a first mechanical fatigue suppression coefficient. The first dynamic compensation instruction set, the first basic pressure correction term, the first flow correction term, and the first response delay compensation amount are cross-coupled in the fusion output layer to output the first initial pump control strategy.

5. The hydraulic pump control optimization method for rare earth magnetic material extraction as described in claim 4, characterized in that, The method includes performing dynamic trend feature analysis on the information of the first physical quantity layer and outputting a first dynamic trend feature group. Based on a preset time window, rolling time-series segmentation is performed on the information of the first physical quantity layer to obtain a continuous data block sequence; The flow pressure derivative feature term is extracted from the continuous data block sequence to obtain the first dynamic derivative feature set; Based on the continuous data block, the over-limit fluctuation state is quantified, and the first over-limit fluctuation duration ratio and the first pressure oscillation spectrum intensity are output. Traverse the continuous data blocks to obtain the first pressurization cycle accumulation count, wherein the first pressurization cycle accumulation count is associated with a first flow trend direction marker; The first dynamic derivative feature set, the first over-limit fluctuation duration ratio, the first pressure oscillation spectrum intensity, the first pressurization cycle cumulative number, and the first flow trend direction marker are encapsulated, and the first dynamic trend feature set is output.

6. The hydraulic pump control optimization method for rare earth magnetic material extraction as described in claim 5, characterized in that, The method includes performing multimodal compensation modeling of the first dynamic trend feature group in the dynamic feature layer and outputting a first dynamic compensation instruction set. The first pressure shock pre-compensation pulse is output based on the pressure change rate matching of the first dynamic trend feature group. Based on the integral time constant of the first over-limit fluctuation continuous proportion regulating flow PID controller, the first PID integral regulation amount is output; The first reverse damping control term is output based on the intensity of the first pressure oscillation spectrum. Based on the cumulative number of the first pressurization cycles and the first flow trend direction marker, fatigue strength attenuation modeling is performed, and the first mechanical fatigue inhibition coefficient is output.

7. The hydraulic pump control optimization method for rare earth magnetic material extraction as described in claim 4, characterized in that, The method involves cross-coupling the first dynamic compensation instruction set, the first basic pressure correction term, the first flow rate correction term, and the first response delay compensation amount at the fusion output layer to output the first initial pump control strategy. After generating an initial disturbance rejection pressure command by linearly superimposing the first basic pressure correction term and the first pressure shock pre-compensation pulse, the initial disturbance rejection pressure command is injected into the first reverse damping control term, and a reliable pressure control command is output. After performing a product operation on the first flow correction term and the first mechanical fatigue suppression coefficient to generate a reference fatigue adaptive flow, the response rate of the reference fatigue adaptive flow is dynamically modulated by the first PID integral adjustment, and the modulated fatigue adaptive flow is output. By integrating the timing components of the first response delay compensation and the first PID integral adjustment, dead time compression parameters are output. The reliable pressure control command, the modulated fatigue adaptive flow rate, and the dead time compression parameters are weighted by confidence to output the first initial pump control strategy.

8. The hydraulic pump control optimization method for rare earth magnetic material extraction as described in claim 1, characterized in that, The multiple process stages include slurry transport, pressure filtration, solvent extraction, and back-extraction and washing.

9. A hydraulic pump control optimization system for rare earth magnetic material extraction, characterized in that, The system is used to implement the hydraulic pump control optimization method for rare earth magnetic material extraction according to any one of claims 1-8, the system comprising: The indicator group association module is used to locate multiple process stages in the rare earth magnetic material extraction system where hydraulic pumps are used, and to directionally associate multiple hydraulic-related indicator groups based on the hydraulic pump functional attributes of the multiple process stages. The pump control strategy output module is used to perform local control strategy fitting by multiple edge computing nodes independently deployed in the multiple process stages based on the multiple hydraulic correlation index groups to collect multiple real-time process information in the multiple process stages and output multiple initial pump control strategies. The pump control strategy optimization module is used to coordinate the control center to receive multiple initial pump control strategies uploaded by the multiple edge computing nodes, and output multiple optimized pump control strategies through cross-stage coupling conflict arbitration. The collaborative control module is used to implement time-series collaborative control of multiple service hydraulic pumps in the multiple process stages after the multiple edge computing nodes receive and adopt multiple optimized pump control strategies issued by the collaborative control center. The closed-loop dynamic update module is used by the multiple edge computing nodes to perform closed-loop dynamic updates on the control parameters of multiple service hydraulic pumps in the rare earth magnetic material extraction system based on the dynamic disturbance range of the multiple hydraulic correlation index groups.

Images