Method for realizing deep treatment and resource recovery of fluoride

Abstract

The invention discloses a method for realizing fluoride deep treatment and resource recovery, and particularly relates to the technical field of fluoride deep treatment and resource recovery, which comprises the following steps: acquiring process data of a fluoride deep treatment device and a resource recovery device in an operation process, continuously acquiring the process data, and acquiring the process data of the fluoride deep treatment device and the resource recovery device; outputting an original monitoring data sequence arranged according to a time sequence; the method comprises the following steps of: performing time window segmentation processing on continuously collected process data, firstly identifying a corresponding operation stage identifier, then performing semantic judgment on a monitoring signal under stage constraint, and driving strategy optimization based on a deviation between a semantic result and a treatment target or a recovery target, so as to obtain a semantic result; therefore, the problems of response state judgment, process switching and recovery path selection distortion caused by understanding the monitoring data in a fixed interpretation mode are avoided.

Description

Technical Field

[0001] This invention relates to the field of advanced treatment and resource recovery of fluorides, and more specifically, to a method for achieving advanced treatment and resource recovery of fluorides. Background Technology

[0002] In the existing process of deep treatment and resource recovery of fluorides, monitoring data is usually directly regarded as a physicochemical indicator that reflects the true state of the system. The control system uses this data to judge the reaction process, sedimentation behavior, crystal growth trend and whether the recovery conditions are met. However, in continuous operation scenarios, the same type of monitoring signal often corresponds to completely different actual meanings in different reaction contexts. For example, a decrease in fluoride ion concentration may come from normal precipitation reaction, or it may come from adsorption residues or false decreases caused by interference from complexing agents; particle size distribution disturbances represent reaction rate adjustments in the deep treatment stage, while in the recovery stage they may mean that the crystal form has begun to shift; sedimentation curves are affected by reaction kinetics in some batches, while in other batches they are affected by structural residues that have not been eliminated in the previous process. Because traditional control systems cannot recognize these superficially similar but fundamentally different semantic differences, they may mistake data changes for stable and uniform physical and chemical process indicators, leading to systematic deviations in reaction state judgment, process switching timing determination, and recycling path selection. It is evident that the core problem lies in the current technology's lack of ability to recognize the semantic changes of monitoring signals in different reaction stages, different chemical environments, and different process contexts. It still interprets data in a fixed way, leading to a distortion of the basis for strategy judgment. Summary of the Invention

[0003] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a method for achieving deep treatment and resource recovery of fluorides. By segmenting continuously collected process data into time windows, the corresponding operating stage identifiers are first identified. Then, semantic judgment is performed on the monitoring signals under stage constraints. Based on the deviation between the semantic results and the treatment or recovery targets, the strategy is optimized, thereby avoiding the problems of distorted reaction state judgment, process switching, and recovery path selection caused by interpreting monitoring data in a fixed way.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for achieving deep treatment and resource recovery of fluorides, comprising: S1. Acquire process data during the operation of the fluoride deep treatment device and resource recovery device, continuously collect the process data, and output the raw monitoring data sequence arranged in chronological order. S2. Remove obvious noise from the original monitoring data sequence, segment the monitoring data according to the preset time window, and compare it with the preset stage division conditions to identify the characteristic data used for stage judgment in each time window. S3. Repeatedly identify the corresponding feature data in multiple adjacent time windows. When the feature data in adjacent time windows meets the corresponding stage division threshold, output the corresponding stage identifier. S4. Based on the current stage identifier, extract feature information for semantic determination from the process data, compare the extracted feature information with the preset semantic determination conditions item by item, and output the corresponding semantic result according to the comparison result. S5. Compare the semantic results with the preset fluoride deep treatment target or resource recovery target, select or combine control strategies from the strategy library, output a set of strategy parameters including deep treatment strategy parameters or resource recovery strategy parameters, and control the deep treatment process or resource recovery process according to the set of strategy parameters.

[0005] In a preferred embodiment, in S1, an independent acquisition channel is set up for the process data, so that various types of process data enter the data acquisition process through an independent acquisition channel; the process data includes six types of data, namely, raw water fluoride ion concentration data, deep treatment effluent fluoride ion concentration data, suspended particle size distribution data, stirring motor current data, settling liquid level height data, and resource recovery mother liquor fluoride ion concentration data. Periodic sampling is performed on the independent acquisition channels corresponding to the six types of process data with a unified sampling period. The six types of process data are acquired in each sampling period, and a corresponding sampling time mark is attached to each type of process data to obtain process data with sampling time mark. Synchronously combine various types of process data acquired under the same sampling time marker and write them into the same monitoring record in a preset order to form a monitoring data point containing six types of process data; The monitoring data points obtained from multiple consecutive sampling cycles are sequentially arranged according to the order of the sampling time markers. The sequentially arranged monitoring data points are then connected to form an original monitoring data sequence covering the operation process of the fluoride deep treatment device and the resource recovery device.

[0006] In a preferred embodiment, in S2, each monitoring data point in the original monitoring data sequence is analyzed one by one, and it is checked whether the sampled values ​​of the six types of process data in the corresponding monitoring data point fall within the corresponding physical allowable range and measurement range. If any sampled value exceeds the corresponding range, the process data of the corresponding category is marked as obvious noise and removed. After removing the process data of the corresponding category that is marked as obvious noise, check whether each monitoring data point still contains the sampling values ​​corresponding to the six categories of process data. If the sampling values ​​of any category are missing, the corresponding monitoring data point is removed as a whole to form a monitoring data sequence. According to the preset time window size, the monitoring data sequence is segmented according to the order of sampling time markers. Monitoring data points whose sampling time falls within the same time window range are included in a time window segment, and a corresponding window identifier is assigned to each time window segment. Within each time window, the first monitoring data point sorted by sampling time within the time window is taken as the starting point, and the last monitoring data point within the time window is taken as the ending point. The difference in the sampling values ​​of the six types of process data between the starting point and the ending point is calculated. The corresponding trend is determined according to the magnitude of the difference between the various types of process data. When the difference is greater than zero, it is recorded as an upward trend; when the difference is equal to zero, it is recorded as a constant trend; and when the difference is less than zero, it is recorded as a downward trend. Within the same time window, for all sampled values ​​of the stirring motor current data, the difference between the upper and lower limits is calculated to obtain the current fluctuation amplitude. The candidate feature data set for the corresponding time window segment is composed of the differences of six types of process data, the corresponding trends of change, and the current fluctuation amplitude. The candidate feature data set includes difference features, trend features, and current fluctuation amplitude features.

[0007] In a preferred embodiment, S2 further includes, within each time window, filtering the difference features, trend features, and current fluctuation amplitude features in the candidate feature data set against preset threshold conditions in the stage division conditions for the deep governance response stage, the deep governance and resource recovery transition stage, and the resource recovery separation stage, respectively, wherein: For the difference feature, the difference of the six types of process data is compared with the upper and lower limits of the difference threshold interval preset for each stage in the stage division conditions. When the value of a difference feature falls into the difference threshold interval corresponding to any stage, the corresponding difference feature is recorded as a difference feature that meets the stage division conditions; otherwise, the corresponding difference feature is removed from the candidate feature data set. For the trend characteristics, the trend corresponding to the six types of process data is compared with the set of allowed trend characteristics preset for each stage in the stage division conditions. When a trend characteristic belongs to the set of allowed trend characteristics for any stage, the corresponding trend characteristic is recorded as a trend characteristic that meets the stage division conditions; otherwise, the corresponding trend characteristic is removed from the candidate feature data set. For the current fluctuation amplitude characteristics, the current fluctuation amplitude is compared with the upper and lower limits of the current fluctuation amplitude threshold range preset for each stage in the stage division conditions. When the current fluctuation amplitude value falls into the current fluctuation amplitude threshold range corresponding to any stage, the corresponding current fluctuation amplitude characteristic is recorded as a current fluctuation characteristic that meets the stage division conditions; otherwise, the corresponding current fluctuation amplitude characteristic is removed from the candidate feature data set. After filtering the difference characteristics, trend characteristics, and current fluctuation amplitude characteristics, for each time window segment, the filtered and retained difference characteristics, trend characteristics, and current fluctuation amplitude characteristics are used together as the characteristic data for stage judgment of the corresponding time window segment.

[0008] In a preferred embodiment, in S3, the feature data corresponding to each time window segment for stage judgment is read sequentially according to the order of the time window identifiers, all time window segments are arranged in sequence to form a window processing sequence, and a preset initial stage identifier is set for the first time window segment in the window processing sequence as the previous stage identifier. When processing the current time window segment in the window processing sequence, the difference features, trend features and current fluctuation amplitude features in the feature data used for stage judgment in the current time window segment are classified into difference feature set, trend feature set and current fluctuation amplitude feature set respectively. For each stage in the deep governance reaction stage, the deep governance and resource recovery transition stage, and the resource recovery separation stage, the number of difference features whose values ​​fall into the corresponding difference threshold interval of the current time window segment is counted in the difference feature set; the number of change trend features whose change trends belong to the allowable change trend set of the corresponding stage is counted in the change trend feature set; and the number of current fluctuation amplitude features whose values ​​fall into the corresponding current fluctuation amplitude threshold interval of the corresponding stage is counted in the current fluctuation amplitude feature set. The cumulative feature number of the corresponding stage is obtained by summing the number of features that meet the conditions of the difference threshold interval, the allowable change trend set, and the current fluctuation amplitude threshold interval. Among the cumulative feature counts of the three stages, the stage with the upper limit of the cumulative feature count is selected as the candidate stage of the current time window. When the cumulative feature count of the stage with the upper limit is greater than zero and there is only one stage with the maximum cumulative feature count, the stage identifier corresponding to the candidate stage with the upper limit is determined as the stage identifier of the current time window. When the cumulative feature count of all three stages is zero, or when there are two or more stages with the same cumulative feature count at the upper limit, no stage switching is performed on the current time window segment, and the previous stage identifier is directly used as the stage identifier of the current time window segment. Write the stage identifier of the current time window segment into the stage identifier sequence, and update the stage identifier of the current time window segment to the previous stage identifier of the next time window segment. Repeat the stage identifier identification process for the remaining time window segments in the window processing sequence until the stage identifier of all time window segments has been determined.

[0009] In a preferred embodiment, in S4, after determining the stage identifier corresponding to the current time window segment, the monitoring data points within the current time window segment are read, and the starting monitoring data point and the ending monitoring data point are determined according to the order of the sampling time markers. The sampling values ​​corresponding to the fluoride ion concentration data of the deep-treated effluent, the particle size distribution data of suspended particles, the sedimentation liquid level height data, and the fluoride ion concentration data of the resource recovery mother liquor are extracted from the starting monitoring data point and the ending monitoring data point, respectively. Based on the sampled values ​​between the initial and final monitoring data points, the changes in fluoride ion concentration, the rate of change in concentration per unit time, the change in suspended particle size, the particle size change trend, the settling velocity, and the change in fluoride ion concentration in the mother liquor from resource recovery were calculated, including: The change in fluoride ion concentration is the difference between the fluoride ion concentration samples in the effluent from the deep-treated water at the end of the monitoring data point and the beginning of the monitoring data point. The rate of change of concentration per unit time is the ratio of the change in fluoride ion concentration to the difference in sampling time between the two monitoring data points; The change in suspended particle size is the difference in particle size statistics between the end monitoring data point and the beginning monitoring data point. The trend of particle size change is determined as an upward trend, a constant trend, or a downward trend based on the sign of the change in suspended particle size; when the change in suspended particle size is greater than zero, it is determined as an upward trend; when the change in suspended particle size is equal to zero, it is determined as a constant trend; and when the change in suspended particle size is less than zero, it is determined as a downward trend. The settling velocity is the ratio of the difference in liquid level between the initial and final monitoring data points to the difference in sampling time. The change in fluoride ion concentration in the mother liquor from resource recovery is the difference between the fluoride ion concentration in the mother liquor at the starting and ending monitoring data points. The changes in fluoride ion concentration, the rate of change in concentration per unit time, the change in suspended particle size, the settling velocity, and the change in fluoride ion concentration in the mother liquor from resource recovery were used as numerical features; the trend of particle size change was used as a trend feature.

[0010] In a preferred embodiment, S4 further includes, when the current stage is identified as the deep treatment reaction stage, comparing the corresponding numerical features and trend features with preset semantic type determination conditions based on the change in fluoride ion concentration, the rate of change in concentration per unit time, the change in suspended particle size, and the particle size change trend. Each semantic type corresponds to at least one set of semantic determination sub-conditions consisting of a set of numerical threshold intervals or a set of change trends. When all semantic determination sub-conditions corresponding to a certain semantic type are satisfied, the semantic result of the reaction stage corresponding to the current time window segment is recorded. When the current stage is identified as the resource recovery and separation stage, the corresponding numerical features are compared with the preset semantic type judgment conditions based on the sedimentation rate and the change in fluoride ion concentration of the resource recovery mother liquor. When all semantic judgment sub-conditions corresponding to any semantic type are met, the semantic result of the separation stage of the corresponding semantic type for the current time window is recorded. When the current stage identifier belongs to the transition stage between deep governance and resource recycling, semantic type determination is not performed, and an empty semantic result set is directly written to the semantic result sequence. The semantic results corresponding to each time window segment are sorted according to the time order of the time window segment to obtain a sequence of semantic results arranged in chronological order.

[0011] In a preferred embodiment, in S5, after determining the stage identifier and semantic result of the current time window segment, when the stage identifier belongs to the deep treatment reaction stage, the sampled value of the fluoride ion concentration in the deep treatment effluent is read from the monitoring data point corresponding to the current time window segment, and the sampled value is compared with the target concentration range set for the effluent ion concentration in the preset fluoride deep treatment target. When the effluent ion concentration is higher than the upper limit of the target concentration range, the effluent concentration deviation is recorded as a positive deviation. When the effluent ion concentration is lower than the lower limit of the target concentration range, the effluent concentration deviation is recorded as a negative deviation. When the effluent ion concentration falls within the target concentration range, the effluent concentration deviation is recorded as zero deviation. When the current stage is identified as the resource recovery and separation stage, the settling velocity and the change in fluoride ion concentration of the mother liquor in the resource recovery are read from the monitoring data points corresponding to the current time window. These data are then compared with the corresponding target intervals in the preset resource recovery targets. When the settling velocity is lower than the lower limit of the target interval, the settling velocity deviation is recorded as a negative deviation; when the settling velocity is higher than the upper limit of the target interval, the settling velocity deviation is recorded as a positive deviation; and when the settling velocity falls within the target interval, the settling velocity deviation is recorded as zero deviation. Similarly, when the change in fluoride ion concentration of the mother liquor is lower than the lower limit of the target interval, the mother liquor concentration change deviation is recorded as a negative deviation; when the change in fluoride ion concentration of the mother liquor is higher than the upper limit of the target interval, the mother liquor concentration change deviation is recorded as a positive deviation; and when the change in fluoride ion concentration of the mother liquor falls within the target interval, the mother liquor concentration change deviation is recorded as zero deviation. Based on the directional markers of effluent concentration deviation, settling velocity deviation, and mother liquor concentration change deviation, the strategy optimization requirements for the current time window are determined. For each directional marker, strategy optimization requirements are generated according to the following rules, and the deviation index type corresponding to that directional marker is recorded as the current deviation index: When the effluent concentration deviation is a positive deviation, the strategy optimization requirement will be recorded as enhancing the deep treatment effect, and the deviation index type will be recorded as the effluent concentration deviation index. When the effluent concentration deviation is negative, the strategy optimization requirement will be recorded as weakening the deep treatment effect, and the deviation index type will be recorded as the effluent concentration deviation index. When the settlement velocity deviation is negative, it will be recorded as an enhanced settlement condition in the strategy optimization requirements, and the deviation index type will be recorded as the settlement velocity deviation index. When the settlement velocity deviation is positive, it will be recorded as a condition to reduce settlement in the strategy optimization requirements, and the deviation index type will be recorded as the settlement velocity deviation index. When the deviation of the mother liquor concentration change is negative, the strategy optimization requirement will be recorded as enhancing resource recovery capability, and the deviation index type will be recorded as the mother liquor concentration change deviation index. When the deviation of the mother liquor concentration change is a positive deviation, the strategy optimization requirement will be recorded as a weakening of resource recovery capability, and the deviation index type will be recorded as the mother liquor concentration change deviation index. When any deviation indicator is zero, no strategy optimization requirement is generated for the corresponding deviation indicator.

[0012] In a preferred embodiment, S5 further includes retrieving a strategy optimization record matching the corresponding strategy optimization requirement from a pre-built strategy library for each strategy optimization requirement in the strategy optimization requirements; the strategy optimization record includes at least the deviation index type, applicable stage identifier, strategy optimization category, parameter name, parameter adjustment direction, and parameter adjustment step size; The deviation index type is used to characterize which deviation source the strategy optimization record applies to, including at least one of the following: effluent concentration deviation index, sedimentation velocity deviation index, and mother liquor concentration change deviation index; The applicable stage identifier is used to indicate the applicable operational stage of the corresponding strategy optimization record, including at least one of the deep governance reaction stage or the resource recovery and separation stage; The strategy optimization category is used to indicate which strategy optimization requirement the corresponding strategy optimization record is used to respond to, including at least one of the following: enhancing the role of deep governance, weakening the role of deep governance, enhancing settlement conditions, weakening settlement conditions, enhancing resource recovery capacity, or weakening resource recovery capacity. The parameter name indicates the specific control parameters to be adjusted in this strategy optimization record, including at least one of the following: dosage, dosing interval, stirring speed, settling time, start and stop timing of resource recovery unit, recovery flow rate, crystal classification method, and mother liquor discharge and reflux distribution ratio; The parameter adjustment direction is used to identify the specific direction of adjustment performed on the parameter, such as increasing, decreasing, shortening, extending, raising, lowering, advancing, or delaying. The parameter adjustment step size is used to specify the adjustment range for each adjustment of the corresponding parameter. During the retrieval process, when the deviation indicator type in a certain strategy optimization record is consistent with the current deviation indicator, the applicable stage identifier is consistent with the current stage identifier, and the strategy optimization category is consistent with the current strategy optimization requirement, the corresponding strategy optimization record will be added to the strategy candidate set. Read each policy optimization record from the policy candidate set and execute parameter adjustment operations sequentially according to the order of the policy optimization records in the policy candidate set; for each policy optimization record, perform a deterministic adjustment process on the specified parameter based on the parameter name, parameter adjustment direction, and parameter adjustment step size in the policy optimization record, specifically including: When the parameter name is dosage and the parameter adjustment direction is increasing, the current dosage is added to the parameter adjustment step size to obtain the new dosage; when the parameter name is dosage and the parameter adjustment direction is decreasing, the current dosage is subtracted from the parameter adjustment step size to obtain the new dosage. When the parameter name is dosing interval and the parameter adjustment direction is shortening, the current dosing interval is subtracted from the parameter adjustment step size to obtain the new dosing interval; when the parameter name is dosing interval and the parameter adjustment direction is extending, the current dosing interval is added to the parameter adjustment step size to obtain the new dosing interval. When the parameter name is stirring speed and the parameter adjustment direction is increasing, the current stirring speed is added to the parameter adjustment step size to obtain the new stirring speed; when the parameter name is stirring speed and the parameter adjustment direction is decreasing, the current stirring speed is subtracted from the parameter adjustment step size to obtain the new stirring speed. When the parameter name is Settlement Residence Time and the parameter adjustment direction is to extend, the current Settlement Residence Time is added to the parameter adjustment step size to obtain the new Settlement Residence Time; when the parameter name is Settlement Residence Time and the parameter adjustment direction is to shorten, the current Settlement Residence Time is subtracted from the parameter adjustment step size to obtain the new Settlement Residence Time. When the parameter name is the start / stop timing of the resource recycling unit and the parameter adjustment direction is advanced, the current start / stop time is subtracted from the parameter adjustment step size to obtain the new start / stop time; when the parameter name is the start / stop timing of the resource recycling unit and the parameter adjustment direction is delayed, the current start / stop time is added to the parameter adjustment step size to obtain the new start / stop time. When the parameter name is "Recovery Flow" and the parameter adjustment direction is "Increase", the current recovery flow is added to the parameter adjustment step size to obtain the new recovery flow; when the parameter name is "Recovery Flow" and the parameter adjustment direction is "Decrease", the current recovery flow is subtracted from the parameter adjustment step size to obtain the new recovery flow. When the parameter name is Crystal Classification Method, the parameter adjustment direction will move one level forward or backward in the preset classification method sequence according to the index order, and the moved classification method will be used as the new crystal classification method. When the parameter name is Mother Liquor Discharge and Recirculation Distribution Ratio and the parameter adjustment direction is to increase the recirculation ratio, the current recirculation ratio is added to the parameter adjustment step size and limited to the allowable range to obtain a new recirculation ratio. At the same time, the discharge ratio is updated to one minus the new recirculation ratio. When the parameter name is Mother Liquor Discharge and Recirculation Distribution Ratio and the parameter adjustment direction is to decrease the recirculation ratio, the current recirculation ratio is subtracted from the parameter adjustment step size and limited to the allowable range to obtain a new recirculation ratio. At the same time, the discharge ratio is updated to one minus the new recirculation ratio. After adjusting the parameters corresponding to a single strategy optimization record, the adjusted parameter values ​​are compared with the allowed value range of the corresponding parameters item by item. When the adjusted parameter value is greater than the upper limit of the allowed value range of the corresponding parameter, the adjusted parameter value is set to the upper limit of the allowed value range of the corresponding parameter; when the adjusted parameter value is less than the lower limit of the allowed value range of the corresponding parameter, the adjusted parameter value is set to the lower limit of the allowed value range of the corresponding parameter. When multiple policy optimization records in the policy candidate set point to the same parameter, the absolute value operation is performed on the deviation value corresponding to each policy optimization record to obtain the absolute value of the deviation. The magnitudes of all absolute deviation values ​​are compared, and the policy optimization record with the absolute value of the deviation at the upper limit is selected as the current policy optimization record to be executed. The parameter adjustment operation is then performed on the parameter corresponding to the currently executed policy optimization record. Other policy optimization records that point to the same parameter are not executed.

[0013] The technical effects and advantages of this invention are as follows: This solution introduces a stage identification and semantic judgment mechanism to interpret the same type of monitoring signals within the corresponding reaction stage and process context. This avoids treating similar numerical changes under different reaction mechanisms as the same physicochemical process indicator, and fundamentally solves the problem of systematic deviations in reaction state judgment, process switching timing, and recovery path selection caused by fixed interpretation methods of monitoring data in existing technologies. By segmenting continuous monitoring data into time windows and determining stage identifiers based on the cumulative satisfaction of multiple features within adjacent time windows, the switching between the deep governance response stage, transition stage, and resource recovery separation stage is established on the basis of multi-feature consistency, avoiding stage misjudgment due to instantaneous fluctuations or single signal anomalies, and improving the stability and repeatability of stage identification under continuous operation conditions. After identifying the stage identifier, the feature information corresponding to that stage is further extracted, and the reaction mechanism state is identified through semantic judgment conditions. This allows changes in fluoride ion concentration, particle size evolution, and sedimentation behavior to no longer be used merely as numerical trends, but to be mapped as semantic results with clear process meanings. This provides a distinguishable semantic basis for subsequent control strategies rather than a purely numerical basis. Based on the deviation direction between semantic results and governance or recovery objectives, strategy optimization requirements are generated, and parameter adjustments are achieved through pre-structured strategy optimization records in the strategy library. This transforms control decisions from whether numerical values ​​exceed limits to whether the current semantic state deviates from the target semantics, thereby reducing inappropriate control actions caused by false degradation, stage misalignment, or residual structure effects. During strategy execution, an absolute value comparison of deviations and a parameter conflict resolution mechanism are introduced. When multiple strategies simultaneously point to the same control parameter, only the strategy with the largest deviation is executed. This avoids parameter oscillation or over-adjustment caused by the superposition of multiple strategies, ensuring that the parameter adjustment direction is clear and the magnitude is controllable during continuous optimization of the deep governance and resource recycling process. Attached Figure Description

[0014] Figure 1 This is a flowchart outlining the method steps of the present invention. Detailed Implementation

[0015] 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.

[0016] Refer to the instruction manual appendix Figure 1 An embodiment of the present invention provides a method for achieving deep treatment and resource recovery of fluorides, comprising: S1. Acquire process data during the operation of the fluoride deep treatment device and the resource recovery device, continuously collect the process data, and output the raw monitoring data sequence arranged in chronological order; the process data includes the fluoride ion concentration in the raw water, the fluoride ion concentration in the deep treated water, the particle size distribution of suspended particles, the current of the stirring motor, the height of the settling liquid level, and the fluoride ion concentration in the resource recovery mother liquor. S2. Remove obvious noise from the original monitoring data sequence, segment the monitoring data according to the preset time window, and compare it with the preset stage division conditions to identify the characteristic data used for stage judgment in each time window; wherein the stage division conditions include at least the characteristic thresholds for the deep governance response stage, the deep governance and resource recovery transition stage, and the resource recovery separation stage. S3. Repeatedly identify the corresponding feature data within multiple adjacent time windows. When the feature data within an adjacent time window meets the corresponding stage division threshold, output the corresponding stage identifier; otherwise, keep the previous stage identifier unchanged. The corresponding stage identifier is used to distinguish between the deep governance response stage, the deep governance and resource recovery transition stage, and the resource recovery separation stage. S4. Based on the current stage identifier, extract feature information for semantic determination from the process data, compare the extracted feature information with the preset semantic determination conditions item by item, and output the corresponding semantic result according to the comparison result. S5. Compare the semantic results with the preset fluoride deep treatment target or resource recovery target, select or combine control strategies from the strategy library, output a set of strategy parameters including deep treatment strategy parameters or resource recovery strategy parameters, and control the deep treatment process or resource recovery process according to the set of strategy parameters. After the control output, continue to collect new process data for subsequent execution steps. The deep treatment strategy parameters include at least the dosage, dosing time interval, stirring speed and settling residence time, and the resource recovery strategy parameters include at least the start and stop time of the resource recovery unit, recovery flow rate, crystal classification method and recovery mother liquor discharge or return path.

[0017] In S1, an independent acquisition channel is set up for process data, so that various types of process data enter the data acquisition process through an independent acquisition channel. The process data includes six types of data, namely raw water fluoride concentration data, deep treatment effluent fluoride concentration data, suspended particle size distribution data, stirring motor current data, settling liquid level height data, and resource recovery mother liquor fluoride concentration data. Periodic sampling is performed on the independent acquisition channels corresponding to the six types of process data with a unified sampling period. The six types of process data are acquired in each sampling period, and a corresponding sampling time mark is attached to each type of process data to obtain process data with sampling time mark. Synchronously combine various process data acquired under the same sampling time marker and write them into the same monitoring record in a preset order to form a monitoring data point containing the above six types of process data; wherein the monitoring record refers to a structured data unit formed by organizing the six types of process data in a preset order under the same sampling time marker. The data unit includes at least a sampling time marker for identifying the sampling time and multiple data fields for storing the corresponding sampling values ​​of the six types of process data, which are used to represent the overall operating status of the fluoride deep treatment device and the resource recovery device at a certain sampling time; The monitoring data points obtained from multiple consecutive sampling cycles are sequentially arranged according to the order of the sampling time markers. The sequentially arranged monitoring data points are then connected to form an original monitoring data sequence covering the operation of the fluoride deep treatment device and the resource recovery device. This sequence serves as a unified input data source for subsequent steps such as removing obvious noise, segmenting the data according to a preset time window, and performing stage judgment, semantic determination, and strategy optimization. It should be noted that, in this solution, the fluoride deep treatment device refers to a set of treatment units used to treat fluoride-containing raw water in multiple stages to ensure that the fluoride ion concentration in the effluent meets the preset discharge or reuse standards. Specifically, it may include, but is not limited to, a raw water inlet unit, a chemical dosing unit, a mixing reaction zone, a flocculation zone, a sedimentation zone, and the supporting stirring mechanism and piping structure. The fluoride deep treatment device creates stable hydraulic and chemical reaction conditions in the reaction zone, causing the fluoride ions in the raw water to undergo precipitation or adsorption reactions with the added chemicals, generating fluoride-containing particles of controllable size. Solid-liquid separation is then achieved through the sedimentation zone, thereby forming deeply treated effluent with a controllable fluoride ion concentration on the effluent side. A resource recovery device refers to a set of processing units that further treat fluoride-containing precipitates or high-fluoride mother liquor generated during deep treatment to recover fluoride crystals or high-value fluoride-containing products. Specifically, it may include, but is not limited to, concentration units, classification units, solid-liquid separation units, washing units, and mother liquor circulation units. The resource recovery device separates fluoride crystals that meet the recovery conditions from the system by adjusting the solid-liquid ratio, controlling the crystal residence time, and classification conditions. It also monitors and adjusts the fluoride ion concentration in the recovered mother liquor to achieve mother liquor discharge or recirculation while meeting the resource utilization target. Process data refers to the set of operating parameters recorded in real time by corresponding monitoring points during the operation of fluoride deep treatment and resource recovery devices. These parameters reflect the characteristics of raw water, reaction conditions, suspended particle state, equipment load, and the properties of the solution in the recovery area. This process data includes at least the following: Raw water fluoride concentration: used to characterize the initial fluoride load in the raw water entering the deep treatment device, and is an important basic parameter for assessing the reaction dosage requirement, sedimentation process intensity and recovery potential; Fluoride ion concentration in effluent from deep treatment: used to reflect the treatment effect and the stabilization characteristics of the reaction stage, and used in the semantic determination stage to identify the mechanism of precipitation reaction, surface adsorption or complexation interference; Suspended particle size distribution: Collected by optical or laser particle size sensors, it reflects the growth, aggregation, breakage and sedimentation behavior of particles in the reaction zone, and is an important basis for stage judgment and resource recovery determination. Stirring motor current: It reflects the actual load change of the stirring device and has a direct coupling relationship with reaction viscosity, particle concentration and agglomeration degree, which can indicate deviations in the reaction process; Settling liquid level height: This reflects the position of the solid-liquid interface and the change in the settling rate in the settling unit. It is an important parameter for judging whether the settling is stable and whether the conditions for graded recycling are met. Fluoride ion concentration in mother liquor for resource recovery: used to determine the degree of fluoride removal and recovery efficiency during the recovery process, and in semantic determination to distinguish whether the resource recovery target has been achieved.

[0018] In S2, each monitoring data point in the original monitoring data sequence is analyzed one by one. It is checked whether the sampled values ​​of the six types of process data in the corresponding monitoring data point fall within the corresponding physical allowable range and measurement range. If any sampled value exceeds the corresponding range, the process data of the corresponding category is marked as obvious noise and removed. The sampled value refers to the single measurement value of each type of process data collected and recorded in the monitoring data point by the corresponding detection point in each sampling period. This value corresponds one-to-one with the sampling time mark and is used to represent the actual measurement result of each process data at that sampling time. The physical allowable range is the range of values ​​of the six types of process data preset based on the normal operating conditions of the fluoride deep treatment device and the resource recovery device. The measurement range refers to the range of values ​​allowed by the measurement means used to obtain various types of process data in terms of physical measurement capability, including the upper limit and lower limit of measurement, and is used to determine whether the sampled value is within the effective measurement range. After removing the process data of the corresponding category that is marked as obvious noise, check whether each monitoring data point still contains the sampling values ​​corresponding to the six categories of process data. If the sampling values ​​of any category are missing, the corresponding monitoring data point is removed as a whole, forming a monitoring data sequence after noise removal and integrity verification. According to the preset time window size, the monitoring data sequence that has passed the integrity verification is segmented according to the order of the sampling time markers. Monitoring data points whose sampling time falls within the same time window range are included in a time window segment, and a corresponding window identifier is assigned to each time window segment. Within each time window, the first monitoring data point sorted by sampling time within the time window is taken as the starting point, and the last monitoring data point within the time window is taken as the ending point. The difference in the sampling values ​​of the six types of process data between the starting point and the ending point is calculated. The corresponding trend is determined according to the magnitude of the difference between the various types of process data. When the difference is greater than zero, it is recorded as an upward trend; when the difference is equal to zero, it is recorded as a constant trend; and when the difference is less than zero, it is recorded as a downward trend. Within the same time window, for all sampled values ​​of the stirring motor current data, the difference between the upper and lower limits is calculated to obtain the current fluctuation amplitude. The candidate feature data set for the corresponding time window segment is composed of the differences of six types of process data, the corresponding trends of change, and the current fluctuation amplitude. This provides a basis for the data analysis required for subsequent stage judgment steps and the optimization of the execution strategy based on stage identifiers. The candidate feature data set includes difference features, trend features, and current fluctuation amplitude features.

[0019] S2 also includes, within each time window, filtering the difference features, trend features, and current fluctuation amplitude features in the candidate feature data set against preset threshold conditions for the deep governance response stage, the deep governance and resource recovery transition stage, and the resource recovery separation stage, respectively, in accordance with the stage division conditions. For the difference feature, the difference of the six types of process data is compared with the upper and lower limits of the difference threshold interval preset for each stage in the stage division conditions. When the value of a difference feature falls into the difference threshold interval corresponding to any stage, the corresponding difference feature is recorded as a difference feature that meets the stage division conditions; otherwise, the corresponding difference feature is removed from the candidate feature data set. For the trend characteristics, the trend corresponding to the six types of process data is compared with the set of allowed trend characteristics preset for each stage in the stage division conditions. When a trend characteristic belongs to the set of allowed trend characteristics for any stage, the corresponding trend characteristic is recorded as a trend characteristic that meets the stage division conditions; otherwise, the corresponding trend characteristic is removed from the candidate feature data set. For the current fluctuation amplitude characteristics, the current fluctuation amplitude is compared with the upper and lower limits of the current fluctuation amplitude threshold range preset for each stage in the stage division conditions. When the current fluctuation amplitude value falls into the current fluctuation amplitude threshold range corresponding to any stage, the corresponding current fluctuation amplitude characteristic is recorded as a current fluctuation characteristic that meets the stage division conditions; otherwise, the corresponding current fluctuation amplitude characteristic is removed from the candidate feature data set. After filtering the difference features, trend features, and current fluctuation amplitude features, for each time window segment, the filtered and retained difference features, trend features, and current fluctuation amplitude features are used together as feature data for the corresponding time window segment for stage judgment. They are then output to the stage judgment step in the order of the time window segments. This data is used to generate stage identifiers for the deep treatment reaction stage, the deep treatment and resource recovery transition stage, or the resource recovery separation stage of fluoride based on the feature data of adjacent time window segments. It also serves as the data foundation for the execution of strategy selection and parameter generation in the subsequent semantic judgment and strategy optimization steps.

[0020] In S3, the feature data corresponding to each time window segment for stage judgment is read sequentially according to the order of the time window identifiers. All time window segments are arranged in order to form a window processing sequence, and a preset initial stage identifier is set for the first time window segment in the window processing sequence as the previous stage identifier. When processing the current time window segment in the window processing sequence, the difference features, trend features and current fluctuation amplitude features in the feature data used for stage judgment in the current time window segment are classified into difference feature set, trend feature set and current fluctuation amplitude feature set respectively. For each stage in the deep governance reaction stage, the deep governance and resource recovery transition stage, and the resource recovery separation stage, the number of difference features whose values ​​fall into the corresponding difference threshold interval of the current time window segment is counted in the difference feature set; the number of change trend features whose change trends belong to the allowable change trend set of the corresponding stage is counted in the change trend feature set; and the number of current fluctuation amplitude features whose values ​​fall into the corresponding current fluctuation amplitude threshold interval of the corresponding stage is counted in the current fluctuation amplitude feature set. The cumulative feature number of the corresponding stage is obtained by summing the number of features that meet the conditions of the difference threshold interval, the allowable change trend set, and the current fluctuation amplitude threshold interval. Among the cumulative feature counts of the three stages, the stage with the upper limit of the cumulative feature count is selected as the candidate stage of the current time window. When the cumulative feature count of the stage with the upper limit is greater than zero and there is only one stage with the maximum cumulative feature count, the stage identifier corresponding to the candidate stage with the upper limit is determined as the stage identifier of the current time window. When the cumulative feature count of all three stages is zero, or when there are two or more stages with the same cumulative feature count at the upper limit, no stage switching is performed on the current time window segment, and the previous stage identifier is directly used as the stage identifier of the current time window segment. The stage identifier of the current time window segment is written into the stage identifier sequence, and the stage identifier of the current time window segment is updated to the previous stage identifier of the next time window segment. The stage identifier identification process of the corresponding time window segment is repeated for the remaining time window segments in the window processing sequence until the stage identifier of all time window segments is determined. This enables the stage identifier sequence to distinguish the fluoride deep treatment reaction stage, the deep treatment and resource recovery transition stage, and the resource recovery separation stage in the time dimension, and provides a stage identifier basis for subsequent semantic judgment and strategy optimization.

[0021] In S4, after determining the stage identifier corresponding to the current time window segment, the monitoring data points within the current time window segment are read. The starting and ending monitoring data points are determined according to the order of the sampling time markers. The sampling values ​​corresponding to the fluoride ion concentration data of the deep-treated effluent, the particle size distribution data of suspended particles, the sedimentation liquid level height data, and the fluoride ion concentration data of the resource recovery mother liquor are extracted from the starting and ending monitoring data points respectively, which are used to form the original data basis required for semantic determination. Based on the sampled values ​​between the initial and final monitoring data points, the changes in fluoride ion concentration, the rate of change in concentration per unit time, the change in suspended particle size, the particle size change trend, the settling velocity, and the change in fluoride ion concentration in the mother liquor from resource recovery were calculated, including: The change in fluoride ion concentration is the difference between the fluoride ion concentration samples in the effluent from the deep-treated water at the end of the monitoring data point and the beginning of the monitoring data point. The rate of change of concentration per unit time is the ratio of the change in fluoride ion concentration to the difference in sampling time between the two monitoring data points; The change in suspended particle size is the difference in particle size statistics between the end monitoring data point and the beginning monitoring data point. The particle size change trend is determined as an upward trend, a constant trend, or a downward trend based on the sign of the change in suspended particle size. When the change in suspended particle size is greater than zero, it is determined as an upward trend; when the change in suspended particle size is equal to zero, it is determined as a constant trend; and when the change in suspended particle size is less than zero, it is determined as a downward trend. The sign refers to the directional marker determined by the positive or negative sign of the change in suspended particle size, which is used to indicate the direction of particle size change within the time window and is the result after extracting the sign of the particle size change. The settling velocity is the ratio of the difference in liquid level between the initial and final monitoring data points to the difference in sampling time. The change in fluoride ion concentration in the mother liquor from resource recovery is the difference between the fluoride ion concentration in the mother liquor at the starting and ending monitoring data points. The changes in fluoride ion concentration, the rate of change in concentration per unit time, the change in suspended particle size, the settling velocity, and the change in fluoride ion concentration in the mother liquor from resource recovery were used as numerical features; the trend of particle size change was used as a trend feature.

[0022] S4 also includes, when the current stage is identified as the deep treatment reaction stage, comparing the corresponding numerical features and trend features with the preset semantic type determination conditions based on the change in fluoride ion concentration, the rate of change in concentration per unit time, the change in suspended particle size, and the particle size change trend. Each semantic type corresponds to at least one set of semantic determination sub-conditions consisting of a set of numerical threshold intervals or a set of change trends. When all the semantic determination sub-conditions corresponding to a certain semantic type are satisfied, the semantic result of the reaction stage for the current time window segment is recorded. The semantic type refers to the state category used to characterize the operating mechanism of the deep treatment reaction stage, including at least one reaction mechanism state that can be distinguished based on the above features. When the current stage is identified as the resource recovery and separation stage, the corresponding numerical features are compared with the preset semantic type judgment conditions based on the sedimentation rate and the change in fluoride ion concentration of the resource recovery mother liquor. When all semantic judgment sub-conditions corresponding to any semantic type are met, the semantic result of the separation stage of the corresponding semantic type for the current time window is recorded. When the current stage identifier belongs to the transition stage of deep governance and resource recycling, semantic type determination is not performed, and an empty semantic result set is directly written to the semantic result sequence; where the empty semantic result set refers to the placeholder result that does not contain any semantic type identifier, which is used to indicate that semantic determination does not need to be performed within this time window. The semantic results corresponding to each time window segment are sorted according to the time order of the time window segments to obtain a sequence of semantic results arranged in time order. The sequence of semantic results is used as the input basis for generating deep governance strategy parameters or resource recovery strategy parameters in subsequent strategy optimization steps, and is used to provide a semantic judgment basis corresponding to the running stage for the strategy optimization process.

[0023] In S5, after determining the stage identifier and semantic result of the current time window segment, when the stage identifier belongs to the deep treatment reaction stage, the sampled value of the fluoride ion concentration in the deep treatment effluent is read from the monitoring data point corresponding to the current time window segment, and the sampled value is compared with the target concentration range set for the effluent ion concentration in the preset fluoride deep treatment target. When the effluent ion concentration is higher than the upper limit of the target concentration range, the effluent concentration deviation is recorded as a positive deviation. When the effluent ion concentration is lower than the lower limit of the target concentration range, the effluent concentration deviation is recorded as a negative deviation. When the effluent ion concentration falls within the target concentration range, the effluent concentration deviation is recorded as zero deviation. When the current stage is identified as the resource recovery and separation stage, the settling velocity and the change in fluoride ion concentration of the mother liquor in the resource recovery are read from the monitoring data points corresponding to the current time window. These data are then compared with the corresponding target intervals in the preset resource recovery targets. When the settling velocity is lower than the lower limit of the target interval, the settling velocity deviation is recorded as a negative deviation; when the settling velocity is higher than the upper limit of the target interval, the settling velocity deviation is recorded as a positive deviation; and when the settling velocity falls within the target interval, the settling velocity deviation is recorded as zero deviation. Similarly, when the change in fluoride ion concentration of the mother liquor is lower than the lower limit of the target interval, the mother liquor concentration change deviation is recorded as a negative deviation; when the change in fluoride ion concentration of the mother liquor is higher than the upper limit of the target interval, the mother liquor concentration change deviation is recorded as a positive deviation; and when the change in fluoride ion concentration of the mother liquor falls within the target interval, the mother liquor concentration change deviation is recorded as zero deviation. The strategy optimization requirements for the current time window are determined based on the directional markers of effluent concentration deviation, settling velocity deviation, and mother liquor concentration change deviation. The directional markers indicate whether the deviation is positive, zero, or negative. For each directional marker, strategy optimization requirements are generated according to the following rules, and the deviation index type corresponding to that directional marker is recorded as the current deviation index: When the effluent concentration deviation is a positive deviation, the strategy optimization requirement will be recorded as enhancing the deep treatment effect, and the deviation index type will be recorded as the effluent concentration deviation index. When the effluent concentration deviation is negative, the strategy optimization requirement will be recorded as weakening the deep treatment effect, and the deviation index type will be recorded as the effluent concentration deviation index. When the settlement velocity deviation is negative, it will be recorded as an enhanced settlement condition in the strategy optimization requirements, and the deviation index type will be recorded as the settlement velocity deviation index. When the settlement velocity deviation is positive, it will be recorded as a condition to reduce settlement in the strategy optimization requirements, and the deviation index type will be recorded as the settlement velocity deviation index. When the deviation of the mother liquor concentration change is negative, the strategy optimization requirement will be recorded as enhancing resource recovery capability, and the deviation index type will be recorded as the mother liquor concentration change deviation index. When the deviation of the mother liquor concentration change is a positive deviation, the strategy optimization requirement will be recorded as a weakening of resource recovery capability, and the deviation index type will be recorded as the mother liquor concentration change deviation index. When any deviation indicator is zero, no strategy optimization requirement is generated for the corresponding deviation indicator.

[0024] S5 also includes retrieving strategy optimization records matching the corresponding strategy optimization requirement from a pre-built strategy library for each strategy optimization requirement in the strategy optimization requirements. The strategy optimization record includes at least six fields: deviation index type, applicable stage identifier, strategy optimization category, parameter name, parameter adjustment direction, and parameter adjustment step size. The strategy library refers to a structured collection of strategy records organized according to the six fields: deviation index type, applicable stage identifier, strategy optimization category, parameter name, parameter adjustment direction, and parameter adjustment step size. Each strategy optimization record corresponds to an executable control action, which is used to trigger the adjustment of the corresponding process parameters when a process deviation occurs. All strategy optimization records in the strategy library can be indexed and called according to the deviation source type and operation stage, which is used to realize the dynamic control of the deep treatment process and the resource recovery process. The deviation index type is used to characterize which deviation source the strategy optimization record applies to, including at least one of the following: effluent concentration deviation index, sedimentation velocity deviation index, and mother liquor concentration change deviation index; The applicable stage identifier is used to indicate the applicable operational stage of the corresponding strategy optimization record, including at least one of the deep governance reaction stage or the resource recovery and separation stage; The strategy optimization category is used to indicate which strategy optimization requirement the corresponding strategy optimization record is used to respond to, including at least one of the following: enhancing the role of deep governance, weakening the role of deep governance, enhancing settlement conditions, weakening settlement conditions, enhancing resource recovery capacity, or weakening resource recovery capacity. The parameter name indicates the specific control parameters to be adjusted in this strategy optimization record, including at least one of the following: dosage, dosing interval, stirring speed, settling time, start and stop timing of resource recovery unit, recovery flow rate, crystal classification method, and mother liquor discharge and reflux distribution ratio; The parameter adjustment direction is used to identify the specific direction of adjustment performed on the parameter, such as increasing, decreasing, shortening, extending, raising, lowering, advancing, or delaying. The parameter adjustment step size is used to specify the adjustment range for each adjustment of the corresponding parameter. During the retrieval process, when the deviation indicator type in a certain strategy optimization record is consistent with the current deviation indicator, the applicable stage identifier is consistent with the current stage identifier, and the strategy optimization category is consistent with the current strategy optimization requirement, the corresponding strategy optimization record will be added to the strategy candidate set. Read each policy optimization record from the policy candidate set and execute parameter adjustment operations sequentially according to the order of the policy optimization records in the policy candidate set; for each policy optimization record, perform a deterministic adjustment process on the specified parameter based on the parameter name, parameter adjustment direction, and parameter adjustment step size in the policy optimization record, specifically including: When the parameter name is dosage and the parameter adjustment direction is increasing, the current dosage is added to the parameter adjustment step size to obtain the new dosage; when the parameter name is dosage and the parameter adjustment direction is decreasing, the current dosage is subtracted from the parameter adjustment step size to obtain the new dosage. When the parameter name is dosing interval and the parameter adjustment direction is shortening, the current dosing interval is subtracted from the parameter adjustment step size to obtain the new dosing interval; when the parameter name is dosing interval and the parameter adjustment direction is extending, the current dosing interval is added to the parameter adjustment step size to obtain the new dosing interval. When the parameter name is stirring speed and the parameter adjustment direction is increasing, the current stirring speed is added to the parameter adjustment step size to obtain the new stirring speed; when the parameter name is stirring speed and the parameter adjustment direction is decreasing, the current stirring speed is subtracted from the parameter adjustment step size to obtain the new stirring speed. When the parameter name is Settlement Residence Time and the parameter adjustment direction is to extend, the current Settlement Residence Time is added to the parameter adjustment step size to obtain the new Settlement Residence Time; when the parameter name is Settlement Residence Time and the parameter adjustment direction is to shorten, the current Settlement Residence Time is subtracted from the parameter adjustment step size to obtain the new Settlement Residence Time. When the parameter name is the start / stop timing of the resource recycling unit and the parameter adjustment direction is advanced, the current start / stop time is subtracted from the parameter adjustment step size to obtain the new start / stop time; when the parameter name is the start / stop timing of the resource recycling unit and the parameter adjustment direction is delayed, the current start / stop time is added to the parameter adjustment step size to obtain the new start / stop time. When the parameter name is "Recovery Flow" and the parameter adjustment direction is "Increase", the current recovery flow is added to the parameter adjustment step size to obtain the new recovery flow; when the parameter name is "Recovery Flow" and the parameter adjustment direction is "Decrease", the current recovery flow is subtracted from the parameter adjustment step size to obtain the new recovery flow. When the parameter name is "Crystal Classification Method," the parameter is moved one level forward or backward in the preset classification method sequence according to the index order, based on the parameter adjustment direction. The moved classification method is then used as the new crystal classification method. The classification method sequence refers to an ordered set of classification methods that can be used in the resource recovery process, sorted according to crystal size, morphology, or sedimentation characteristics. Different classification methods correspond to different crystal size cutoff conditions or classification paths, which are used to adjust the crystal separation effect during resource recovery. The index order of this classification method sequence is used to determine the classification method upgrade or downgrade logic when adjusting parameters. When the parameter name is "Mother Liquor Discharge and Recirculation Allocation Ratio" and the parameter adjustment direction is to increase the recirculation ratio, the current recirculation ratio is added to the parameter adjustment step size and limited within the allowable range to obtain a new recirculation ratio. Simultaneously, the discharge ratio is updated to one minus the new recirculation ratio. When the parameter name is "Mother Liquor Discharge and Recirculation Allocation Ratio" and the parameter adjustment direction is to decrease the recirculation ratio, the current recirculation ratio is subtracted from the parameter adjustment step size and limited within the allowable range to obtain a new recirculation ratio. Simultaneously, the discharge ratio is updated to one minus the new recirculation ratio. The recirculation ratio refers to the proportion of mother liquor allocated to the recirculation path during resource recovery, used to control the balance between the mother liquor circulation volume and the new liquor replenishment volume in the system. "Limited within the allowable range" means that the adjusted parameter value is constrained within the feasible physical range or process-permissible range of the parameter. When it exceeds the upper limit, it is limited to the upper limit value; when it is below the lower limit value, it is limited to the lower limit value. The discharge ratio refers to the proportion of mother liquor allocated to the discharge path after the recirculation ratio is determined. Its value is equal to one minus the recirculation ratio, used to represent the percentage of mother liquor allowed to be discharged under the current operating conditions. After adjusting the parameters corresponding to a single strategy optimization record, the adjusted parameter values ​​are compared with the corresponding allowed value range item by item. When the adjusted parameter value is greater than the upper limit of the corresponding allowed value range, the adjusted parameter value is set to the upper limit of the corresponding allowed value range; when the adjusted parameter value is less than the lower limit of the corresponding allowed value range, the adjusted parameter value is set to the lower limit of the corresponding allowed value range. The corresponding allowed value range refers to the range of values ​​that can be preset for each control parameter based on process requirements, equipment capabilities, or physical limitations, used to ensure that parameter adjustments do not lead to an operating state that exceeds the controllable range of the system. When multiple policy optimization records in the policy candidate set point to the same parameter, the absolute value of the deviation value corresponding to each policy optimization record is calculated. The absolute value of the deviation is then compared with the absolute value of all deviations. The policy optimization record with the absolute value of the deviation at the upper limit is selected as the current policy optimization record to be executed. The parameter adjustment operation is then performed on the parameter corresponding to the currently executed policy optimization record. Other policy optimization records pointing to the same parameter are not executed. The deviation value refers to the difference between the current sampled value and the boundary of the corresponding target interval. When the sampled value is higher than the upper limit of the target interval, the deviation value is the sampled value minus the upper limit of the target interval. When the sampled value is lower than the lower limit of the target interval, the deviation value is the sampled value minus the lower limit of the target interval. When the sampled value falls within the target interval, the deviation value is zero. The parameter adjustment operation refers to performing an adjustment process on the corresponding control parameter according to the parameter name, parameter adjustment direction, and parameter adjustment step size in the policy optimization record. After adjustment, the updated parameter value is limited to a range according to the allowed value range.

[0025] It should be noted that, including but not limited to, this solution adds a complete data semantics and strategy optimization chain to the traditional deep treatment and resource recovery process for fluorides. This transforms the original crude control method that only looks at numerical curves into a full-process control structure that includes stage identification, semantic judgment, and strategy optimization. The specific implementation method is as follows: This solution first uses S1 to abstract the operating status of the fluoride deep treatment unit and the resource recovery unit into a raw monitoring data sequence containing six types of process data, so that the process status at each sampling moment can be described in a structured form. This step is not a simple data collection, but provides a unified and traceable data input basis for subsequent stage judgment, semantic judgment and strategy optimization, avoiding ambiguity in subsequent analysis between different data sources and time bases. Subsequently, in S2, this scheme does not directly make process judgments based on the monitoring data. Instead, it first organizes the data's reliability and time structure through noise removal, integrity verification, and time window segmentation. Then, it constructs a candidate feature data set based on the overall change and current fluctuation amplitude within the time window, and selects feature data for stage judgment based on the stage division conditions. The significance of this process is that it transforms the originally scattered instantaneous values ​​into a feature set that can reflect the evolution trend of the reaction over a period of time, thereby creating conditions for stable stage identification. Building on this, S3 further transforms the feature data into a phase identifier sequence. By cumulatively judging the feature matching within adjacent time windows, it distinguishes between the deep treatment reaction phase, the transition phase, and the resource recovery and separation phase. It also maintains phase continuity when there is uncertainty, avoiding frequent fluctuations in process status between adjacent windows. This step establishes a clear timeline of operational phases, providing contextual constraints for subsequent interpretation of data meaning. In the following S4, this solution does not directly map the monitoring values ​​to control actions. Instead, under the constraint of stage identifiers, it executes different semantic judgment logics for the same type of data changes. For example, only when it is confirmed that the deep treatment reaction stage is in progress are changes in fluoride ion concentration and particle size used to determine the reaction mechanism; only when it is confirmed that the resource recovery and separation stage is in progress are changes in sedimentation rate and mother liquor fluoride ion concentration used to determine the recovery status. Through this structure of first judging the stage and then interpreting the semantics, the different meanings of the same monitoring signal in different process contexts are clearly distinguished, avoiding the fixed interpretation of data in the prior art. Finally, S5 compares the semantic results with the preset deep governance or resource recovery targets to generate strategy optimization requirements. These requirements are then transformed into specific parameter adjustment actions through the strategy library. Since strategy generation is based on stage identifiers and semantic results, the adjustment of control parameters is no longer a direct response to numerical deviations, but a targeted optimization based on the current stage and the type of deviation that has occurred. This allows for complete optimization control of the deep governance and resource recovery process while ensuring process safety boundaries.

[0026] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for achieving deep treatment and resource recovery of fluorides, characterized in that, include: S1. Acquire process data during the operation of the fluoride deep treatment device and resource recovery device, continuously collect the process data, and output the raw monitoring data sequence arranged in chronological order. S2. Remove obvious noise from the original monitoring data sequence, segment the monitoring data according to the preset time window, and compare it with the preset stage division conditions to identify the characteristic data used for stage judgment in each time window. S3. Repeatedly identify the corresponding feature data in multiple adjacent time windows. When the feature data in adjacent time windows meets the corresponding stage division threshold, output the corresponding stage identifier. S4. Based on the current stage identifier, extract feature information for semantic determination from the process data, compare the extracted feature information with the preset semantic determination conditions item by item, and output the corresponding semantic result according to the comparison result. S5. Compare the semantic results with the preset fluoride deep treatment target or resource recovery target, select or combine control strategies from the strategy library, output a set of strategy parameters including deep treatment strategy parameters or resource recovery strategy parameters, and control the deep treatment process or resource recovery process according to the set of strategy parameters.

2. The method for achieving deep treatment and resource recovery of fluorides according to claim 1, characterized in that: In S1, an independent acquisition channel is set up for process data, so that various types of process data enter the data acquisition process through an independent acquisition channel. The process data includes six categories: raw water fluoride concentration data, deep-treated effluent fluoride concentration data, suspended particle size distribution data, stirring motor current data, settling liquid level height data, and resource recovery mother liquor fluoride concentration data. Periodic sampling is performed on the independent acquisition channels corresponding to the six types of process data with a unified sampling period. The six types of process data are acquired in each sampling period, and a corresponding sampling time mark is attached to each type of process data to obtain process data with sampling time mark. Synchronously combine various types of process data acquired under the same sampling time marker and write them into the same monitoring record in a preset order to form a monitoring data point containing six types of process data; The monitoring data points obtained from multiple consecutive sampling cycles are sequentially arranged according to the order of the sampling time markers. The sequentially arranged monitoring data points are then connected to form an original monitoring data sequence covering the operation of the fluoride deep treatment device and the resource recovery device.

3. The method for achieving deep treatment and resource recovery of fluorides according to claim 2, characterized in that: In S2, each monitoring data point in the original monitoring data sequence is analyzed one by one. The sampling values ​​of the six types of process data in the corresponding monitoring data point are checked to see if they fall within the corresponding physical allowable range and measurement range. If any sampling value exceeds the corresponding range, the process data of the corresponding category is marked as obvious noise and removed. After removing the process data of the corresponding category that is marked as obvious noise, check whether each monitoring data point still contains the sampling values ​​corresponding to the six categories of process data. If the sampling values ​​of any category are missing, the corresponding monitoring data point is removed as a whole to form a monitoring data sequence. According to the preset time window size, the monitoring data sequence is segmented according to the order of sampling time markers. Monitoring data points whose sampling time falls within the same time window range are included in a time window segment, and a corresponding window identifier is assigned to each time window segment. Within each time window, the first monitoring data point sorted by sampling time within the time window is taken as the starting point, and the last monitoring data point within the time window is taken as the ending point. The difference in the sampling values ​​of the six types of process data between the starting point and the ending point is calculated respectively. The corresponding trend is determined based on the magnitude of the differences between various process data. When the difference is greater than zero, it is recorded as an upward trend; when the difference is equal to zero, it is recorded as a constant trend; and when the difference is less than zero, it is recorded as a downward trend. Within the same time window, for all sampled values ​​of the stirring motor current data, the difference between the upper and lower limits is calculated to obtain the current fluctuation amplitude. The candidate feature data set for the corresponding time window segment is composed of the differences of six types of process data, the corresponding trends of change, and the current fluctuation amplitude. The candidate feature data set includes difference features, trend features, and current fluctuation amplitude features.

4. The method for achieving deep treatment and resource recovery of fluorides according to claim 3, characterized in that: S2 also includes, within each time window, filtering the difference features, trend features, and current fluctuation amplitude features in the candidate feature data set against preset threshold conditions for the deep governance response stage, the deep governance and resource recovery transition stage, and the resource recovery separation stage, respectively, in accordance with the stage division conditions. For the difference feature, the difference of the six types of process data is compared with the upper and lower limits of the difference threshold interval preset for each stage in the stage division conditions. When the value of a difference feature falls into the difference threshold interval corresponding to any stage, the corresponding difference feature is recorded as a difference feature that meets the stage division conditions; otherwise, the corresponding difference feature is removed from the candidate feature data set. For the trend characteristics, the trend corresponding to the six types of process data is compared with the set of allowed trend characteristics preset for each stage in the stage division conditions. When a trend characteristic belongs to the set of allowed trend characteristics for any stage, the corresponding trend characteristic is recorded as a trend characteristic that meets the stage division conditions; otherwise, the corresponding trend characteristic is removed from the candidate feature data set. For the current fluctuation amplitude characteristics, the current fluctuation amplitude is compared with the upper and lower limits of the current fluctuation amplitude threshold range preset for each stage in the stage division conditions. When the current fluctuation amplitude value falls into the current fluctuation amplitude threshold range corresponding to any stage, the corresponding current fluctuation amplitude characteristic is recorded as a current fluctuation characteristic that meets the stage division conditions; otherwise, the corresponding current fluctuation amplitude characteristic is removed from the candidate feature data set. After filtering the difference characteristics, trend characteristics, and current fluctuation amplitude characteristics, for each time window segment, the filtered and retained difference characteristics, trend characteristics, and current fluctuation amplitude characteristics are used together as the characteristic data for stage judgment of the corresponding time window segment.

5. A method for achieving deep treatment and resource recovery of fluorides according to claim 4, characterized in that: In S3, the feature data corresponding to each time window segment for stage judgment is read sequentially according to the order of the time window identifiers. All time window segments are arranged in order to form a window processing sequence, and a preset initial stage identifier is set for the first time window segment in the window processing sequence as the previous stage identifier. When processing the current time window segment in the window processing sequence, the difference features, trend features and current fluctuation amplitude features in the feature data used for stage judgment in the current time window segment are classified into difference feature set, trend feature set and current fluctuation amplitude feature set respectively. For each stage in the deep governance reaction stage, the deep governance and resource recovery transition stage, and the resource recovery separation stage, the number of difference features whose values ​​fall into the corresponding difference threshold interval of the current time window segment is counted in the difference feature set; the number of change trend features whose change trends belong to the allowable change trend set of the corresponding stage is counted in the change trend feature set; and the number of current fluctuation amplitude features whose values ​​fall into the corresponding current fluctuation amplitude threshold interval of the corresponding stage is counted in the current fluctuation amplitude feature set. The cumulative feature number of the corresponding stage is obtained by summing the number of features that meet the conditions of the difference threshold interval, the allowable change trend set, and the current fluctuation amplitude threshold interval. Among the cumulative feature counts of the three stages, the stage with the upper limit of the cumulative feature count is selected as the candidate stage of the current time window. When the cumulative feature count of the stage with the upper limit is greater than zero and there is only one stage with the maximum cumulative feature count, the stage identifier corresponding to the candidate stage with the upper limit is determined as the stage identifier of the current time window. When the cumulative feature count of all three stages is zero, or when there are two or more stages with the same cumulative feature count at the upper limit, no stage switching is performed on the current time window segment, and the previous stage identifier is directly used as the stage identifier of the current time window segment. Write the stage identifier of the current time window segment into the stage identifier sequence, and update the stage identifier of the current time window segment to the previous stage identifier of the next time window segment. Repeat the stage identifier identification process for the remaining time window segments in the window processing sequence until the stage identifier of all time window segments has been determined.

6. The method for achieving deep treatment and resource recovery of fluorides according to claim 5, characterized in that: In S4, after determining the stage identifier corresponding to the current time window segment, the monitoring data points within the current time window segment are read. The starting and ending monitoring data points are determined according to the order of the sampling time markers. The sampling values ​​corresponding to the fluoride ion concentration data, suspended particle size distribution data, sedimentation liquid level height data, and fluoride ion concentration data of the resource recovery mother liquor in the starting and ending monitoring data points are extracted respectively. Based on the sampled values ​​between the initial and final monitoring data points, the changes in fluoride ion concentration, the rate of change in concentration per unit time, the change in suspended particle size, the particle size change trend, the settling velocity, and the change in fluoride ion concentration in the mother liquor from resource recovery were calculated, including: The change in fluoride ion concentration is the difference between the fluoride ion concentration samples in the effluent from the deep-treated water at the end of the monitoring data point and the beginning of the monitoring data point. The rate of change of concentration per unit time is the ratio of the change in fluoride ion concentration to the difference in sampling time between the two monitoring data points; The change in suspended particle size is the difference in particle size statistics between the end monitoring data point and the beginning monitoring data point. The trend of particle size change is determined as an upward trend, a constant trend, or a downward trend based on the sign of the change in suspended particle size; when the change in suspended particle size is greater than zero, it is determined as an upward trend; when the change in suspended particle size is equal to zero, it is determined as a constant trend; and when the change in suspended particle size is less than zero, it is determined as a downward trend. The settling velocity is the ratio of the difference in liquid level between the initial and final monitoring data points to the difference in sampling time. The change in fluoride ion concentration in the mother liquor from resource recovery is the difference between the fluoride ion concentration in the mother liquor at the starting and ending monitoring data points. The changes in fluoride ion concentration, the rate of change in concentration per unit time, the change in suspended particle size, the settling velocity, and the change in fluoride ion concentration in the mother liquor from resource recovery were used as numerical features; the trend of particle size change was used as a trend feature.

7. A method for achieving deep treatment and resource recovery of fluorides according to claim 6, characterized in that: S4 also includes, when the current stage is identified as the deep treatment reaction stage, comparing the corresponding numerical features and trend features with the preset semantic type determination conditions based on the change in fluoride ion concentration, the rate of change in concentration per unit time, the change in suspended particle size, and the particle size change trend. Each semantic type corresponds to at least one set of semantic determination sub-conditions consisting of a set of numerical threshold intervals or a set of change trends. When all the semantic determination sub-conditions corresponding to a certain semantic type are satisfied, the semantic result of the reaction stage for the current time window segment is recorded. When the current stage is identified as the resource recovery and separation stage, the corresponding numerical features are compared with the preset semantic type judgment conditions based on the sedimentation rate and the change in fluoride ion concentration of the resource recovery mother liquor. When all semantic judgment sub-conditions corresponding to any semantic type are met, the semantic result of the separation stage of the corresponding semantic type for the current time window is recorded. When the current stage identifier belongs to the transition stage between deep governance and resource recycling, semantic type determination is not performed, and an empty semantic result set is directly written to the semantic result sequence. The semantic results corresponding to each time window segment are sorted according to the time order of the time window segment to obtain a sequence of semantic results arranged in chronological order.

8. The method for achieving deep treatment and resource recovery of fluorides according to claim 7, characterized in that: In S5, after determining the stage identifier and semantic result of the current time window segment, when the stage identifier belongs to the deep treatment reaction stage, the sampled value of the fluoride ion concentration in the deep treatment effluent is read from the monitoring data point corresponding to the current time window segment, and the sampled value is compared with the target concentration range set for the effluent ion concentration in the preset fluoride deep treatment target. When the effluent ion concentration is higher than the upper limit of the target concentration range, the effluent concentration deviation is recorded as a positive deviation. When the effluent ion concentration is lower than the lower limit of the target concentration range, the effluent concentration deviation is recorded as a negative deviation. When the effluent ion concentration falls within the target concentration range, the effluent concentration deviation is recorded as zero deviation. When the current stage is identified as the resource recovery and separation stage, the settling velocity and the change in fluoride ion concentration of the mother liquor in the resource recovery are read from the monitoring data points corresponding to the current time window. These data are then compared with the corresponding target intervals in the preset resource recovery targets. When the settling velocity is lower than the lower limit of the target interval, the settling velocity deviation is recorded as a negative deviation; when the settling velocity is higher than the upper limit of the target interval, the settling velocity deviation is recorded as a positive deviation; and when the settling velocity falls within the target interval, the settling velocity deviation is recorded as zero deviation. Similarly, when the change in fluoride ion concentration of the mother liquor is lower than the lower limit of the target interval, the mother liquor concentration change deviation is recorded as a negative deviation; when the change in fluoride ion concentration of the mother liquor is higher than the upper limit of the target interval, the mother liquor concentration change deviation is recorded as a positive deviation; and when the change in fluoride ion concentration of the mother liquor falls within the target interval, the mother liquor concentration change deviation is recorded as zero deviation. Based on the directional markers of effluent concentration deviation, settling velocity deviation, and mother liquor concentration change deviation, the strategy optimization requirements for the current time window are determined. For each directional marker, strategy optimization requirements are generated according to the following rules, and the deviation index type corresponding to that directional marker is recorded as the current deviation index: When the effluent concentration deviation is a positive deviation, the strategy optimization requirement will be recorded as enhancing the deep treatment effect, and the deviation index type will be recorded as the effluent concentration deviation index. When the effluent concentration deviation is negative, the strategy optimization requirement will be recorded as weakening the deep treatment effect, and the deviation index type will be recorded as the effluent concentration deviation index. When the settlement velocity deviation is negative, it will be recorded as an enhanced settlement condition in the strategy optimization requirements, and the deviation index type will be recorded as the settlement velocity deviation index. When the settlement velocity deviation is positive, it will be recorded as a condition to reduce settlement in the strategy optimization requirements, and the deviation index type will be recorded as the settlement velocity deviation index. When the deviation of the mother liquor concentration change is negative, the strategy optimization requirement will be recorded as enhancing resource recovery capability, and the deviation index type will be recorded as the mother liquor concentration change deviation index. When the deviation of the mother liquor concentration change is a positive deviation, the strategy optimization requirement will be recorded as a weakening of resource recovery capability, and the deviation index type will be recorded as the mother liquor concentration change deviation index. When any deviation indicator is zero, no strategy optimization requirement is generated for the corresponding deviation indicator.

9. A method for achieving deep treatment and resource recovery of fluorides according to claim 8, characterized in that: S5 also includes retrieving strategy optimization records from a pre-built strategy library that match each strategy optimization requirement in the strategy optimization requirements; the strategy optimization record includes at least the deviation index type, applicable stage identifier, strategy optimization category, parameter name, parameter adjustment direction, and parameter adjustment step size; The deviation index type is used to characterize which deviation source the strategy optimization record applies to, including at least one of the following: effluent concentration deviation index, sedimentation velocity deviation index, and mother liquor concentration change deviation index; The applicable stage identifier is used to indicate the applicable operational stage of the corresponding strategy optimization record, including at least one of the deep governance reaction stage or the resource recovery and separation stage; The strategy optimization category is used to indicate which strategy optimization requirement the corresponding strategy optimization record is used to respond to, including at least one of the following: enhancing the role of deep governance, weakening the role of deep governance, enhancing settlement conditions, weakening settlement conditions, enhancing resource recovery capacity, or weakening resource recovery capacity. The parameter name indicates the specific control parameters to be adjusted in this strategy optimization record, including at least one of the following: dosage, dosing interval, stirring speed, settling time, start and stop timing of resource recovery unit, recovery flow rate, crystal classification method, and mother liquor discharge and reflux distribution ratio; The parameter adjustment direction is used to identify the specific direction of adjustment performed on the parameter, such as increasing, decreasing, shortening, extending, raising, lowering, advancing, or delaying. The parameter adjustment step size is used to specify the adjustment range for each adjustment of the corresponding parameter. During the retrieval process, when the deviation indicator type in a certain strategy optimization record is consistent with the current deviation indicator, the applicable stage identifier is consistent with the current stage identifier, and the strategy optimization category is consistent with the current strategy optimization requirement, the corresponding strategy optimization record will be added to the strategy candidate set. Read each policy optimization record from the policy candidate set and execute parameter adjustment operations sequentially according to the order of the policy optimization records in the policy candidate set; for each policy optimization record, perform a deterministic adjustment process on the specified parameter based on the parameter name, parameter adjustment direction, and parameter adjustment step size in the policy optimization record, specifically including: When the parameter name is dosage and the parameter adjustment direction is increasing, the current dosage is added to the parameter adjustment step size to obtain the new dosage; when the parameter name is dosage and the parameter adjustment direction is decreasing, the current dosage is subtracted from the parameter adjustment step size to obtain the new dosage. When the parameter name is dosing interval and the parameter adjustment direction is shortening, the current dosing interval is subtracted from the parameter adjustment step size to obtain the new dosing interval; when the parameter name is dosing interval and the parameter adjustment direction is extending, the current dosing interval is added to the parameter adjustment step size to obtain the new dosing interval. When the parameter name is stirring speed and the parameter adjustment direction is increasing, the current stirring speed is added to the parameter adjustment step size to obtain the new stirring speed; when the parameter name is stirring speed and the parameter adjustment direction is decreasing, the current stirring speed is subtracted from the parameter adjustment step size to obtain the new stirring speed. When the parameter name is Settlement Residence Time and the parameter adjustment direction is to extend, the current Settlement Residence Time is added to the parameter adjustment step size to obtain the new Settlement Residence Time; when the parameter name is Settlement Residence Time and the parameter adjustment direction is to shorten, the current Settlement Residence Time is subtracted from the parameter adjustment step size to obtain the new Settlement Residence Time. When the parameter name is the start / stop timing of the resource recycling unit and the parameter adjustment direction is advanced, the current start / stop time is subtracted from the parameter adjustment step size to obtain the new start / stop time; when the parameter name is the start / stop timing of the resource recycling unit and the parameter adjustment direction is delayed, the current start / stop time is added to the parameter adjustment step size to obtain the new start / stop time. When the parameter name is "Recovery Flow" and the parameter adjustment direction is "Increase", the current recovery flow is added to the parameter adjustment step size to obtain the new recovery flow; when the parameter name is "Recovery Flow" and the parameter adjustment direction is "Decrease", the current recovery flow is subtracted from the parameter adjustment step size to obtain the new recovery flow. When the parameter name is Crystal Classification Method, the parameter adjustment direction will move one level forward or backward in the preset classification method sequence according to the index order, and the moved classification method will be used as the new crystal classification method. When the parameter name is Mother Liquor Discharge and Recirculation Distribution Ratio and the parameter adjustment direction is to increase the recirculation ratio, the current recirculation ratio is added to the parameter adjustment step size and limited to the allowable range to obtain a new recirculation ratio. At the same time, the discharge ratio is updated to one minus the new recirculation ratio. When the parameter name is Mother Liquor Discharge and Recirculation Distribution Ratio and the parameter adjustment direction is to decrease the recirculation ratio, the current recirculation ratio is subtracted from the parameter adjustment step size and limited to the allowable range to obtain a new recirculation ratio. At the same time, the discharge ratio is updated to one minus the new recirculation ratio. After adjusting the parameters corresponding to a single strategy optimization record, the adjusted parameter values ​​are compared with the allowed value range of the corresponding parameters item by item. When the adjusted parameter value is greater than the upper limit of the allowed value range of the corresponding parameter, the adjusted parameter value is set to the upper limit of the allowed value range of the corresponding parameter; when the adjusted parameter value is less than the lower limit of the allowed value range of the corresponding parameter, the adjusted parameter value is set to the lower limit of the allowed value range of the corresponding parameter. When multiple policy optimization records in the policy candidate set point to the same parameter, the absolute value operation is performed on the deviation value corresponding to each policy optimization record to obtain the absolute value of the deviation. The magnitudes of all absolute deviation values ​​are compared, and the policy optimization record with the absolute value of the deviation at the upper limit is selected as the current policy optimization record to be executed. The parameter adjustment operation is then performed on the parameter corresponding to the currently executed policy optimization record. Other policy optimization records that point to the same parameter are not executed.

Images