Sewage treatment AI management platform and data-driven intelligent control method

Abstract

The application discloses a sewage treatment AI management platform and a data-driven intelligent regulation and control method, relates to the technical field of environmental engineering, and comprises the following steps: collecting continuous water quality change records in an inflow load mutation stage, retaining original fluctuation amplitudes and change rates in chronological order, setting mutation change marks at the end of the continuous water quality change records, and forming original change tracks without smoothing treatment.The application retains original fluctuation tracks in the inflow mutation stage, constructs impact change tracks, determines a toxicity influence starting position in combination with microbial activity changes, realizes accurate identification of toxicity impact, determines an intervention starting point through time backtracking, rearranges dosing rhythm, implements alternating regulation of inhibition and recovery in the impact duration stage, and thus controls a toxicity influence range and improves operation adaptive capacity.

Description

Technical Field

[0001] This invention relates to the field of environmental engineering technology, specifically to an AI management platform for wastewater treatment and a data-driven intelligent control method. Background Technology

[0002] The wastewater treatment AI management platform and data-driven intelligent control refer to the construction of a comprehensive management platform that integrates data acquisition, status perception, intelligent analysis, and dynamic adjustment on the basis of the traditional wastewater treatment process operation system. This platform collects multi-source data in real time, including influent water quality parameters, flow rate changes, dissolved oxygen, sludge concentration, energy consumption indicators, and equipment operating status. This data is then aggregated, cleaned, and structured using big data processing technology. Based on this, artificial intelligence methods such as machine learning, trend prediction, and correlation analysis are used to identify process load fluctuations, changes in microbial activity, and abnormal operational risks. Furthermore, it enables dynamic control decisions for key operating parameters such as aeration intensity, return flow ratio, chemical dosage, and sludge discharge cycle, feeding control commands back to the on-site execution system. This transforms the wastewater treatment process from being driven by human experience to being driven by data models, thereby constructing an intelligent operation management system with continuous perception, prediction, and adaptive optimization capabilities.

[0003] The existing technology has the following shortcomings: Under current technological conditions, intelligent wastewater treatment management systems typically use outlier identification mechanisms to automatically filter collected data. When a short-term toxic shock occurs in the influent, its water quality indicators may exhibit sudden and drastic fluctuations, which the system may easily interpret as sensor acquisition errors or occasional noise and automatically delete. Because this true toxicity signal is not effectively identified, the system will continue to maintain the original dosing ratio and operating parameters according to the established control logic, failing to promptly increase the buffer or emergency adjustment intensity. This can easily lead to continuous toxic inhibition or even poisoning and inactivation of microorganisms in the reaction tank, thereby causing the rapid spread of sludge bulking, resulting in deterioration of effluent water quality and posing a significant operational risk.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide an AI management platform for wastewater treatment and a data-driven intelligent control method to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a wastewater treatment AI management platform and a data-driven intelligent control method, comprising the following steps: Collect continuous water quality change records during the abrupt change phase of influent load, retain the original fluctuation amplitude and change rate in chronological order, and set abrupt change marker at the end of the continuous water quality change record to form the original change trajectory without smoothing. Based on the abrupt change markers, the original change trajectory is processed by time expansion. The fluctuations of each time period in the original change trajectory are reordered according to the rate of change, and the impact segments that deviate from the daily operating rhythm are extracted to form the impact change trajectory. Based on the impact change trajectory, the operation data of the reaction tank were synchronously compared to screen the time period in which the impact change trajectory and the decrease in microbial activity occurred simultaneously, and the starting point of the toxicity effect was determined. A time-backtracking analysis of the impact trajectory was conducted around the initial location of the toxicity effect to calculate the time difference between the water quality surge stage and the existing dosing rhythm, and to determine the intervention starting point. The frequency and dosage of subsequent drug administration are adjusted based on the starting point of intervention. During the sustained impact phase, the single dosage is gradually reduced and the recovery rhythm is released intermittently. The operating rhythm is rearranged by alternating between the inhibition and recovery phases to control the range of toxic effects.

[0007] Preferably, the steps for forming the original change trajectory are as follows: The water quality parameters of the influent are continuously collected and recorded in chronological order. The changes and rates of change between adjacent time points are recorded simultaneously to form a continuous water quality change record that includes timestamps, original fluctuation amplitudes and rates of change. The rate of change in continuous water quality change records is compared point by point. The time range of the load change stage is determined based on the difference in the rate of change, and the set of continuous water quality change records corresponding to the load change stage is extracted. A sudden change marker is set at the end of the continuous water quality change record set. The sudden change marker is associated with the time series number, and all continuous water quality change records before the sudden change marker are retained. The entire set of continuous water quality change records is sealed around the abrupt change markers, keeping the original fluctuation amplitude and change rate unchanged, forming the original change trajectory without smoothing in chronological order.

[0008] Preferably, the continuous water quality change records are not processed by moving average, trend line fitting, or fluctuation reduction during the recording process. The abrupt change markers serve as the endpoint identifiers of the load abrupt change phase and maintain a corresponding relationship with the time series number. The original fluctuation amplitude and change rate are kept intact in the continuous water quality change record set.

[0009] Preferably, the time-expanding process for the original change trajectory based on abrupt change markers is as follows: Using abrupt change markers as the reference points for time expansion, bidirectional time expansion is performed on time nodes before and after the abrupt change markers, with the addition of time distance parameters, forming a time expansion trajectory that includes time distance parameters, original fluctuation amplitude, and rate of change. The time nodes in the time-distributed trajectory are reordered according to their rate of change, while maintaining the time distance parameter in sync with the original fluctuation amplitude, forming a sequence of change rates arranged according to their rate of change. Based on the rate of change sequence, a daily operating rhythm interval is constructed, continuous time nodes that exceed the daily operating rhythm interval are collected, and impact candidate segments are sorted out in combination with time distance parameters. The time sequence around the candidate impact segment is restored according to the time distance parameter, the original fluctuation amplitude and change rate are preserved, and the impact change trajectory is formed with the abrupt change marker as the time reference point.

[0010] Preferably, the collection of impact candidate segments is combined with the time distance parameter for continuity determination. Time nodes with continuous time distance parameters and a rate of change exceeding the daily operating rhythm range are identified as impact segments, and the continuous correlation structure of impact segments in the time unfolding trajectory is maintained around the abrupt change marker.

[0011] Preferably, the steps for determining the initiation site of toxic effects are as follows: Data on the operation of the reaction tank within the corresponding time range were extracted from all time points in the shock change trajectory. The microbial activity characterization parameters were time-aligned with the original fluctuation amplitude and rate of change in the shock change trajectory to form a synchronous control sequence. By tracking the changing trends of microbial activity characterization parameters around the time nodes in the synchronous control sequence, the time segments of continuous decline in microbial activity characterization parameters are identified and associated with the corresponding time nodes in the shock change trajectory to form candidate time segments; Based on the candidate time intervals, the continuous time nodes in the impact change trajectory where the rate of change is within the impact interval and the microbial activity characterization parameters continue to decrease are grouped into time intervals that occur synchronously. By tracing back along the trajectory of the impact change from the starting point of the synchronous time segment to the time point when the rate of change first entered the impact segment range, the starting position of the toxic effect was determined and marked.

[0012] Preferably, around each time node in the synchronous time segment, combined with the distribution of change rate in the impact change trajectory, the duration of continuous decline in microbial activity characterization parameters is continuously organized, and the earliest time node where the change rate is continuously within the impact segment range and the microbial activity characterization parameters continue to decline is determined as the starting position of toxicity effect.

[0013] Preferably, the steps for performing a time-backtracking analysis of the impact trajectory around the initial location of the toxic effects are as follows: Starting from the initial location of the toxicity effect, we traced back along the impact trajectory node by node to determine the time point when the rate of change entered the impact zone and define the time interval of the water quality surge stage. Existing dosing rhythm records corresponding to the time interval of the water quality surge stage were extracted, and the time nodes of the dosing action were aligned with the time nodes of the impact change trajectory to form a time mapping relationship; Based on the time interval between the start time of the water quality surge and the start location of the toxicity effect, the time nodes of the dosing actions in the existing dosing rhythm records are compared to form a time difference sequence between the water quality surge and the existing dosing rhythm. By combining the time difference sequence with the time node corresponding to the starting position of toxic effects, the starting time position of the intervention can be determined.

[0014] Preferably, the time difference between the starting time of the sudden rise in water quality and the time of the most recent dosing action in the existing dosing rhythm is extracted, and the time point corresponding to the starting position of the toxicity effect is compared with the time point, and the time point not covered by the existing dosing rhythm is selected as the intervention starting point.

[0015] Preferably, the steps for adjusting the subsequent dosing frequency and dosage based on the intervention starting point are as follows: The impact duration phase is divided by using the intervention start point as the control benchmark time node, and the existing drug administration rhythm records are rearranged around the intervention start point to form a new drug administration rhythm sequence. A progressive adjustment rule is set up around the dosing action during the continuous impact phase. The single dosing amount is reduced in chronological order and the dosing interval is redistributed so that the single dosing amount and the dosing interval correspond to each other. Based on the progressive adjustment rules, the inhibition phase and the recovery phase are divided. During the inhibition phase, the dosing frequency is kept concentrated, and during the recovery phase, the dosing interval is widened, forming an alternating structure of inhibition and recovery phases. By arranging the frequency and dosage of drug administration during the sustained impact phase around the alternating structure of the inhibition and recovery phases, a rhythmic rearrangement trajectory with the intervention starting point as the control benchmark is formed, thereby controlling the range of toxic effects.

[0016] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention preserves the original fluctuation amplitude and rate of change during the abrupt change phase of the influent load, and constructs the original unsmoothed change trajectory using the abrupt change marker as a time reference. This prevents short-term violent fluctuations from being reduced or replaced, thus preserving the temporal structure characteristics of the real impact process. Furthermore, it extracts the impact segment by reordering the time expansion and rate of change, and determines the starting position of the toxicity effect by combining it with the operating data of the reaction tank. This establishes a direct temporal correlation between abnormal fluctuations in the influent and changes in microbial activity, improving the accuracy and targetedness of toxicity impact identification and avoiding the neglect of risk signals that lead to control lag.

[0017] This invention uses the starting point of toxicity impact as the core time anchor point. Through time backtracking analysis, it clarifies the time difference between the stage of sudden water quality rise and the existing dosing rhythm. Using the intervention starting point as the control benchmark, it rearranges the rhythm of subsequent dosing frequency and dosage. During the impact duration, it adopts an alternating adjustment method of inhibition and recovery stages, transforming the operating rhythm from a fixed mode to a dynamic rhythm structure. This limits the range of toxicity impact in the time dimension, reduces the duration of microbial activity inhibition, stabilizes the effluent state, and enhances the adaptive control capability of the operation process. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0019] Figure 1 This is a flowchart of the wastewater treatment AI management platform and data-driven intelligent control method of the present invention. Detailed Implementation

[0020] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0021] This invention provides, for example Figure 1 The wastewater treatment AI management platform and data-driven intelligent control method shown include the following steps: Collect continuous water quality change records during the abrupt change phase of influent load, retain the original fluctuation amplitude and change rate in chronological order, and set abrupt change marker at the end of the continuous water quality change record to form the original change trajectory without smoothing. The original, unsmoothed change trajectory was constructed from continuous water quality change records around the abrupt change in influent load. The specific implementation steps are as follows: Throughout the entire process of influent entering the treatment process, influent water quality parameters are continuously and uninterruptedly collected. Water quality values ​​at each collection point are recorded sequentially according to time. During recording, the changes in values ​​between adjacent time points and the rate of change per unit time are calculated simultaneously. Water quality values, changes, and rates of change are bound and stored using the same time series numbering method, ensuring that each time point simultaneously contains the original water quality value, the original fluctuation amplitude, and the rate of change information. This forms a continuous water quality change record containing timestamps, original fluctuation amplitudes, and rates of change. During this process, no moving average processing, trend line fitting, or fluctuation reduction processing is applied to the collected water quality values, ensuring that the original fluctuation amplitudes are fully preserved. Furthermore, an incremental time approach ensures that the continuous water quality change record possesses an irreversible sequential structure in the time dimension, laying the foundation for subsequent abrupt change identification.

[0022] Based on the formation of continuous water quality change records, the rate of change in the continuous water quality change records is compared point by point to identify the time period in which the rate of change jumps concentrated. The difference in the rate of change between consecutive time nodes is used as the trigger for the sudden change. When the rate of change of adjacent time nodes continuously exceeds the preset stable interval, the corresponding time node is marked as the starting interval of the load sudden change phase. At the same time, the rate of change of subsequent time nodes is continuously tracked until the rate of change returns to the normal operating rhythm interval. The ending interval of the load sudden change phase is determined at the return time node. The complete time range of the load sudden change phase is locked in the above method. After locking the load sudden change phase, all continuous water quality change records within this time range are extracted as the continuous water quality change record set of the influent load sudden change phase, so that the original fluctuation amplitude and rate of change within the sudden change phase are maintained in their original arrangement on the time axis.

[0023] A sudden change marker is added to the end of the continuous water quality change record set of the extracted influent load sudden change phase. This sudden change marker is associated with the time sequence number of the continuous water quality change record, so that the sudden change marker points to the time node corresponding to the end of the load sudden change phase. The sudden change marker is retained in the continuous water quality change record set as an independent identifier. At the same time, all continuous water quality change records before the sudden change marker are completely retained in chronological order, so that the sudden change marker becomes the time boundary that distinguishes the daily operation phase from the load sudden change phase. By setting a sudden change marker at the end of the continuous water quality change record, the original fluctuation amplitude and change rate form a complete sudden change trajectory structure in the time dimension, thereby realizing the time sealing of the entire process of the load sudden change phase.

[0024] After the abrupt change markers are set, the continuous water quality change records containing the abrupt change markers are sealed as a whole. During the sealing process, the original fluctuation amplitude and rate of change remain unchanged. At the same time, smoothing replacement, trend reduction, or abnormal deletion operations on the water quality values ​​within the sealed interval are prohibited. This ensures that the water quality values, changes, and rates of change within this time interval are continuously preserved in an unsmoothed form, forming a continuously extending original change trajectory in chronological order. In this original change trajectory, any time point can be traced back to the corresponding original water quality value and rate of change. The abrupt change marker is located at the end of the original change trajectory and serves as the endpoint marker of the load abrupt change phase. This completes the process of preserving continuous water quality change records during the influent load abrupt change phase, ensuring that the original fluctuation amplitude and rate of change remain completely traceable within the abrupt change phase, and forming an unsmoothed original change trajectory. This provides the basic data for subsequent time-based processing around the abrupt change markers.

[0025] Based on the abrupt change markers, the original change trajectory is processed by time expansion. The fluctuations of each time period in the original change trajectory are reordered according to the rate of change, and the impact segments that deviate from the daily operating rhythm are extracted to form the impact change trajectory. Based on the original change trajectory containing abrupt change markers, the original change trajectory is subjected to time expansion processing to extract the impact segment. The specific implementation steps are as follows: Using abrupt change marker as the reference point for time unfolding, all time nodes before the abrupt change marker are subjected to reverse time unfolding, and time nodes after the abrupt change marker are subjected to forward time unfolding. This allows the original change trajectory to form a bidirectional extended time unfolding sequence around the abrupt change marker. During the time unfolding process, the original fluctuation amplitude and change rate in the original change trajectory are kept constant. At the same time, a time distance parameter is added to each time node. The time distance parameter represents the time interval between the current time node and the abrupt change marker. By introducing the time distance parameter, the original change trajectory forms a continuous time unfolding structure centered on the abrupt change marker in the time dimension. This constructs a time unfolding trajectory that includes the time distance parameter, the original fluctuation amplitude, and the change rate, laying the foundation for subsequent reordering according to the change rate.

[0026] Based on the established time-distributed trajectory, the time nodes in the trajectory are layered and organized according to their rate of change. All time nodes are arranged according to the magnitude of their rate of change, while maintaining the synchronous movement of the time distance parameter corresponding to each time node with the original fluctuation amplitude information. This rearrangement ensures that the sequence reflects both the distribution of the rate of change and the original temporal positional relationships. Through this rearrangement, the fluctuation intensity structure hidden beneath the time sequence in the original trajectory is revealed. In this process, the original fluctuation amplitude and rate of change are not changed; only the arrangement order is reconstructed, resulting in a rate of change sequence rearranged according to the rate of change, forming a distributed matrix structure corresponding to the time distance parameter.

[0027] After reordering according to the rate of change, the daily operating rhythm interval is used as a reference benchmark to perform rhythm deviation analysis on the reordered rate of change sequence. By extracting the distribution range of the rate of change in the stable operating phase in the time unfolded trajectory, the daily operating rhythm interval is constructed. The reordered rate of change sequence is compared with the daily operating rhythm interval, and continuous time nodes that exceed the daily operating rhythm interval are collected as impact candidate segments. At the same time, the time continuity of the impact candidate segments is sorted in combination with the time distance parameter, so that the impact candidate segments form continuous segments in the time dimension. Discrete fluctuation segments that are too far away from the abrupt change marker are further eliminated. Finally, the impact segments that deviate from the daily operating rhythm are extracted, so that the impact segments not only meet the condition that the rate of change exceeds the daily operating rhythm interval, but also form a continuous association with the abrupt change marker in the time unfolded structure.

[0028] After extracting the impact segments, the time nodes corresponding to the impact segments are restored to their chronological order according to the time distance parameter, and the original fluctuation amplitude and rate of change information of each time node are retained, so that the impact segments form a continuous impact change trajectory in the time dimension. The impact change trajectory uses the abrupt change marker as the time reference point, covering the starting interval of the load abrupt change stage forward and the attenuation interval of the load abrupt change stage backward, thus forming a complete impact change trajectory structure. In this impact change trajectory, the original fluctuation amplitude and rate of change remain in an unsmoothed state, the time order is restored, and the fluctuation intensity level revealed by the reordering of the rate of change is mapped back to the time axis, realizing the structured extraction of the impact segments inside the original change trajectory. Finally, the time unfolding processing based on the abrupt change marker and the impact change trajectory construction process are completed, laying the data foundation for subsequent synchronous comparison of reaction pool operation data around the impact change trajectory.

[0029] Based on the impact change trajectory, the operation data of the reaction tank were synchronously compared to screen the time period in which the impact change trajectory and the decrease in microbial activity occurred simultaneously, and the starting point of the toxicity effect was determined. Based on the established impact change trajectory, in order to achieve synchronous comparison between the impact change trajectory and the reaction tank operation data, and to determine the starting location of the toxicity effect, the specific implementation steps are as follows: Around all time points included in the impact change trajectory, reaction tank operation data corresponding exactly to the time range of the impact change trajectory were extracted in chronological order. The microbial activity characterization parameters in the reaction tank operation data were time-aligned with the original fluctuation amplitude and change rate in the impact change trajectory, so that each time point of the impact change trajectory corresponds to the microbial activity status information at the same moment. During the time alignment process, the chronological order of the impact change trajectory was kept unchanged, and the microbial activity characterization parameters in the reaction tank operation data were presented in their original recorded state. This formed a synchronous control sequence with the impact change trajectory as the main line and the microbial activity status as the control object, so that the impact change trajectory and the reaction tank operation data established a one-to-one correspondence in the time dimension, providing a basis for subsequent screening of synchronously occurring time segments.

[0030] After forming a synchronous control sequence, the analysis started from the time points in the impact trajectory where the rate of change was within the impact segment. The trend of microbial activity characterization parameters at the corresponding time points was observed point by point. The time points in the impact trajectory were divided into the pre-impact, mid-impact, and post-impact segments. The microbial activity characterization parameters in each stage were continuously tracked. When the microbial activity characterization parameters began to show a continuous decline in a certain time segment of the impact trajectory, the continuous decline segment was identified and associated with the corresponding time point of the impact trajectory. This established a time-bound relationship between the microbial activity decline segment and the specific fluctuation position in the impact trajectory, thereby marking the candidate time segments that may produce toxic effects within the impact trajectory.

[0031] Based on the already marked candidate time segments, time nodes that completely overlap with the time segment of microbial activity decline in the impact change trajectory are centrally organized. Continuous time nodes in the impact change trajectory where the rate of change remains within the impact segment and the corresponding time node's microbial activity characterization parameters continuously decline are merged into synchronously occurring time segments. These synchronously occurring time segments are then distinguished by independent identifiers in the impact change trajectory. During the merging process, the temporal continuity is maintained, ensuring that the synchronously occurring time segments satisfy both the fluctuation characteristics of the impact change trajectory and the temporal continuity of microbial activity decline. This eliminates interference from single-point fluctuations or isolated time nodes, making the synchronously occurring time segments appear as continuous intervals in the time unfolding structure.

[0032] After completing the compilation of the time segments of synchronous occurrence, the starting time node of the synchronous occurrence time segment is used as the candidate starting position of toxicity effect. The time sequence of the impact change trajectory is traced back to the time node when the rate of change in the impact change trajectory first enters the impact segment range. By comparing the time relationship between this first entry time node and the starting time node of the synchronous occurrence time segment, the final time point of the starting position of toxicity effect is determined. After determining the starting position of toxicity effect, this time node is marked separately in the impact change trajectory, making the starting position of toxicity effect a key time anchor point connecting the impact change trajectory and the reaction tank operation data. This completes the synchronous comparison process of reaction tank operation data based on the impact change trajectory. By screening the time segments where the impact change trajectory and the decrease in microbial activity occur synchronously, the starting position of toxicity effect is accurately determined, providing a time basis for subsequent time backtracking analysis and intervention starting point determination around the starting position of toxicity effect.

[0033] A time-backtracking analysis of the impact trajectory was conducted around the initial location of the toxicity effect to calculate the time difference between the water quality surge stage and the existing dosing rhythm, and to determine the intervention starting point. Based on the established location of the onset of toxic effects and the clear temporal correspondence between the impact trajectory and the onset location of toxic effects, the specific implementation steps for determining a reasonable intervention starting point are as follows: Using the starting point of the toxic impact as the starting reference point for the time backtracking analysis, the analysis proceeds node by node along the time sequence of the impact change trajectory. All time nodes before the starting point of the toxic impact are continuously retrieved, and the original fluctuation amplitude and rate of change information in the impact change trajectory are kept unchanged during the backtracking process. At the same time, the time node when the rate of change in the impact change trajectory begins to enter the impact zone is marked, and this time node is defined as the starting time point of the water quality surge stage. After determining the starting time point of the water quality surge stage, the analysis continues to trace back along the impact change trajectory to the last time node when the rate of change remains within the normal operating rhythm range, so that the water quality surge stage forms a clear start and end boundary in the time dimension. Thus, a complete time backtracking interval is constructed with the starting point of the toxic impact as the end point and the starting time point of the water quality surge stage as the starting point.

[0034] Based on the clearly defined time interval of the water quality surge phase, existing dosing rhythm records corresponding to this time interval are extracted simultaneously. The dosing action time nodes in the existing dosing rhythm records are aligned one by one with the time nodes in the impact change trajectory, so that each time node of the water quality surge phase corresponds to the dosing action status at that time. During the alignment process, the dosing start time, dosing end time, and dosing interval time in the existing dosing rhythm records are arranged in chronological order and superimposed on the time axis of the water quality surge phase. Through this superposition method, a continuous mapping relationship is established between the water quality surge phase and the existing dosing rhythm in the time dimension, providing a complete time reference framework for subsequent time difference calculations.

[0035] After establishing a time mapping relationship between the water quality surge phase and the existing dosing rhythm, the time interval between the starting time of the water quality surge phase and the starting position of the toxicity effect is used to compare the time nodes of each dosing action in the existing dosing rhythm record point by point. The time difference between the starting time of the water quality surge phase and the time node of the most recent dosing action is calculated, as is the time difference between the starting position of the toxicity effect and the time node of the most recent dosing action. These time differences are then arranged in chronological order to form a continuous time difference sequence between the water quality surge phase and the existing dosing rhythm. During the formation of the time difference sequence, the time unit is kept consistent, and the original time marker in the impact change trajectory is kept consistent with the time marker in the existing dosing rhythm record, so that the time difference calculation results have continuity and traceability in the time dimension.

[0036] After completing the time difference sequence, the intervention starting point is determined based on the time difference between the start time of the water quality surge and the existing dosing rhythm, combined with the time node corresponding to the starting position of the toxicity effect. When there is an uncovered time segment between the start time of the water quality surge and the existing dosing rhythm, the start time node of the uncovered time segment is defined as the intervention starting point. When the starting position of the toxicity effect falls within the time segment after the existing dosing rhythm, the end time node of the most recent dosing action before the starting position of the toxicity effect is used as a reference, and the time node immediately following the reference time node is defined as the intervention starting point. In this way, the determination of the intervention starting point is based on both the time retrospective analysis results of the impact change trajectory and the calculation results of the time difference between the water quality surge and the existing dosing rhythm, thus forming a clear intervention starting point position in the time dimension. The intervention starting point becomes a key time node connecting the impact change trajectory, the water quality surge, and the existing dosing rhythm, providing a time basis for subsequent adjustments to the dosing frequency and dosage around the intervention starting point.

[0037] The frequency and dosage of subsequent drug administration were adjusted based on the intervention starting point as the control benchmark. During the continuous shock phase, the single dosage was gradually reduced and the recovery rhythm was released at intervals. The operation rhythm was rearranged by alternating between the inhibition and recovery phases to control the range of toxic effects. Based on the established intervention starting point and the time difference analysis between the sudden rise in water quality and the existing dosing schedule, the specific implementation steps are as follows to rearrange the operational schedule and control the scope of toxicity impact: Using the intervention start point as the control benchmark time node, the entire time interval after the intervention start point is divided into the impact duration phase and the recovery transition phase. A new dosing rhythm sequence is established between the intervention start point and the end of the impact duration phase. Based on the existing dosing rhythm records, the subsequent dosing frequency and dosage are rearranged, making the intervention start point the starting time marker of the new rhythm sequence. During the rearrangement process, the time order in the impact change trajectory remains unchanged, and the time distance between the starting position of toxicity effect and the intervention start point is used as a reference scale to rearrange the subsequent dosing actions on the time axis. This means that the dosing actions no longer follow the fixed interval method of the existing dosing rhythm, but are rearranged around the intervention start point, thereby constructing a control benchmark structure with the intervention start point as the core.

[0038] After rearranging the dosing rhythm based on the intervention starting point as the control benchmark, a progressive adjustment rule is set for each dosing action during the sustained impact phase. This causes the single dosing magnitude to decrease sequentially over time, while the dosing interval is redistributed in a progressive manner, ensuring a correspondence between the single dosing magnitude and the dosing interval. In the initial stage of the sustained impact phase, the dosing frequency is kept relatively concentrated to form the rhythm framework of the suppression phase. In subsequent time intervals, the dosing interval is gradually increased, while the single dosing magnitude is simultaneously compressed, resulting in a decreasing distribution of dosing actions over time. Throughout this process, the original fluctuation amplitude and rate of change recorded in the impact trajectory are used as a reference to ensure that the time position of the dosing action corresponds to the fluctuation segment in the impact trajectory. This ensures that the adjustment process of dosing frequency and dosage is synchronized with the sustained impact phase.

[0039] After the single dosage amplitude gradually decreases and forms an intermittent release structure, the impact duration phase is divided into two alternating segments: the suppression phase and the recovery phase. In the suppression phase, the dosage frequency is kept concentrated and the single dosage amplitude structure is maintained, so that the suppression phase forms a closed control segment against the toxic effects. In the recovery phase, the dosage interval is further increased, and the single dosage amplitude is kept in a stable range after the decrease, so that the recovery phase forms a rhythm buffer segment on the time axis. By alternating between the suppression phase and the recovery phase, the dosage action presents a periodic alternating structure in the time dimension, so that the operating rhythm in the impact duration phase changes from a single continuous rhythm to an alternating rhythm structure, thereby realizing the rearrangement of the operating rhythm.

[0040] After the alternating adjustment of the inhibition and recovery phases is completed, the frequency and dosage distribution of the drug administration throughout the entire impact duration are arranged as a whole, forming a complete rhythmic rearrangement trajectory from the intervention start point to the end of the impact duration. In this rhythmic rearrangement trajectory, the single dosage amplitude decreases successively, the drug administration interval gradually expands, and the inhibition and recovery phases alternate in chronological order, thus confining the toxic effects within the impact duration in the time dimension. Through the above rhythmic rearrangement method with the intervention start point as the control benchmark, a one-to-one correspondence is formed between the impact change trajectory and subsequent drug administration actions in the time dimension, ensuring that the change trajectory of drug administration frequency and dosage covers the key time segment after the onset of the toxic effects. This forms a control structure for the range of toxic effects at the operational rhythm level, completing the rhythmic rearrangement and toxic effect range control process of the impact duration.

[0041] This invention preserves the original fluctuation amplitude and rate of change during the abrupt change phase of the influent load, and constructs the original unsmoothed change trajectory using the abrupt change marker as a time reference. This prevents short-term violent fluctuations from being reduced or replaced, thus preserving the temporal structure characteristics of the real impact process. Furthermore, it extracts the impact segment by reordering the time expansion and rate of change, and determines the starting position of the toxicity effect by combining it with the operating data of the reaction tank. This establishes a direct temporal correlation between abnormal fluctuations in the influent and changes in microbial activity, improving the accuracy and targetedness of toxicity impact identification and avoiding the neglect of risk signals that lead to control lag.

[0042] This invention uses the starting point of toxicity impact as the core time anchor point. Through time backtracking analysis, it clarifies the time difference between the stage of sudden water quality rise and the existing dosing rhythm. Using the intervention starting point as the control benchmark, it rearranges the rhythm of subsequent dosing frequency and dosage. During the impact duration, it adopts an alternating adjustment method of inhibition and recovery stages, transforming the operating rhythm from a fixed mode to a dynamic rhythm structure. This limits the range of toxicity impact in the time dimension, reduces the duration of microbial activity inhibition, stabilizes the effluent state, and enhances the adaptive control capability of the operation process.

[0043] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A wastewater treatment AI management platform and data-driven intelligent control method, characterized in that, Includes the following steps: Collect continuous water quality change records during the abrupt change phase of influent load, retain the original fluctuation amplitude and change rate in chronological order, and set a sudden change marker at the end of the continuous water quality change record to form the original change trajectory; Based on the abrupt change markers, the original change trajectory is processed by time expansion. The fluctuations of each time period in the original change trajectory are reordered according to the rate of change, and the impact segments that deviate from the daily operating rhythm are extracted to form the impact change trajectory. Based on the impact change trajectory, the operation data of the reaction tank were synchronously compared to screen the time period in which the impact change trajectory and the decrease in microbial activity occurred simultaneously, and the starting point of the toxicity effect was determined. A time-backtracking analysis of the impact trajectory was conducted around the initial location of the toxicity effect to calculate the time difference between the water quality surge stage and the existing dosing rhythm, and to determine the intervention starting point. The frequency and dosage of subsequent drug administration are adjusted based on the starting point of intervention. During the sustained impact phase, the single dosage is gradually reduced and the recovery rhythm is released intermittently. The operating rhythm is rearranged by alternating between the inhibition and recovery phases to control the range of toxic effects.

2. The wastewater treatment AI management platform and data-driven intelligent control method according to claim 1, characterized in that, The steps for forming the original change trajectory are as follows: The water quality parameters of the influent are continuously collected and recorded in chronological order. The amount and rate of change between adjacent time points are also recorded to form a continuous water quality change record. The rate of change in continuous water quality change records is compared point by point. The time range of the load change stage is determined based on the difference in the rate of change, and the set of continuous water quality change records corresponding to the load change stage is extracted. A sudden change marker is set at the end of the continuous water quality change record set. The sudden change marker is associated with the time series number, and all continuous water quality change records before the sudden change marker are retained. The entire set of continuous water quality change records is sealed around the abrupt change markers, keeping the original fluctuation amplitude and change rate unchanged, forming the original change trajectory arranged in chronological order.

3. The wastewater treatment AI management platform and data-driven intelligent control method according to claim 2, characterized in that, During the recording process, the continuous water quality change records are not processed by moving average, trend line fitting, or fluctuation reduction. The abrupt change markers serve as the endpoint identifiers of the load abrupt change phase and maintain a corresponding relationship with the time series number. The original fluctuation amplitude and change rate are completely preserved in the continuous water quality change record set.

4. The wastewater treatment AI management platform and data-driven intelligent control method according to claim 2, characterized in that, The steps for performing time-expansion processing on the original change trajectory based on abrupt change markers are as follows: Using abrupt change markers as the reference points for time expansion, bidirectional time expansion is performed on time nodes before and after the abrupt change markers, with the addition of time distance parameters to form a time expansion trajectory; The time nodes in the time-distributed trajectory are reordered according to their rate of change, while maintaining the time distance parameter in sync with the original fluctuation amplitude, forming a sequence of change rates arranged according to their rate of change. Based on the rate of change sequence, a daily operating rhythm interval is constructed, continuous time nodes that exceed the daily operating rhythm interval are collected, and impact candidate segments are sorted out in combination with time distance parameters. The time sequence around the candidate impact segment is restored according to the time distance parameter, the original fluctuation amplitude and change rate are preserved, and the impact change trajectory is formed with the abrupt change marker as the time reference point.

5. The wastewater treatment AI management platform and data-driven intelligent control method according to claim 4, characterized in that, The collection of candidate impact segments is combined with the time distance parameter for continuity determination. Time nodes with continuous time distance parameters and a rate of change exceeding the daily operating rhythm range are identified as impact segments. The continuous correlation structure of impact segments in the time unfolding trajectory is maintained around the abrupt change marker.

6. The wastewater treatment AI management platform and data-driven intelligent control method according to claim 4, characterized in that, The steps for determining the initiation point of toxic effects are as follows: Data on the operation of the reaction tank within the corresponding time range were extracted from all time points in the shock change trajectory. The microbial activity characterization parameters were time-aligned with the original fluctuation amplitude and rate of change in the shock change trajectory to form a synchronous control sequence. By tracking the changing trends of microbial activity characterization parameters around the time nodes in the synchronous control sequence, the time segments of continuous decline in microbial activity characterization parameters are identified and associated with the corresponding time nodes in the shock change trajectory to form candidate time segments; Based on the candidate time intervals, the continuous time nodes in the impact change trajectory where the rate of change is within the impact interval and the microbial activity characterization parameters continue to decrease are grouped into time intervals that occur synchronously. By tracing back along the trajectory of the impact change from the starting point of the synchronous time segment to the time point when the rate of change first entered the impact segment range, the starting position of the toxic effect was determined and marked.

7. The wastewater treatment AI management platform and data-driven intelligent control method according to claim 6, characterized in that, By considering the distribution of change rates in the impact trajectory and the time nodes within the synchronous time interval, the duration of continuous decline in microbial activity characterization parameters is continuously analyzed. The earliest time node where the change rate remains within the impact interval and the microbial activity characterization parameters continue to decline is determined as the starting point of the toxicity effect.

8. The wastewater treatment AI management platform and data-driven intelligent control method according to claim 6, characterized in that, The steps for a time-backtracking analysis of the impact trajectory around the initial location of the toxic effects are as follows: Starting from the initial location of the toxicity effect, we traced back along the impact trajectory node by node to determine the time point when the rate of change entered the impact zone and define the time interval of the water quality surge stage. Existing dosing rhythm records corresponding to the time interval of the water quality surge stage were extracted, and the time nodes of the dosing action were aligned with the time nodes of the impact change trajectory to form a time mapping relationship; Based on the time interval between the start time of the water quality surge and the start location of the toxicity effect, the time nodes of the dosing actions in the existing dosing rhythm records are compared to form a time difference sequence between the water quality surge and the existing dosing rhythm. By combining the time difference sequence with the time node corresponding to the starting position of toxic effects, the starting time position of the intervention can be determined.

9. The wastewater treatment AI management platform and data-driven intelligent control method according to claim 8, characterized in that, The time difference between the starting point of the sudden rise in water quality and the time point of the most recent dosing action in the existing dosing rhythm was extracted, and the location was compared with the time point corresponding to the starting point of the toxicity effect. The time point not covered by the existing dosing rhythm was selected as the intervention starting point.

10. The wastewater treatment AI management platform and data-driven intelligent control method according to claim 8, characterized in that, The steps for adjusting the frequency and dosage of subsequent medication based on the intervention starting point are as follows: The impact duration phase is divided by using the intervention start point as the control benchmark time node, and the existing drug administration rhythm records are rearranged around the intervention start point to form a new drug administration rhythm sequence. A progressive adjustment rule is set up around the dosing action during the continuous impact phase. The single dosing amount is reduced in chronological order and the dosing interval is redistributed so that the single dosing amount and the dosing interval correspond to each other. Based on the progressive adjustment rules, the inhibition phase and the recovery phase are divided. During the inhibition phase, the dosing frequency is kept concentrated, and during the recovery phase, the dosing interval is widened, forming an alternating structure of inhibition and recovery phases. By arranging the frequency and dosage of drug administration during the sustained impact phase around the alternating structure of the inhibition and recovery phases, a rhythmic rearrangement trajectory with the intervention starting point as the control benchmark is formed, thereby controlling the range of toxic effects.

Images