A monitoring system for detecting the dose-effect relationship of phosphate ore flotation reagents and its process

Abstract

The application discloses a monitoring system for detecting the amount-effect relationship of phosphate ore flotation reagent and a process thereof, relates to the technical field of intelligent control of mineral processing, and comprises a multilevel influence net construction module, a dynamic relationship mapping module, a collaborative evolution driving module, an abnormal phenomenon capturing module, a decision flow activating module and a collaborative regulation instruction set integrating module. The system drives the collaborative evolution of the influence net through a multilevel influence net, a progressive action chain, dynamic mapping of reagent action and real-time injection of process parameter flow into the system, and captures two types of abnormal phenomena, i.e. response lag of the reagent action unit and migration blockage of the mineral phase node, in the process. The system generates instant correction instructions for reagent parameters aiming at the response lag phenomenon, generates action chain topology adjustment instructions aiming at the migration blockage phenomenon, integrates the two types of instructions into a collaborative regulation instruction set and issues the instruction set to a production execution end, and thus realizes dynamic and visual monitoring and accurate collaborative regulation of the amount-effect relationship of the flotation process reagent.

Description

Technical Field

[0001] This invention belongs to the field of intelligent control technology for mineral processing, and in particular relates to a monitoring system and process for detecting the dosage-effect relationship of flotation reagents in phosphate rock. Background Technology

[0002] Existing technologies for monitoring the dosage-efficiency relationship in phosphate rock flotation processes primarily rely on threshold monitoring and regression analysis of key parameters such as reagent dosage, pulp pH, final concentrate grade, and recovery rate. Essentially, these methods establish a static statistical correlation between end-point indicators and input parameters, managing the flotation process as a whole or divided into several isolated units. Their technical foundation lies in empirical models of parameters and results, lacking real-time descriptions of the continuous evolution of mineral particle states during flotation and the dynamic changes in reagent efficiency at specific stages.

[0003] Due to the failure to construct a dynamic mechanism model synchronized with the physical process in real time, existing technologies have a fundamental limitation: they cannot track and quantify the actual transmission and consumption of reagent effects in each flotation step in real time. When production indicators are abnormal, the system can only alarm for parameters exceeding limits or results deviating, but cannot identify the root cause of the problem—whether it is insufficient reagent adsorption in the grinding stage, selective failure in the cleaning stage, or disordered mineral migration paths caused by internal process circulation. This leads to vague fault diagnosis and a lack of targeted control measures.

[0004] Existing anomaly identification and control logic is relatively simplistic. Systems typically only respond to parameter exceeding limits and trigger fixed adjustment strategies. This approach fails to distinguish between two different types of problems: "insufficient absolute reagent dosage" and "low reagent efficiency in a specific process step," and it cannot identify "obstructed mineral migration paths" caused by mismatched process structure or kinetic conditions. Therefore, control interventions often remain at the level of superficial parameter compensation, failing to achieve systematic synergistic optimization between reagent efficacy and process status. Summary of the Invention

[0005] The purpose of this invention is to provide a monitoring system and process for detecting the dosage-effect relationship of phosphate rock flotation reagents, so as to solve the problems mentioned in the background art.

[0006] To solve the above-mentioned technical problems, the present invention is achieved through the following technical solution:

[0007] This invention relates to a monitoring system for detecting the dosage-effect relationship of reagents in phosphate rock flotation. The system comprises a multi-level influence network construction module, a dynamic relationship mapping module, a collaborative evolution driving module, an anomaly capture module, a decision flow activation module, and a collaborative control command integration module. The multi-level influence network construction module establishes a multi-level influence network for the phosphate rock flotation process, which is composed of mineral phase nodes, reagent action units, and flotation environmental factors. The dynamic relationship mapping module establishes progressive action chains between different mineral phase nodes based on the physical flow of the flotation process, and establishes a dynamic mapping relationship between reagent action units and progressive action chains. The collaborative evolution driving module injects process parameter streams into the flotation production line in real time and drives the multi-level influence network to undergo collaborative evolution through these process parameter streams.

[0008] The anomaly capture module is used to capture the sluggish response of reagent action units and the migration and blockage of mineral phase nodes during the co-evolution process; the decision flow activation module is used to activate a preset compensation decision flow to generate real-time correction instructions for reagent parameters in response to the sluggish response of reagent action units, and to activate a preset reconstruction decision flow to generate topology adjustment instructions for the action chain in response to the migration and blockage of mineral phase nodes; the co-control instruction integration module is used to integrate the real-time correction instructions for reagent parameters and the topology adjustment instructions for the action chain into a co-control instruction set, and send the co-control instruction set to the flotation production execution end.

[0009] A monitoring process for detecting the dose-effect relationship of phosphate rock flotation reagents, wherein the process is an implementation process of the aforementioned monitoring system for detecting the dose-effect relationship of phosphate rock flotation reagents, and the steps for establishing a multi-level influence network for the phosphate rock flotation process in the multi-level influence network construction module are as follows:

[0010] A1. Set the constituent elements of mineral phase nodes. The constituent elements include the composition of the minerals entering the ore, the degree of liberation of the target minerals, the activity of gangue minerals, and the surface potential of minerals.

[0011] A2. Set the constituent elements of the drug action unit, which include the molecular structure of the collector, the functional group of the inhibitor, the interfacial tension parameter of the foaming agent, the drug addition rate profile and the cumulative amount of drug consumption.

[0012] A3. Set the constituent elements of the flotation environment factors, including pulp pH trajectory, pulp temperature fluctuation curve, flotation machine aeration intensity distribution, and time series data of froth layer thickness;

[0013] A4. Construct hierarchical transition rules between mineral phase nodes. The hierarchical transition rules are defined as follows.

[0014] A41. When the degree of dissociation of the target mineral is higher than the set degree of dissociation threshold and the surface potential of the mineral is within the preset potential adsorption window, the mineral phase node is allowed to jump to the concentrate enrichment level.

[0015] A42. When the activity of gangue minerals is lower than the set activity inhibition threshold and the surface potential of the minerals deviates from the potential adsorption window, the mineral phase nodes are forced to migrate to the tailings waste layer.

[0016] A5. Construct a mechanism for evaluating the efficacy of drug action units on a progressive chain of action, specifically as follows:

[0017] A51. In the dynamic mapping relationship, an efficacy evaluation operator is embedded. The efficacy evaluation operator calculates the efficacy coefficient of the agent action unit in real time based on the ratio between the cumulative amount of agent consumption and the incremental recovery of the target mineral.

[0018] A52. When the efficacy coefficient is lower than the preset efficacy warning line, mark the corresponding drug action unit with efficacy decay.

[0019] Furthermore, the specific steps of the collaborative evolution driving module for driving the multi-level influence network through process parameter flow to perform collaborative evolution are as follows:

[0020] B1. Continuously acquire process parameter streams from the distributed sensor network on the flotation production line. The process parameter streams include real-time slurry grade spectrum, dosing valve opening timing signals, flotation cell liquid level change sequence, and foam image feature stream.

[0021] B2. Inject the real-time slurry grade spectrum into the corresponding mineral phase node and update the mineral composition and surface potential data in the mineral phase node.

[0022] B3. The timing signal of the dosing valve opening is correlated and compared with the drug addition rate profile. If there is a continuous deviation between the opening signal and the rate profile, the rate calibration process of the drug action unit is triggered to update the drug addition rate profile in the drug action unit.

[0023] B4. The sequence of changes in the liquid level in the flotation cell is fused with the time series data of the thickness of the foam layer to generate a slurry fluid stability index. The slurry fluid stability index is then used to correct the aeration intensity distribution parameter in the flotation environmental factors.

[0024] B5. Based on the updated agent addition rate profile and the corrected aeration intensity distribution parameters, recalculate the effectiveness coefficient of the agent action unit, and adjust the connection strength of the dynamic mapping relationship according to the latest effectiveness coefficient.

[0025] B6. Based on the updated mineral composition and surface potential data, re-evaluate the triggering conditions of the hierarchical transition rules and simulate the real-time migration path of mineral phase nodes in the multi-level influence network.

[0026] Furthermore, the process of the anomaly detection module is as follows:

[0027] C1. Establish periodic monitoring windows during the collaborative evolution of multi-level influence networks;

[0028] C2. Within each periodic monitoring window, perform response hysteresis detection, specifically as follows:

[0029] C21. Track the slope of the change in the efficacy coefficient of the drug action unit over time;

[0030] C22. When the slope of change changes from positive to negative, and the duration of the negative change exceeds the preset hysteresis judgment time, and the value of the efficacy coefficient is lower than the efficacy warning line, it is determined that the response hysteresis phenomenon of the drug action unit has occurred.

[0031] C23. Record the identifier of the drug action unit, the time point when the slope of change turns negative, and the current value of the action efficacy coefficient when the drug action unit response hysteresis occurs, to form a response hysteresis record;

[0032] C3. Within each periodic monitoring window, synchronously perform migration obstruction detection, specifically as follows:

[0033] C31. Track the dwell time of mineral phase nodes on the simulated real-time migration path;

[0034] C32. When the dwell time of a certain mineral phase node at a non-target level exceeds the preset migration timeout threshold, it is determined that a mineral phase node migration blockage phenomenon has occurred.

[0035] C33. Record the identifier of the mineral phase node, its non-target level, and its current dwell time when the mineral phase node migration blockage phenomenon occurs, forming a migration blockage record.

[0036] Furthermore, the step of activating the preset compensation decision flow in the decision flow activation module to generate an instantaneous correction instruction for the drug parameters is as follows:

[0037] D1. Retrieve the response hysteresis records and analyze the identified drug action units;

[0038] D2. Query the drug addition rate profile and cumulative drug consumption of the drug action unit currently bound;

[0039] D3. Start the compensation decision flow, which includes the following parallel processing threads.

[0040] D31, Incremental Compensation Thread, calculates the suggested compensation increment of the drug addition rate based on the negative degree of the change slope, and generates a rate increase instruction containing the drug action unit identifier and the suggested compensation increment.

[0041] D32, Timing Optimization Thread, analyzes the mapping position of the reagent action unit on the progressive action chain, combines the pulp fluid stability index in the current flotation environment factors, recalculates the optimal timing of reagent action, and generates timing adjustment instructions;

[0042] D33. Associate the wake-up thread, retrieve other drug action units that have a strong connection with the drug action unit that has experienced response delay, send warning signals to these associated units, and request their current action efficacy coefficient data;

[0043] D4. Collection rate increase command, timing adjustment command, and effect coefficient data returned from the associated unit;

[0044] D5. If the efficacy coefficient data returned by the associated unit is also generally low, then the collaborative compensation amount is superimposed on the rate increase instruction to form the final version of the real-time correction instruction for the drug parameters.

[0045] D6. If the efficacy coefficient data returned by the associated unit is normal, the rate increase instruction and timing adjustment instruction are directly output as immediate correction instructions for the drug parameters.

[0046] Furthermore, the steps in the decision flow activation module for activating the preset reconstructed decision flow to generate the topology adjustment instruction for the action chain are as follows:

[0047] E1. Retrieve migration blocking records and parse the mineral phase nodes identified therein and their non-target levels;

[0048] E2. Obtain the current constituent element data of mineral phase nodes, and focus on analyzing the liberation degree of target minerals and the activity of gangue minerals;

[0049] E3. Initiate the refactoring decision flow. The refactoring decision flow will perform the following diagnostic and refactoring steps.

[0050] E31. Precursor link diagnosis: trace back the progressive action chain that caused the current mineral phase node to migrate and block, analyze the hierarchical transition state of all preceding mineral phase nodes on the chain, locate the node that first showed abnormal transition, and mark it as the blocking source node.

[0051] E32. Environmental factor interference assessment: Extract data of flotation environmental factors in the current time period and assess whether the pulp acidity and alkalinity trajectory and pulp temperature fluctuation curve deviate from the optimal environmental range required for mineral phase node level transition.

[0052] E33, Agent Action Mapping Review: Review the dynamic mapping relationship of all agent action units associated with the current mineral phase node, and check for mapping relationships with decaying action efficiency coefficients or weak connection strength.

[0053] E4. Based on the diagnostic results, generate topology adjustment instructions for the action chain. The topology adjustment instructions include at least one of the following operations.

[0054] E41. If the blocking source node is clear, generate an instruction to insert a reinforced preprocessing node in front of the blocking source node in the multi-level influence network and configure a new agent action unit mapping for it.

[0055] E42. If the environmental factor interference assessment confirms that environmental deviation is the main cause, then an instruction is generated to adjust the topology of the progressive action chain and add a parallel environmental buffer path so that the mineral phase node can be temporarily bypassed.

[0056] E43. If the reagent action mapping review finds that a specific mapping relationship is invalid, an instruction is generated to cut off the currently invalid dynamic mapping relationship, and based on the constituent elements of the mineral phase node, a new reagent action unit is rematched from the available reagent library to establish a new dynamic mapping relationship.

[0057] Furthermore, the process of the coordinated control instruction set in the coordinated control instruction integration module includes conflict resolution and sequence optimization steps:

[0058] F1. Establish an instruction conflict detection mechanism to compare the operational objects and expected execution time periods involved in the real-time correction instructions for drug parameters and the topology adjustment instructions for the action chain.

[0059] F2. When a command conflict is detected, the conflict resolution rule base is activated. The conflict resolution rule base contains the following rules:

[0060] F21. Resource Exclusivity Rule: If two instructions compete for control of the flotation machine at the same physical dosing point or within the same time period, the topology adjustment instruction of the action chain shall be executed first, and the immediate correction instruction of the reagent parameters shall be suspended and marked with a delayed execution label.

[0061] F22. Logical Dependency Rule: If the execution of the topology adjustment instruction in the action chain is a prerequisite for the immediate correction instruction of the drug parameter to take effect, then the immediate correction instruction of the drug parameter shall be executed after the topology adjustment instruction is completed, and a logical connection mark shall be established between the two.

[0062] F3. Optimize the execution sequence of instructions that do not conflict or have their conflicts resolved, specifically as follows:

[0063] F31. Based on the physical sequence of the mineral phase nodes processed by the instructions in the flotation process, perform preliminary sorting of the instructions;

[0064] F32. For multiple instructions that process the same mineral phase node or closely related nodes, calculate the expected convergence speed of the network state after execution, and prioritize the instructions with the faster expected convergence speed.

[0065] F4. Based on the equipment readiness status and process switching costs at the flotation production execution end, fine-tune the instruction execution sequence to form the final coordinated control instruction set;

[0066] F5. Add execution condition constraints and result feedback requirements to each instruction in the coordinated control instruction set. The execution condition constraints include the required flotation environmental factor threshold range, and the result feedback requirements include the mineral phase node migration state or reagent action unit efficiency coefficient change that needs to be monitored.

[0067] Furthermore, the step of calculating the efficacy coefficient of the drug action unit in real time is as follows:

[0068] G1. The cumulative amount of drug consumption corresponding to the drug action unit within the set evaluation period;

[0069] G2. Within the same evaluation period, calculate the target mineral recovery increment by comparing the target mineral content before and after the mineral phase node transition.

[0070] G3. Calculate the ratio of the target mineral recovery increment to the cumulative amount of reagent consumption to obtain the original efficacy ratio;

[0071] G4. Introduce a correction factor to standardize the original efficacy ratio, obtaining the standardized efficacy coefficient. The correction factor is obtained by:

[0072] G41. Obtain the pulp fluid stability index in the current flotation environment factors. When the pulp fluid stability index is lower than the stability threshold, adjust the correction factor value.

[0073] G42. Obtain the target mineral liberation degree of the current mineral phase node. When the liberation degree is lower than the standard liberation degree, adjust the correction factor value.

[0074] G43. Multiply the original efficacy ratio by the correction factor to obtain the final efficacy coefficient.

[0075] Furthermore, the coordinated control command integration module also includes a system feedback learning and multi-level influence network self-updating process:

[0076] H1. While executing the coordinated control instruction set at the flotation production execution end, the feedback data capture thread is started simultaneously;

[0077] H2. The feedback data capture thread collects the feedback process parameter stream generated by the execution of instructions in real time. The feedback process parameter stream includes the real-time changes in the slurry grade spectrum after execution, the response curve of the dosing valve opening adjustment, and signs of accelerated or unblocked migration of mineral phase nodes.

[0078] H3. Feedback process parameters are injected back into the multi-level influence network;

[0079] H4. Based on the feedback process parameter flow from the reinjection, trigger the self-update of the multi-level influence network. The self-update content includes:

[0080] H41. Update the efficacy coefficient of the relevant drug action unit and reassess whether it is still in a state of efficacy decay;

[0081] H42. Recalculate the hierarchical transition probability of the mineral phase nodes affected by the instruction and adjust their expected migration paths on the progressive action chain.

[0082] H43. Evaluate the connection strength of the newly established or adjusted dynamic mapping relationship, and strengthen or weaken it based on feedback data;

[0083] H5. Store the execution records of this coordinated control instruction set, the key features of the feedback process parameter flow, and the self-updating results of the multi-level influence network into the case library, which will be used to optimize the activation logic of subsequent compensation decision flow and reconstruction decision flow.

[0084] Furthermore, the steps for prioritizing instructions with faster expected convergence speed are as follows:

[0085] I1. Establish a network state convergence speed calculation module and perform the following operations.

[0086] I11. Obtain multiple instructions to be sorted, a real-time snapshot of the current multi-level influence network, and the current data of flotation environment factors;

[0087] I12. Construct a convergence rate estimator, which performs simulations for each instruction to be sorted:

[0088] I13. Virtually inject the instructions to be sorted into the real-time status snapshot of the current multi-level influence network;

[0089] I14. Based on the instruction type and parameters, simulate the impact of the execution of the instruction to be sorted on the trigger probability of the hierarchical transition rules of the target mineral phase node or associated node group, and generate the predicted node migration acceleration.

[0090] I15. Based on the instruction type and parameters, simulate the execution of the instructions to be ordered, and generate the predicted efficacy recovery rate based on the effect on the efficacy coefficient of the associated drug action unit.

[0091] I16. The predicted node migration acceleration and the predicted effectiveness recovery speed are weighted and fused to generate a quantitative value of the expected network state convergence speed of the instructions to be sorted.

[0092] I2. Establish an instruction sorter. The instruction sorter performs the following operations.

[0093] I21. Receive all the instructions to be sorted output by the convergence speed evaluator and their corresponding quantified values of the expected network state convergence speed.

[0094] I22. Sort all the instructions to be sorted in descending order according to their corresponding quantified values of the expected network state convergence speed.

[0095] I23. Output the arranged instruction sequence as the optimized execution sequence.

[0096] The present invention has the following beneficial effects:

[0097] 1. By establishing a progressive action chain that strictly corresponds to the physical flow direction of the flotation process and dynamically mapping the reagent action units to specific links of the chain, the system constructs a digital twin that runs through the entire process. The process parameter flow injected in real time drives the synchronous evolution of the mineral phase nodes and reagent action units on the chain, making the internal state of the system keep in sync with the actual dynamic process of the production line. It transforms the traditional black-box or static correlation model into a transparent and concrete dynamic process mirror. This can reflect in real time and continuously the dynamic migration trajectories of each type of mineral phase and the application and attenuation of the reagent effect at each step during the entire process from raw ore to concentrate, providing unprecedented process visualization and quantification capabilities for understanding the dose-effect relationship.

[0098] 2. By defining and capturing two types of deep abnormal phenomena based on process mechanisms, namely, the response hysteresis of reagent action units and the migration blockage of mineral phase nodes, the system realizes the fine diagnosis of fault modes. The response hysteresis phenomenon accurately locates the specific link where the reagent efficiency fails to meet the expectation, going beyond the simple judgment of insufficient reagent dosage; the migration blockage phenomenon reveals the kinetic bottleneck or path interference inside the process. For these two different types of abnormalities, the system activates the preset compensation decision flow and reconstruction decision flow respectively. The compensation decision flow directly generates correction instructions for the reagent parameters of specific links, while the reconstruction decision flow generates adjustment instructions for the topological structure of the action chain. This precise matching of abnormal patterns and decision logics upgrades the regulation from the rough correction of a single parameter to the coordinated and hierarchical intervention of parameter efficiency and process structure, improving the accuracy and systematicness of the regulation.

[0099] Of course, it is not necessary for any product implementing the present invention to achieve all the above-mentioned advantages simultaneously. Description of the Drawings

[0100] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the 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.

[0101] Figure 1 This is a timing diagram of the monitoring system for detecting the dose-effect relationship of phosphate rock flotation reagents according to the present invention;

[0102] Figure 2 A flowchart for constructing a multi-level influence network;

[0103] Figure 3 A flowchart illustrating the process parameter flow-driven collaborative evolution.

[0104] Figure 4 A heatmap showing the correlation between reagent efficacy coefficient and flotation environmental factors;

[0105] Figure 5 A heat map showing the probability of mineral phase node transitions in a multi-stage influence network for phosphate rock flotation. Detailed Implementation

[0106] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0107] Please see Figure 1-5As shown, this invention is a monitoring system for detecting the dosage-effect relationship of phosphate rock flotation reagents. It includes a multi-level influence network construction module, a dynamic relationship mapping module, a collaborative evolution driving module, an anomaly capture module, a decision flow activation module, and a collaborative control command integration module. The multi-level influence network construction module is used to establish a multi-level influence network for the phosphate rock flotation process. This multi-level influence network is composed of mineral phase nodes, reagent action units, and flotation environmental factors. The dynamic relationship mapping module establishes progressive action chains between different mineral phase nodes based on the physical flow direction of the flotation process, and establishes a dynamic mapping relationship between reagent action units and progressive action chains. The collaborative evolution driving module is used to inject real-time data into the flotation production process. The process parameter flow is generated online, and a multi-level influence network is driven to evolve collaboratively through the process parameter flow. The anomaly capture module is used to capture the response lag phenomenon of reagent action unit and the migration blockage phenomenon of mineral phase nodes during the collaborative evolution process. The decision flow activation module is used to activate a preset compensation decision flow to generate an instant correction instruction for reagent parameters in response to the response lag phenomenon of reagent action unit, and to activate a preset reconstruction decision flow to generate a topology adjustment instruction for the action chain in response to the migration blockage phenomenon of mineral phase nodes. The collaborative control instruction integration module is used to integrate the instant correction instruction for reagent parameters and the topology adjustment instruction for the action chain into a collaborative control instruction set, and send the collaborative control instruction set to the flotation production execution terminal.

[0108] A monitoring process for detecting the dose-effect relationship of phosphate rock flotation reagents, wherein the process is an implementation process of the aforementioned monitoring system for detecting the dose-effect relationship of phosphate rock flotation reagents, and the steps for establishing a multi-level influence network for the phosphate rock flotation process in the multi-level influence network construction module are as follows:

[0109] A1. Set the constituent elements of mineral phase nodes. The constituent elements include the composition of the minerals entering the ore, the degree of liberation of the target minerals, the activity of gangue minerals, and the surface potential of minerals.

[0110] A2. Set the constituent elements of the drug action unit, which include the molecular structure of the collector, the functional group of the inhibitor, the interfacial tension parameter of the foaming agent, the drug addition rate profile and the cumulative amount of drug consumption.

[0111] A3. Set the constituent elements of the flotation environment factors, including pulp pH trajectory, pulp temperature fluctuation curve, flotation machine aeration intensity distribution, and time series data of froth layer thickness;

[0112] A4. Construct hierarchical transition rules between mineral phase nodes. The hierarchical transition rules are defined as follows.

[0113] A41. When the degree of dissociation of the target mineral is higher than the set degree of dissociation threshold and the surface potential of the mineral is within the preset potential adsorption window, the mineral phase node is allowed to jump to the concentrate enrichment level.

[0114] A42. When the activity of gangue minerals is lower than the set activity inhibition threshold and the surface potential of the minerals deviates from the potential adsorption window, the mineral phase nodes are forced to migrate to the tailings waste layer.

[0115] A5. Construct a mechanism for evaluating the efficacy of drug action units on a progressive chain of action, specifically as follows:

[0116] A51. In the dynamic mapping relationship, an efficacy evaluation operator is embedded. The efficacy evaluation operator calculates the efficacy coefficient of the agent action unit in real time based on the ratio between the cumulative amount of agent consumption and the incremental recovery of the target mineral.

[0117] A52. When the efficacy coefficient is lower than the preset efficacy warning line, mark the corresponding drug action unit with efficacy decay.

[0118] The specific steps of the collaborative evolution driving module to drive the multi-level influence network through process parameter flow for collaborative evolution are as follows:

[0119] B1. Continuously acquire process parameter streams from the distributed sensor network on the flotation production line. The process parameter streams include real-time slurry grade spectrum, dosing valve opening timing signals, flotation cell liquid level change sequence, and foam image feature stream.

[0120] B2. Inject the real-time slurry grade spectrum into the corresponding mineral phase node and update the mineral composition and surface potential data in the mineral phase node.

[0121] B3. The timing signal of the dosing valve opening is correlated and compared with the drug addition rate profile. If there is a continuous deviation between the opening signal and the rate profile, the rate calibration process of the drug action unit is triggered to update the drug addition rate profile in the drug action unit.

[0122] B4. The sequence of changes in the liquid level in the flotation cell is fused with the time series data of the thickness of the foam layer to generate a slurry fluid stability index. The slurry fluid stability index is then used to correct the aeration intensity distribution parameter in the flotation environmental factors.

[0123] B5. Based on the updated agent addition rate profile and the corrected aeration intensity distribution parameters, recalculate the effectiveness coefficient of the agent action unit, and adjust the connection strength of the dynamic mapping relationship according to the latest effectiveness coefficient.

[0124] B6. Based on the updated mineral composition and surface potential data, re-evaluate the triggering conditions of the hierarchical transition rules and simulate the real-time migration path of mineral phase nodes in the multi-level influence network.

[0125] The process of the anomaly detection module is as follows:

[0126] C1. Establish periodic monitoring windows during the collaborative evolution of multi-level influence networks;

[0127] C2. Within each periodic monitoring window, perform response hysteresis detection, specifically as follows:

[0128] C21. Track the slope of the change in the efficacy coefficient of the drug action unit over time;

[0129] C22. When the slope of change changes from positive to negative, and the duration of the negative change exceeds the preset hysteresis judgment time, and the value of the efficacy coefficient is lower than the efficacy warning line, it is determined that the response hysteresis phenomenon of the drug action unit has occurred.

[0130] C23. Record the identifier of the drug action unit, the time point when the slope of change turns negative, and the current value of the action efficacy coefficient when the drug action unit response hysteresis occurs, to form a response hysteresis record;

[0131] C3. Within each periodic monitoring window, synchronously perform migration obstruction detection, specifically as follows:

[0132] C31. Track the dwell time of mineral phase nodes on the simulated real-time migration path;

[0133] C32. When the dwell time of a certain mineral phase node at a non-target level exceeds the preset migration timeout threshold, it is determined that a mineral phase node migration blockage phenomenon has occurred.

[0134] C33. Record the identifier of the mineral phase node, its non-target level, and its current dwell time when the mineral phase node migration blockage phenomenon occurs, forming a migration blockage record.

[0135] The steps in the decision flow activation module to activate the preset compensation decision flow to generate immediate correction instructions for the drug parameters are as follows:

[0136] D1. Retrieve the response hysteresis records and analyze the identified drug action units;

[0137] D2. Query the drug addition rate profile and cumulative drug consumption of the drug action unit currently bound;

[0138] D3. Start the compensation decision flow, which includes the following parallel processing threads.

[0139] D31, Incremental Compensation Thread, calculates the suggested compensation increment of the drug addition rate based on the negative degree of the change slope, and generates a rate increase instruction containing the drug action unit identifier and the suggested compensation increment.

[0140] D32, Timing Optimization Thread, analyzes the mapping position of the reagent action unit on the progressive action chain, combines the pulp fluid stability index in the current flotation environment factors, recalculates the optimal timing of reagent action, and generates timing adjustment instructions;

[0141] D33. Associate the wake-up thread, retrieve other drug action units that have a strong connection with the drug action unit that has experienced response delay, send warning signals to these associated units, and request their current action efficacy coefficient data;

[0142] D4. Collection rate increase command, timing adjustment command, and effect coefficient data returned from the associated unit;

[0143] D5. If the efficacy coefficient data returned by the associated unit is also generally low, then the collaborative compensation amount is superimposed on the rate increase instruction to form the final version of the real-time correction instruction for the drug parameters.

[0144] D6. If the efficacy coefficient data returned by the associated unit is normal, the rate increase instruction and timing adjustment instruction are directly output as immediate correction instructions for the drug parameters.

[0145] The steps in the decision flow activation module to activate the preset reconstruction decision flow to generate the topology adjustment instructions for the action chain are as follows:

[0146] E1. Retrieve migration blocking records and parse the mineral phase nodes identified therein and their non-target levels;

[0147] E2. Obtain the current constituent element data of mineral phase nodes, and focus on analyzing the liberation degree of target minerals and the activity of gangue minerals;

[0148] E3. Initiate the refactoring decision flow. The refactoring decision flow will perform the following diagnostic and refactoring steps.

[0149] E31. Precursor link diagnosis: trace back the progressive action chain that caused the current mineral phase node to migrate and block, analyze the hierarchical transition state of all preceding mineral phase nodes on the chain, locate the node that first showed abnormal transition, and mark it as the blocking source node.

[0150] E32. Environmental factor interference assessment: Extract data of flotation environmental factors in the current time period and assess whether the pulp acidity and alkalinity trajectory and pulp temperature fluctuation curve deviate from the optimal environmental range required for mineral phase node level transition.

[0151] E33, Agent Action Mapping Review: Review the dynamic mapping relationship of all agent action units associated with the current mineral phase node, and check for mapping relationships with decaying action efficiency coefficients or weak connection strength.

[0152] E4. Based on the diagnostic results, generate topology adjustment instructions for the action chain. The topology adjustment instructions include at least one of the following operations.

[0153] E41. If the blocking source node is clear, generate an instruction to insert a reinforced preprocessing node in front of the blocking source node in the multi-level influence network and configure a new agent action unit mapping for it.

[0154] E42. If the environmental factor interference assessment confirms that environmental deviation is the main cause, then an instruction is generated to adjust the topology of the progressive action chain and add a parallel environmental buffer path so that the mineral phase node can be temporarily bypassed.

[0155] E43. If the reagent action mapping review finds that a specific mapping relationship is invalid, an instruction is generated to cut off the currently invalid dynamic mapping relationship, and based on the constituent elements of the mineral phase node, a new reagent action unit is rematched from the available reagent library to establish a new dynamic mapping relationship.

[0156] The process of the coordinated control instruction set in the coordinated control instruction integration module includes conflict resolution and sequence optimization steps:

[0157] F1. Establish an instruction conflict detection mechanism to compare the operational objects and expected execution time periods involved in the real-time correction instructions for drug parameters and the topology adjustment instructions for the action chain.

[0158] F2. When a command conflict is detected, the conflict resolution rule base is activated. The conflict resolution rule base contains the following rules:

[0159] F21. Resource Exclusivity Rule: If two instructions compete for control of the flotation machine at the same physical dosing point or within the same time period, the topology adjustment instruction of the action chain shall be executed first, and the immediate correction instruction of the reagent parameters shall be suspended and marked with a delayed execution label.

[0160] F22. Logical Dependency Rule: If the execution of the topology adjustment instruction in the action chain is a prerequisite for the immediate correction instruction of the drug parameter to take effect, then the immediate correction instruction of the drug parameter shall be executed after the topology adjustment instruction is completed, and a logical connection mark shall be established between the two.

[0161] F3. Optimize the execution sequence of instructions that do not conflict or have their conflicts resolved, specifically as follows:

[0162] F31. Based on the physical sequence of the mineral phase nodes processed by the instructions in the flotation process, perform preliminary sorting of the instructions;

[0163] F32. For multiple instructions that process the same mineral phase node or closely related nodes, calculate the expected convergence speed of the network state after execution, and prioritize the instructions with the faster expected convergence speed.

[0164] F4. Based on the equipment readiness status and process switching costs at the flotation production execution end, fine-tune the instruction execution sequence to form the final coordinated control instruction set;

[0165] F5. Add execution condition constraints and result feedback requirements to each instruction in the coordinated control instruction set. The execution condition constraints include the required flotation environmental factor threshold range, and the result feedback requirements include the mineral phase node migration state or reagent action unit efficiency coefficient change that needs to be monitored.

[0166] The steps for real-time calculation of the efficacy coefficient of the drug action unit are as follows:

[0167] G1. The cumulative amount of drug consumption corresponding to the drug action unit within the set evaluation period;

[0168] G2. Within the same evaluation period, calculate the target mineral recovery increment by comparing the target mineral content before and after the mineral phase node transition.

[0169] G3. Calculate the ratio of the target mineral recovery increment to the cumulative amount of reagent consumption to obtain the original efficacy ratio;

[0170] G4. Introduce a correction factor to standardize the original efficacy ratio, obtaining the standardized efficacy coefficient. The correction factor is obtained by:

[0171] G41. Obtain the pulp fluid stability index in the current flotation environment factors. When the pulp fluid stability index is lower than the stability threshold, adjust the correction factor value.

[0172] G42. Obtain the target mineral liberation degree of the current mineral phase node. When the liberation degree is lower than the standard liberation degree, adjust the correction factor value.

[0173] G43. Multiply the original efficacy ratio by the correction factor to obtain the final efficacy coefficient.

[0174] The coordinated control command integration module also includes a system feedback learning and multi-level influence network self-updating process:

[0175] H1. While executing the coordinated control instruction set at the flotation production execution end, the feedback data capture thread is started simultaneously;

[0176] H2. The feedback data capture thread collects the feedback process parameter stream generated by the execution of instructions in real time. The feedback process parameter stream includes the real-time changes in the slurry grade spectrum after execution, the response curve of the dosing valve opening adjustment, and signs of accelerated or unblocked migration of mineral phase nodes.

[0177] H3. Feedback process parameters are injected back into the multi-level influence network;

[0178] H4. Based on the feedback process parameter flow from the reinjection, trigger the self-update of the multi-level influence network. The self-update content includes:

[0179] H41. Update the efficacy coefficient of the relevant drug action unit and reassess whether it is still in a state of efficacy decay;

[0180] H42. Recalculate the hierarchical transition probability of the mineral phase nodes affected by the instruction and adjust their expected migration paths on the progressive action chain.

[0181] H43. Evaluate the connection strength of the newly established or adjusted dynamic mapping relationship, and strengthen or weaken it based on feedback data;

[0182] H5. Store the execution records of this coordinated control instruction set, the key features of the feedback process parameter flow, and the self-updating results of the multi-level influence network into the case library, which will be used to optimize the activation logic of subsequent compensation decision flow and reconstruction decision flow.

[0183] The steps for prioritizing instructions with faster expected convergence speed are as follows:

[0184] I1. Establish a network state convergence speed calculation module and perform the following operations.

[0185] I11. Obtain multiple instructions to be sorted, a real-time snapshot of the current multi-level influence network, and the current data of flotation environment factors;

[0186] I12. Construct a convergence rate estimator, which performs simulations for each instruction to be sorted:

[0187] I13. Virtually inject the instructions to be sorted into the real-time status snapshot of the current multi-level influence network;

[0188] I14. Based on the instruction type and parameters, simulate the impact of the execution of the instruction to be sorted on the trigger probability of the hierarchical transition rules of the target mineral phase node or associated node group, and generate the predicted node migration acceleration.

[0189] I15. Based on the instruction type and parameters, simulate the execution of the instructions to be ordered, and generate the predicted efficacy recovery rate based on the effect on the efficacy coefficient of the associated drug action unit.

[0190] I16. The predicted node migration acceleration and the predicted effectiveness recovery speed are weighted and fused to generate a quantitative value of the expected network state convergence speed of the instructions to be sorted.

[0191] I2. Establish an instruction sorter. The instruction sorter performs the following operations.

[0192] I21. Receive all unsorted instructions and their corresponding expected network state convergence speed quantization values ​​output by the convergence speed estimator.

[0193] I22. Sort all instructions to be sorted in descending order according to their corresponding expected network state convergence speed quantization values.

[0194] I23. Output the arranged instruction sequence as the optimized execution sequence.

[0195] Please see Figure 1 As shown, the multi-level influence network construction module is responsible for establishing a multi-level influence network for the phosphate rock flotation process. This network consists of mineral phase nodes characterizing mineral state and properties, reagent action units characterizing reagent performance and additive characteristics, and flotation environmental factors characterizing flotation conditions. The dynamic relationship mapping module establishes a progressive action chain between different mineral phase nodes based on the physical flow of the flotation process to simulate the mineral processing flow. This module establishes a dynamic mapping relationship between reagent action units and specific links in the progressive action chain to characterize the influence of reagents on the mineral processing process. The co-evolution driving module receives process parameter streams from the flotation production line in real time and injects this real-time data into the multi-level influence network, driving the dynamic updating and co-evolution of the states of each component in the network. During the co-evolution process, the anomaly capture module continuously monitors the network state, specifically capturing lag in reagent action unit response and mineral phase node migration blockage. Once an anomaly is detected, the decision flow activation module is triggered: for sluggish response of reagent action units, this module activates a preset compensation decision flow to generate immediate correction instructions for reagent parameters; for mineral phase node migration blockage, it activates a preset reconstruction decision flow to generate topology adjustment instructions for adjusting the action chain structure. The collaborative control instruction integration module is responsible for integrating the generated immediate reagent parameter correction instructions with the action chain topology adjustment instructions to form a coordinated and consistent collaborative control instruction set, which is then distributed to the flotation production execution end to guide the adjustment of production parameters.

[0196] In one embodiment of the present invention, see [reference] Figure 2 In practical implementation, when establishing a multi-level influence network for the phosphate rock flotation process using the multi-level influence network construction module, the first step is:

[0197] A1. Set the constituent elements of each entity unit in the network. The constituent elements of the mineral phase node are set to include the composition of the ore-bearing mineral, the degree of liberation of the target mineral, the activity of the gangue mineral, and the surface potential of the mineral. The composition of the ore-bearing mineral is expressed as the mineral type and mass percentage. The degree of liberation of the target mineral is characterized by the numerical value obtained by the mineral liberation analyzer. The activity of the gangue mineral is quantitatively scored according to its possible reaction with the reagent. The surface potential of the mineral is measured using the zeta potential.

[0198] A2. The constituent elements of the agent action unit are set to include the collector molecular structure, inhibitor functional group, foaming agent interfacial tension parameter, agent addition rate profile, and cumulative agent consumption. The collector molecular structure is described by functional group type and carbon chain length. The inhibitor functional group identifies the chemical group that inhibits its action. The foaming agent interfacial tension parameter is the measured dynamic interfacial tension value. The agent addition rate profile is a function curve of time versus addition rate. The cumulative agent consumption is obtained from the metering device of the production line.

[0199] A3. The components of the flotation environmental factors are set to include pulp pH trajectory, pulp temperature fluctuation curve, flotation machine aeration intensity distribution, and time series data of froth layer thickness. The pulp pH trajectory is a sequence of pH value changes over time. The pulp temperature fluctuation curve is a time series signal collected by temperature sensors. The flotation machine aeration intensity distribution describes the aeration volume of different flotation cell units. The time series data of froth layer thickness is extracted by the froth image analysis system.

[0200] A4. Construct hierarchical transition rules between mineral phase nodes. These rules define the permissible and mandatory conditions for mineral phase node state migration. A41. One permissive rule is configured as follows: when the target mineral liberation degree of a mineral phase node is higher than a set liberation degree threshold, and the mineral surface potential of the mineral phase node is within a preset potential adsorption window, the system allows the mineral phase node to transition to the concentrate enrichment level representing the concentrate product. A42. One mandatory rule is configured as follows: when the gangue mineral activity of a mineral phase node is lower than a set activity inhibition threshold, and the mineral surface potential of the mineral phase node deviates from the potential adsorption window, the system forces the mineral phase node to migrate to the tailings waste level representing waste tailings. The liberation degree threshold, potential adsorption window, and activity inhibition threshold are determined based on specific phosphate mining process theoretical values ​​and historical data.

[0201] A5. In specific implementation, construct an evaluation mechanism for the efficacy of drug action units on the progressive action chain;

[0202] A51. This mechanism is achieved by embedding an efficacy evaluation operator in the dynamic mapping relationship between the agent action unit and the progressive action chain. The efficacy evaluation operator calculates the action efficacy coefficient of the agent action unit in real time based on the ratio between the cumulative amount of agent consumption and the incremental amount of target mineral recovery.

[0203] The calculation process of the effectiveness evaluation operator:

[0204] G1. The evaluation is conducted within a set evaluation period. At the beginning of the evaluation period, the efficacy evaluation operator accumulates the cumulative amount of drug consumption corresponding to the drug action unit.

[0205] G2. At the end of the same evaluation period, the effectiveness evaluation operator calculates the target mineral recovery increment by comparing the target mineral content before and after the hierarchical transition of the associated mineral phase nodes.

[0206] G3. It is understandable that the efficacy evaluation operator calculates the ratio of the target mineral recovery increment to the cumulative amount of reagent consumption to obtain the original efficacy ratio.

[0207] G4. Subsequently, the efficacy evaluation operator introduces a correction factor to standardize the original efficacy ratio in order to obtain the standardized efficacy coefficient.

[0208] The process of obtaining the correction factor includes: G41, the effectiveness evaluation operator obtains the pulp fluid stability index in the current flotation environment factors. When the pulp fluid stability index is lower than the preset stability threshold, the effectiveness evaluation operator lowers the correction factor value; G42, the effectiveness evaluation operator simultaneously obtains the target mineral liberation degree of the currently associated mineral phase node. When the target mineral liberation degree is lower than the preset standard liberation degree, the effectiveness evaluation operator lowers the correction factor value; G43, finally, the effectiveness evaluation operator multiplies the original effectiveness ratio by the correction factor to obtain the effectiveness coefficient of the reagent action unit.

[0209] In some embodiments, the efficacy evaluation operator calculates the final efficacy coefficient using the following formula:

[0210]

[0211] Where: symbol Represents the final calculated efficacy coefficient, symbol Represents the incremental recovery of the target mineral calculated during the assessment period, with the symbol... Represents the cumulative amount of pharmaceuticals consumed within the same assessment period, symbol The correction factor represents the stability index of the slurry fluid. and the degree of liberation of the target mineral The function.

[0212] A52. When the calculated efficacy coefficient is lower than the preset efficacy warning line, the system marks the corresponding drug action unit with an efficacy decay indicator.

[0213] In one embodiment of the present invention, see [reference] Figure 3 In practical implementation, the co-evolution driving module drives the multi-level influence network to undergo co-evolution through process parameter flow. The specific steps of the co-evolution driving module are as follows:

[0214] B1. Continuously acquire process parameter streams from the distributed sensor network on the flotation production line. The process parameter streams include real-time slurry grade spectra periodically output by the online grade analyzer, timing signals of dosing valve opening collected by the dosing valve position sensor, flotation cell level change sequences recorded by the flotation cell level gauge, and foam image feature streams captured by the machine vision system.

[0215] B2. In specific implementation, the co-evolution driving module injects the real-time slurry grade spectrum into the corresponding mineral phase node in the multi-level influence network, updates the mineral composition data in the mineral phase node according to the content percentage of various minerals in the real-time slurry grade spectrum, and updates the mineral surface potential data in the mineral phase node according to the electrochemical signal associated with the real-time slurry grade spectrum.

[0216] B3. The co-evolution driving module correlates and compares the timing signal of the dosing valve opening with the drug addition rate profile stored in the drug action unit. The correlation and comparison process calculates the deviation between the timing signal of the dosing valve opening and the drug addition rate profile at the same time. If the deviation value continuously exceeds the preset deviation threshold in multiple consecutive sampling periods, the co-evolution driving module triggers the rate calibration process of the drug action unit. The rate calibration process smooths and corrects the drug addition rate profile according to the actual trend of the timing signal of the dosing valve opening, thereby updating the drug addition rate profile in the drug action unit.

[0217] B4. In some embodiments, the co-evolution driving module fuses the flotation cell level change sequence with the time-series data of foam layer thickness extracted from the foam image feature stream to generate a slurry fluid stability index. The fusion process uses an algorithm function to comprehensively calculate the fluctuation amplitude of the flotation cell level change sequence and the fluctuation frequency of the foam layer thickness time-series data. The co-evolution driving module uses the generated slurry fluid stability index to correct the aeration intensity distribution parameter in the flotation environmental factors. The correction logic is that when the slurry fluid stability index indicates a decrease in stability, the aeration intensity distribution parameter value of the corresponding flotation cell unit in the flotation environmental factors is correspondingly reduced.

[0218] B5. Understandably, the co-evolution driving module, based on the updated agent addition rate profile and the corrected aeration intensity distribution parameters, re-invokes the efficacy evaluation operator to calculate the efficacy coefficients of the relevant agent-acting units. The co-evolution driving module adjusts the connection strength of the dynamic mapping relationship based on the latest efficacy coefficient values. The adjustment strategy is to strengthen the connection strength of the dynamic mapping relationship when the efficacy coefficient increases, and to weaken the connection strength of the dynamic mapping relationship when the efficacy coefficient decreases.

[0219] B6. The co-evolution driving module re-evaluates the triggering conditions of hierarchical transition rules based on updated mineral composition and surface potential data. This re-evaluation process compares the current target mineral liberation degree of the mineral phase node with the liberation degree threshold and the mineral surface potential with the potential adsorption window. Based on the re-evaluation results, the co-evolution driving module simulates the real-time migration path of mineral phase nodes in a multi-level influence network. This simulation reveals the probability and direction of mineral phase nodes migrating from the current level to concentrate enrichment levels or tailings waste levels.

[0220] In one embodiment of the present invention, in a specific implementation, the anomaly capture module captures the response lag of the agent action unit and the mineral phase node migration blockage phenomenon that occur during the cooperative evolution process. The process of the anomaly capture module is as follows:

[0221] C1. Establish a periodic monitoring window during the collaborative evolution of the multi-level influence network. The duration of the periodic monitoring window is configured according to the kinetic characteristics of the flotation process.

[0222] C2. Within each periodic monitoring window, the anomaly capture module performs the response hysteresis detection step as follows: C21. The response hysteresis detection process involves tracking the slope of the change in the efficacy coefficient of a specified drug action unit over time. The slope is obtained by calculating the difference in efficacy coefficients between adjacent monitoring time points. C22. When the slope changes from positive to negative, and the negative duration exceeds the preset hysteresis judgment duration, while the efficacy coefficient value is below the efficacy warning line, the anomaly capture module determines that a drug action unit response hysteresis phenomenon has occurred. C23. The anomaly capture module records the identifier of the drug action unit corresponding to the occurrence of the drug action unit response hysteresis phenomenon, the time point when the slope turns negative, and the current efficacy coefficient value, forming a response hysteresis record.

[0223] C3. In some embodiments, within each periodic monitoring window, the anomaly capture module synchronously performs migration blockage detection. The migration blockage detection process is as follows: C31. Track the dwell time of mineral phase nodes on the real-time migration path simulated by the co-evolution driving module; C32. When the dwell time of a certain mineral phase node in a non-target level exceeds a preset migration timeout threshold, the anomaly capture module determines that a mineral phase node migration blockage phenomenon has occurred. The non-target level refers to the intermediate processing level, which is neither the concentrate enrichment level nor the tailings waste level; C33. The anomaly capture module records the identifier of the mineral phase node corresponding to the occurrence of the mineral phase node migration blockage phenomenon, its non-target level, and its current dwell time, forming a migration blockage record.

[0224] It is understandable that the slope of change can be calculated using the following formula:

[0225]

[0226] Where: symbol Represents a point in time The slope of the calculated effect efficacy coefficient change, sign Represents a point in time The obtained efficacy coefficient value, symbol Represents the previous point in time. The obtained efficacy coefficient value. The anomaly detection module continuously monitors. The sign and value are used to determine whether the slope of change changes from positive to negative and the duration of the negative change.

[0227] In some embodiments, the hysteresis judgment duration can be differentiated based on the normal response time of different reagent action units. For example, for fast-acting frothers, the hysteresis judgment duration is set shorter; for inhibitors requiring slow action, the hysteresis judgment duration is set longer. The migration timeout threshold is set based on the typical residence time range of minerals in each stage of flotation. When the residence time of mineral phase nodes in the roughing stage exceeds the roughing stage migration timeout threshold, the mineral phase node migration blockage phenomenon is triggered. Optionally, the boundary of the periodic monitoring window can be synchronized with the key process switching signal of the flotation production line to achieve more accurate stage performance evaluation. Optionally, in addition to basic information, the generated response hysteresis records and migration blockage records can also include snapshot data of the flotation environmental factors at that time for in-depth analysis.

[0228] In one embodiment of the present invention, the decision flow activation module activates a preset compensation decision flow to generate an instantaneous correction instruction for the drug action unit response hysteresis in response to the following steps:

[0229] D1. The decision flow activation module first retrieves the response hysteresis record generated by the anomaly capture module. The response hysteresis record is then parsed to identify the specific agent action unit, such as the agent action unit identified as "collector PA-1".

[0230] D2. The decision flow activation module queries the currently bound drug addition rate profile and cumulative drug consumption of the "Collector PA-1" drug action unit. The drug addition rate profile is displayed as a preset addition curve, and the cumulative drug consumption is displayed as the cumulative value from the beginning of the month to the present.

[0231] D3. The decision flow activation module initiates the compensation decision flow, which includes three parallel processing threads: an incremental compensation thread, a timing optimization thread, and an associated wake-up thread. D31. The incremental compensation thread calculates the suggested compensation increment for the reagent addition rate based on the negative degree of the change slope in the response hysteresis record. The negative degree is comprehensively evaluated by the absolute value of the change slope and its duration, generating a rate increase instruction that includes the "collector PA-1" reagent action unit identifier and the specific suggested compensation increment. D32. The timing optimization thread analyzes the mapping position of the "collector PA-1" reagent action unit on the progressive action chain. The mapping position indicates that the reagent mainly acts on the coarsening stage. Combining the pulp fluid stability index in the current flotation environment factors, the timing optimization thread recalculates the optimal action timing for "collector PA-1" and generates a timing adjustment instruction to slightly advance the addition point. D33. The associated wake-up thread searches for other drug action units that have a strong connection with the "collector PA-1" drug action unit. The strong connection is defined according to historical action data. The associated wake-up thread sends a warning signal to these associated units and requests their current action efficacy coefficient data.

[0232] D4. In some embodiments, the decision flow activation module collects rate increase instructions, timing adjustment instructions, and effectiveness coefficient data returned from the associated unit.

[0233] D5. The decision flow activation module determines whether the effectiveness coefficient data returned by the associated units are generally low. The basis for this judgment is to compare them with the historical benchmark values ​​of these associated units.

[0234] D6. If the efficacy coefficient data returned by the associated units are also generally low, the decision flow activation module will add a collaborative compensation amount to the compensation increment suggested by the rate increase instruction, forming the final version of the real-time correction instruction for the drug parameters. If the efficacy coefficient data returned by the associated units is normal, the decision flow activation module will directly output the rate increase instruction and the timing adjustment instruction as the real-time correction instruction for the drug parameters. For a comparison of the efficacy coefficient data of the associated units, please refer to Table 1.

[0235] Table 1: Feedback Table of Efficacy Coefficients of Related Drug Action Units

[0236]

[0237] In practical implementation, to address the phenomenon of mineral phase node migration blockage, the decision flow activation module activates a preset reconstruction decision flow to generate topology adjustment instructions for the action chain, as follows:

[0238] E1. The decision flow activation module retrieves the migration blocking record. The migration blocking record is parsed to identify the mineral phase node and its non-target level. For example, the node identified as "mineral phase node P-Ore-7" is blocked at the "coarse selection-sweep selection" level.

[0239] E2. The decision flow activation module obtains the current constituent element data of "mineral phase node P-Ore-7" and focuses on analyzing its target mineral liberation degree and gangue mineral activity. The data shows that its target mineral liberation degree is 65% and its gangue mineral activity score is 75.

[0240] E3. Understandably, the decision flow activation module initiates a reconstructed decision flow, which performs pre-link diagnosis, environmental factor interference assessment, and reagent action mapping review. E31. Pre-link diagnosis traces back the progressive action chain that caused the migration blockage at "mineral phase node P-Ore-7," analyzes the hierarchical transition states of all preceding mineral phase nodes on the chain, locates the earliest abnormal transition node, and marks it as the blockage source node, for example, "mineral phase node P-Ore-5." E32. Environmental factor interference assessment extracts flotation environmental factor data for the current time period, assesses whether the pulp pH trajectory and pulp temperature fluctuation curve deviate from the optimal environmental range required for the hierarchical transition of "mineral phase node P-Ore-7." E33. Reagent action mapping review examines the dynamic mapping relationship of all reagent action units associated with "mineral phase node P-Ore-7," checking for mapping relationships with diminishing effectiveness coefficients or weak connections.

[0241] E4. In some embodiments, a topology adjustment instruction for the action chain is generated based on the diagnostic results. The topology adjustment instruction includes at least one of the following operations: E41. If the preceding link diagnosis clearly identifies "mineral phase node P-Ore-5" as the blocking source node, an instruction is generated to insert a strengthening pretreatment node in front of "mineral phase node P-Ore-5" in the multi-level influence network and configure a new agent action unit mapping for it. E42. If the environmental factor interference assessment confirms that the deviation of slurry pH is the main cause of migration blockage, an instruction is generated to adjust the topology of the progressive action chain, adding a parallel environmental buffer path so that the mineral phase node can temporarily bypass the area affected by pH. E43. If the drug action mapping review finds that the mapping relationship with "Inhibitor IN-3" is invalid, an instruction is generated to sever the current dynamic mapping relationship between "Mineral Phase Node P-Ore-7" and the drug action unit "Inhibitor IN-3", and based on the constituent elements of "Mineral Phase Node P-Ore-7", a new drug action unit is re-matched for it from the available drug library, for example, matched as "Inhibitor IN-3A", and a new dynamic mapping relationship is established.

[0242] Optionally, the formula for calculating the proposed compensation increment can be designed as follows:

[0243]

[0244] Where: symbol The symbol represents the calculated proposed compensation increment. Represents a gain coefficient determined according to the type of drug, symbol The average absolute slope value after the slope of change turns negative, sign... Represents negative duration, symbol This represents the current baseline addition rate in the drug addition rate profile. Optionally, when the associated wake-up thread requests the associated unit's efficacy coefficient data, a response timeout can be set; units that do not respond within the timeout period are treated as having missing data.

[0245] See Figure 4 During the reagent efficacy evaluation phase, the heatmap visually presented the linear correlations between collector PA-1, pH value, temperature, and stability indicators. Specifically, collector PA-1 showed a very strong negative correlation with pH value (correlation coefficient -0.99) and a very strong positive correlation with the stability indicators (correlation coefficient 0.99); pH value showed a completely negative correlation with the stability indicators (correlation coefficient -1.00), reflecting the reverse regulatory effect of flotation environmental factors on reagent efficacy. Temperature showed extremely weak correlations with the other factors (absolute value ≤0.03), indicating that the impact of temperature on reagent efficacy and other environmental factors was negligible during this evaluation phase. At the parameter level, the correlation coefficients ranged from -1.00 to 1.00, and the correlation strength was quantified using a red-white-blue gradient color scheme, with red representing a strong positive correlation, blue representing a strong negative correlation, and white representing no correlation.

[0246] In one embodiment of the present invention, the process by which the collaborative regulation instruction integration module integrates the real-time correction instructions for drug parameters and the topology adjustment instructions for the action chain into a collaborative regulation instruction set includes conflict resolution and sequence optimization steps:

[0247] F1. The collaborative control command integration module establishes a command conflict detection mechanism. The command conflict detection mechanism compares the operation objects and expected execution time periods involved in the real-time correction command of the reagent parameters with the topology adjustment command of the action chain. The operation objects include specific flotation cell numbers, dosing point locations, or reagent action unit identifiers.

[0248] F2. When the instruction conflict detection mechanism detects an instruction conflict, the collaborative control instruction integration module activates the conflict resolution rule base. This rule base includes resource exclusivity rules and logical dependency rules. F21. The resource exclusivity rule stipulates that if two instructions compete for control of the flotation machine at the same physical dosing point or within the same time period, the topology adjustment instruction of the action chain will be executed first, while the immediate correction instruction for reagent parameters will be suspended and marked with a delayed execution tag. F22. The logical dependency rule stipulates that if the execution of the topology adjustment instruction of the action chain is a prerequisite for the immediate correction instruction for reagent parameters to take effect, the immediate correction instruction for reagent parameters will be executed after the topology adjustment instruction is completed, and a logical connection marker will be established between the two instructions.

[0249] F3. In some embodiments, the collaborative control instruction integration module optimizes the execution sequence of instructions that do not conflict or have resolved conflicts. Specifically: F31. The execution sequence optimization process first sorts the instructions according to the physical order of the mineral phase nodes processed by the instructions in the flotation process. Instructions that process mineral phase nodes located at the beginning of the process are arranged at the beginning of the execution sequence. F32. For multiple instructions that process the same mineral phase node or closely related nodes, the collaborative control instruction integration module calculates the expected network state convergence speed after execution and prioritizes instructions with faster expected convergence speeds.

[0250] F4. The collaborative control instruction integration module fine-tunes the instruction execution sequence based on the equipment readiness status and process switching costs at the flotation production execution end, forming the final collaborative control instruction set.

[0251] F5, the collaborative control instruction integration module adds execution condition constraints and result feedback requirements to each instruction in the collaborative control instruction set. The execution condition constraints include the required flotation environmental factor threshold range, and the result feedback requirements include the migration status of mineral phase nodes or the change in the efficiency coefficient of reagent action units that need to be monitored.

[0252] It can be understood that the specific steps for the collaborative control instruction integration module to calculate the expected network state convergence speed are as follows:

[0253] I1. The network state convergence speed calculation module performs the following operations: I11. The network state convergence speed calculation module acquires multiple instructions to be sorted, a real-time snapshot of the current multi-level influence network, and current data of flotation environmental factors. I12. The network state convergence speed calculation module constructs a convergence speed estimator. The convergence speed estimator performs simulation for each instruction to be sorted. I13. During the simulation, the instructions to be sorted are virtually injected into the real-time snapshot of the current multi-level influence network. I14. The convergence speed estimator simulates the impact of the execution of the instructions to be sorted on the trigger probability of hierarchical transition rules of the target mineral phase nodes or associated node groups, based on the type and parameters of the instructions to be sorted, and generates predicted node migration acceleration. I15. The convergence speed estimator simulates the impact of the execution of the instructions to be sorted on the effectiveness coefficient of the associated reagent action units, based on the type and parameters of the instructions to be sorted, and generates predicted effectiveness recovery speed. I16. The convergence speed estimator weights and fuses the predicted node migration acceleration and the predicted effectiveness recovery speed to generate a quantitative value of the expected network state convergence speed of the instructions to be sorted.

[0254] I2. The network state convergence speed calculation module establishes an instruction sorter. The instruction sorter performs the following operations: I21. The instruction sorter receives all instructions to be sorted and their corresponding expected network state convergence speed quantization values ​​output by the convergence speed evaluator. I22. The instruction sorter sorts all instructions to be sorted in descending order according to their corresponding expected network state convergence speed quantization values. I23. The instruction sorter outputs the sorted instruction sequence as the optimized execution sequence.

[0255] In some embodiments, after the coordinated control instruction set is issued to the flotation production execution end, the system initiates a feedback learning and multi-level influence network self-update process:

[0256] H1. While executing the coordinated control instruction set at the flotation production execution end, the system simultaneously starts the feedback data capture thread.

[0257] H2. The feedback data capture thread collects the feedback process parameter stream generated by the execution of instructions in real time. The feedback process parameter stream includes the real-time changes in the slurry grade spectrum after execution, the response curve of the dosing valve opening adjustment, and signs of accelerated or unblocked migration of mineral phase nodes.

[0258] H3, the feedback data capture thread, injects the feedback process parameter stream back into the multi-level influence network.

[0259] H4. Based on the feedback process parameters from the reinjection, a self-updating process of the multi-level influence network is triggered. The self-updating process includes: H41. Updating the effectiveness coefficients of relevant reagent action units and reassessing whether they are still in a state of effectiveness decay. H42. The self-updating process recalculates the hierarchical transition probabilities of mineral phase nodes affected by the command and adjusts their expected migration paths on the progressive action chain. H43. The self-updating process assesses the connection strength of the newly established or adjusted dynamic mapping relationships and strengthens or weakens them based on feedback data.

[0260] H5. The system stores the execution records of this coordinated control instruction set, the key features of the feedback process parameter flow, and the self-updating results of the multi-level influence network into the case library. The stored data is used to optimize the activation logic of subsequent compensation decision flow and reconstruction decision flow.

[0261] See Figure 5 In the self-updating stage of the multi-stage influence network in phosphate rock flotation, the distribution of transition probabilities of mineral phase nodes to different target levels after command execution is presented. Specifically, the figure uses mineral phase nodes as the vertical dimension and concentrate enrichment layer / transition layer / tailings waste layer as the horizontal target levels, with transition probability values ​​ranging from 0.0 to 1.0 corresponding to color gradients (blue to orange-red). In actual analysis, the transition probabilities of mineral phase nodes are calculated based on the feedback process parameter flow after the execution of the coordinated control command: for example, the transition probability of node 1 to the tailings waste layer is 0.64, corresponding to the state where the dissociation degree of its target mineral has not reached the threshold and the surface potential deviates from the adsorption window; the transition probability of node 3 to the concentrate enrichment layer is 0.49, reflecting the characteristic that the activity of its gangue minerals is lower than the inhibition threshold. At the parameter level, the probability values ​​of each node are generated after re-evaluating the triggering conditions of the level transition rules, which can directly reflect the adjustment effect of the command on the migration path of mineral phase nodes.

[0262] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0263] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A monitoring system for detecting the dose-effect relationship of phosphate rock flotation reagents, characterized in that, The system includes a multi-level influence network construction module, a dynamic relationship mapping module, a collaborative evolution driving module, an anomaly capture module, a decision flow activation module, and a collaborative control instruction integration module. The multi-level influence network construction module establishes a multi-level influence network for the phosphate rock flotation process, which is composed of mineral phase nodes, reagent action units, and flotation environmental factors. The dynamic relationship mapping module establishes progressive action chains between different mineral phase nodes based on the physical flow direction of the flotation process, and establishes a dynamic mapping relationship between reagent action units and progressive action chains. The collaborative evolution driving module injects process parameter streams into the flotation production line in real time and drives the multi-level influence network to undergo collaborative evolution through these process parameter streams. The anomaly capture module is used to capture the sluggish response of reagent action units and the migration and blockage of mineral phase nodes during the co-evolution process; the decision flow activation module is used to activate a preset compensation decision flow to generate real-time correction instructions for reagent parameters in response to the sluggish response of reagent action units, and to activate a preset reconstruction decision flow to generate topology adjustment instructions for the action chain in response to the migration and blockage of mineral phase nodes; the co-control instruction integration module is used to integrate the real-time correction instructions for reagent parameters and the topology adjustment instructions for the action chain into a co-control instruction set, and send the co-control instruction set to the flotation production execution end.

2. A monitoring process for detecting the dose-response relationship of phosphate rock flotation reagents, wherein the process is an implementation process of the monitoring system for detecting the dose-response relationship of phosphate rock flotation reagents as described in claim 1, characterized in that, The steps for establishing a multi-level influence network for the phosphate rock flotation process in the multi-level influence network construction module are as follows: A1. Set the constituent elements of mineral phase nodes. The constituent elements include the composition of the minerals entering the ore, the degree of liberation of the target minerals, the activity of gangue minerals, and the surface potential of minerals. A2. Set the constituent elements of the drug action unit, which include the molecular structure of the collector, the functional group of the inhibitor, the interfacial tension parameter of the foaming agent, the drug addition rate profile and the cumulative amount of drug consumption. A3. Set the constituent elements of the flotation environment factors, including pulp pH trajectory, pulp temperature fluctuation curve, flotation machine aeration intensity distribution, and time series data of froth layer thickness; A4. Construct hierarchical transition rules between mineral phase nodes. The hierarchical transition rules are defined as follows. A41. When the degree of dissociation of the target mineral is higher than the set degree of dissociation threshold and the surface potential of the mineral is within the preset potential adsorption window, the mineral phase node is allowed to jump to the concentrate enrichment level. A42. When the activity of gangue minerals is lower than the set activity inhibition threshold and the surface potential of the minerals deviates from the potential adsorption window, the mineral phase nodes are forced to migrate to the tailings waste layer. A5. Construct a mechanism for evaluating the efficacy of drug action units on a progressive chain of action, specifically as follows: A51. In the dynamic mapping relationship, an efficacy evaluation operator is embedded. The efficacy evaluation operator calculates the efficacy coefficient of the agent action unit in real time based on the ratio between the cumulative amount of agent consumption and the incremental recovery of the target mineral. A52. When the efficacy coefficient is lower than the preset efficacy warning line, mark the corresponding drug action unit with efficacy decay.

3. The monitoring process for detecting the dose-effect relationship of phosphate rock flotation reagents according to claim 2, characterized in that, The specific steps of the collaborative evolution driving module to drive the multi-level influence network through process parameter flow for collaborative evolution are as follows: B1. Continuously acquire process parameter streams from the distributed sensor network on the flotation production line. The process parameter streams include real-time slurry grade spectrum, dosing valve opening timing signals, flotation cell liquid level change sequence, and foam image feature stream. B2. Inject the real-time slurry grade spectrum into the corresponding mineral phase node and update the mineral composition and surface potential data in the mineral phase node. B3. The timing signal of the dosing valve opening is correlated and compared with the drug addition rate profile. If there is a continuous deviation between the opening signal and the rate profile, the rate calibration process of the drug action unit is triggered to update the drug addition rate profile in the drug action unit. B4. The sequence of changes in the liquid level in the flotation cell is fused with the time series data of the thickness of the foam layer to generate a slurry fluid stability index. The slurry fluid stability index is then used to correct the aeration intensity distribution parameter in the flotation environmental factors. B5. Based on the updated agent addition rate profile and the corrected aeration intensity distribution parameters, recalculate the effectiveness coefficient of the agent action unit, and adjust the connection strength of the dynamic mapping relationship according to the latest effectiveness coefficient. B6. Based on the updated mineral composition and surface potential data, re-evaluate the triggering conditions of the hierarchical transition rules and simulate the real-time migration path of mineral phase nodes in the multi-level influence network.

4. The monitoring process for detecting the dose-effect relationship of phosphate rock flotation reagents according to claim 3, characterized in that, The process of the anomaly detection module is as follows: C1. Establish periodic monitoring windows during the collaborative evolution of multi-level influence networks; C2. Within each periodic monitoring window, perform response hysteresis detection, specifically as follows: C21. Track the slope of the change in the efficacy coefficient of the drug action unit over time; C22. When the slope of change changes from positive to negative, and the duration of the negative change exceeds the preset hysteresis judgment time, and the value of the efficacy coefficient is lower than the efficacy warning line, it is determined that the response hysteresis phenomenon of the drug action unit has occurred. C23. Record the identifier of the drug action unit, the time point when the slope of change turns negative, and the current value of the action efficacy coefficient when the drug action unit response hysteresis occurs, to form a response hysteresis record; C3. Within each periodic monitoring window, synchronously perform migration obstruction detection, specifically as follows: C31. Track the dwell time of mineral phase nodes on the simulated real-time migration path; C32. When the dwell time of a certain mineral phase node at a non-target level exceeds the preset migration timeout threshold, it is determined that a mineral phase node migration blockage phenomenon has occurred. C33. Record the identifier of the mineral phase node, its non-target level, and its current dwell time when the mineral phase node migration blockage phenomenon occurs, forming a migration blockage record.

5. A monitoring process for detecting the dose-effect relationship of phosphate rock flotation reagents according to claim 4, characterized in that, The steps in the decision flow activation module to activate the preset compensation decision flow to generate immediate correction instructions for the drug parameters are as follows: D1. Retrieve the response hysteresis records and analyze the identified drug action units; D2. Query the drug addition rate profile and cumulative drug consumption of the drug action unit currently bound; D3. Start the compensation decision flow, which includes the following parallel processing threads. D31, Incremental Compensation Thread, calculates the suggested compensation increment of the drug addition rate based on the negative degree of the change slope, and generates a rate increase instruction containing the drug action unit identifier and the suggested compensation increment. D32, Timing Optimization Thread, analyzes the mapping position of the reagent action unit on the progressive action chain, combines the pulp fluid stability index in the current flotation environment factors, recalculates the optimal timing of reagent action, and generates timing adjustment instructions; D33. Associate the wake-up thread, retrieve other drug action units that have a strong connection with the drug action unit that has experienced response delay, send warning signals to these associated units, and request their current action efficacy coefficient data; D4. Collection rate increase command, timing adjustment command, and effect coefficient data returned from the associated unit; D5. If the efficacy coefficient data returned by the associated unit is also generally low, then the collaborative compensation amount is superimposed on the rate increase instruction to form the final version of the real-time correction instruction for the drug parameters. D6. If the efficacy coefficient data returned by the associated unit is normal, the rate increase instruction and timing adjustment instruction are directly output as immediate correction instructions for the drug parameters.

6. A monitoring process for detecting the dose-effect relationship of phosphate rock flotation reagents according to claim 5, characterized in that, The steps in the decision flow activation module to activate the preset reconstruction decision flow to generate the topology adjustment instructions for the action chain are as follows: E1. Retrieve migration blocking records and parse the mineral phase nodes identified therein and their non-target levels; E2. Obtain the current constituent element data of mineral phase nodes, and focus on analyzing the liberation degree of target minerals and the activity of gangue minerals; E3. Initiate the refactoring decision flow. The refactoring decision flow will perform the following diagnostic and refactoring steps. E31. Precursor link diagnosis: trace back the progressive action chain that caused the current mineral phase node to migrate and block, analyze the hierarchical transition state of all preceding mineral phase nodes on the chain, locate the node that first showed abnormal transition, and mark it as the blocking source node. E32. Environmental factor interference assessment: Extract data of flotation environmental factors in the current time period and assess whether the pulp acidity and alkalinity trajectory and pulp temperature fluctuation curve deviate from the optimal environmental range required for mineral phase node level transition. E33, Agent Action Mapping Review: Review the dynamic mapping relationship of all agent action units associated with the current mineral phase node, and check for mapping relationships with decaying action efficiency coefficients or weak connection strength. E4. Based on the diagnostic results, generate topology adjustment instructions for the action chain. The topology adjustment instructions include at least one of the following operations. E41. If the blocking source node is clear, generate an instruction to insert a reinforced preprocessing node in front of the blocking source node in the multi-level influence network and configure a new agent action unit mapping for it. E42. If the environmental factor interference assessment confirms that environmental deviation is the main cause, then an instruction is generated to adjust the topology of the progressive action chain and add a parallel environmental buffer path so that the mineral phase node can be temporarily bypassed. E43. If the reagent action mapping review finds that a specific mapping relationship is invalid, an instruction is generated to cut off the currently invalid dynamic mapping relationship, and based on the constituent elements of the mineral phase node, a new reagent action unit is rematched from the available reagent library to establish a new dynamic mapping relationship.

7. A monitoring process for detecting the dose-effect relationship of phosphate rock flotation reagents according to claim 6, characterized in that, The process of the coordinated control instruction set in the coordinated control instruction integration module includes conflict resolution and sequence optimization steps: F1. Establish an instruction conflict detection mechanism to compare the operational objects and expected execution time periods involved in the real-time correction instructions for drug parameters and the topology adjustment instructions for the action chain. F2. When a command conflict is detected, the conflict resolution rule base is activated. The conflict resolution rule base contains the following rules: F21. Resource Exclusivity Rule: If two instructions compete for control of the flotation machine at the same physical dosing point or within the same time period, the topology adjustment instruction of the action chain shall be executed first, and the immediate correction instruction of the reagent parameters shall be suspended and marked with a delayed execution label. F22. Logical Dependency Rule: If the execution of the topology adjustment instruction in the action chain is a prerequisite for the immediate correction instruction of the drug parameter to take effect, then the immediate correction instruction of the drug parameter shall be executed after the topology adjustment instruction is completed, and a logical connection mark shall be established between the two. F3. Optimize the execution sequence of instructions that do not conflict or have their conflicts resolved, specifically as follows: F31. Based on the physical sequence of the mineral phase nodes processed by the instructions in the flotation process, perform preliminary sorting of the instructions; F32. For multiple instructions that process the same mineral phase node or closely related nodes, calculate the expected convergence speed of the network state after execution, and prioritize the instructions with the faster expected convergence speed. F4. Based on the equipment readiness status and process switching costs at the flotation production execution end, fine-tune the instruction execution sequence to form the final coordinated control instruction set; F5. Add execution condition constraints and result feedback requirements to each instruction in the coordinated control instruction set. The execution condition constraints include the required flotation environmental factor threshold range, and the result feedback requirements include the mineral phase node migration state or reagent action unit efficiency coefficient change that needs to be monitored.

8. A monitoring process for detecting the dose-effect relationship of phosphate rock flotation reagents according to claim 3, characterized in that, The step of calculating the efficacy coefficient of the drug action unit in real time is as follows: G1. The cumulative amount of drug consumption corresponding to the drug action unit within the set evaluation period; G2. Within the same evaluation period, calculate the target mineral recovery increment by comparing the target mineral content before and after the mineral phase node transition. G3. Calculate the ratio of the target mineral recovery increment to the cumulative amount of reagent consumption to obtain the original efficacy ratio; G4. Introduce a correction factor to standardize the original efficacy ratio, obtaining the standardized efficacy coefficient. The correction factor is obtained by: G41. Obtain the pulp fluid stability index in the current flotation environment factors. When the pulp fluid stability index is lower than the stability threshold, adjust the correction factor value. G42. Obtain the target mineral liberation degree of the current mineral phase node. When the liberation degree is lower than the standard liberation degree, adjust the correction factor value. G43. Multiply the original efficacy ratio by the correction factor to obtain the final efficacy coefficient.

9. A monitoring process for detecting the dose-effect relationship of phosphate rock flotation reagents according to claim 2, characterized in that, The coordinated control command integration module also includes a system feedback learning and multi-level influence network self-updating process: H1. While executing the coordinated control instruction set at the flotation production execution end, the feedback data capture thread is started simultaneously; H2. The feedback data capture thread collects the feedback process parameter stream generated by the execution of instructions in real time. The feedback process parameter stream includes the real-time changes in the slurry grade spectrum after execution, the response curve of the dosing valve opening adjustment, and signs of accelerated or unblocked migration of mineral phase nodes. H3. Feedback process parameters are injected back into the multi-level influence network; H4. Based on the feedback process parameter flow from the reinjection, trigger the self-update of the multi-level influence network. The self-update content includes: H41. Update the efficacy coefficient of the relevant drug action unit and reassess whether it is still in a state of efficacy decay; H42. Recalculate the hierarchical transition probability of the mineral phase nodes affected by the instruction and adjust their expected migration paths on the progressive action chain. H43. Evaluate the connection strength of the newly established or adjusted dynamic mapping relationship, and strengthen or weaken it based on feedback data; H5. Store the execution records of this coordinated control instruction set, the key features of the feedback process parameter flow, and the self-updating results of the multi-level influence network into the case library, which will be used to optimize the activation logic of subsequent compensation decision flow and reconstruction decision flow.

10. A monitoring process for detecting the dose-effect relationship of phosphate rock flotation reagents according to claim 7, characterized in that, The steps for prioritizing instructions with faster expected convergence speed are as follows: I1. Establish a network state convergence speed calculation module and perform the following operations. I11. Obtain multiple instructions to be sorted, a real-time snapshot of the current multi-level influence network, and the current data of flotation environment factors; I12. Construct a convergence rate estimator, which performs simulations for each instruction to be sorted: I13. Virtually inject the instructions to be sorted into the real-time status snapshot of the current multi-level influence network; I14. Based on the instruction type and parameters, simulate the impact of the execution of the instruction to be sorted on the trigger probability of the hierarchical transition rules of the target mineral phase node or associated node group, and generate the predicted node migration acceleration. I15. Based on the instruction type and parameters, simulate the execution of the instructions to be ordered, and generate the predicted efficacy recovery rate based on the effect on the efficacy coefficient of the associated drug action unit. I16. The predicted node migration acceleration and the predicted effectiveness recovery speed are weighted and fused to generate a quantitative value of the expected network state convergence speed of the instructions to be sorted. I2. Establish an instruction sorter. The instruction sorter performs the following operations. I21. Receive all unsorted instructions and their corresponding expected network state convergence speed quantization values ​​output by the convergence speed estimator. I22. Sort all instructions to be sorted in descending order according to their corresponding expected network state convergence speed quantization values. I23. Output the arranged instruction sequence as the optimized execution sequence.

Images