Chemical multi-element wastewater treatment method and system

Abstract

The application discloses a chemical multi-element wastewater treatment method and system, and relates to the technical field of wastewater treatment. The method comprises the following steps: uniformly accessing operation data of multi-source wastewater entering a treatment system at a wastewater collection node, synchronously acquiring water inflow and online toxicity detection results, time-aligning the data, and forming water inflow information reflecting the current wastewater state; organizing the water inflow information at continuous time points according to a fixed time window, obtaining a state sequence for describing a toxicity change trend, and generating a toxicity time sequence fingerprint; generating a toxicity load prediction value based on the toxicity time sequence fingerprint, taking the current toxicity time sequence fingerprint and the prediction result as a common judgment basis, and identifying a risk mode of a current operation state. The application realizes the intelligent operation control transformation from passive response to prediction driving by early identification of the toxicity evolution trend of the chemical multi-element wastewater and active scheduling of a treatment path, and improves the stability and adaptability of the treatment system to complex toxicity impact.

Description

Technical Field

[0001] This invention relates to the field of wastewater treatment technology, and in particular to a method and system for treating chemical multi-element wastewater. Background Technology

[0002] With the rapid development of industries such as fine chemicals, pharmaceuticals, and new materials, the wastewater generated during chemical production has become increasingly complex. Wastewater sources are diverse, compositions vary widely, and pollutant types often overlap. Especially when multiple units operate simultaneously and production conditions frequently switch, the quality of wastewater entering the treatment system often fluctuates significantly and exhibits a degree of uncertainty.

[0003] For this type of complex chemical wastewater, the industry currently widely adopts physicochemical methods, biological methods, or combinations thereof for treatment, and relies on online monitoring and automatic control to regulate the operating status. Some treatment systems are already capable of continuously monitoring the influent flow rate and conventional water quality indicators, and adjusting the aeration intensity, reflux ratio, or chemical dosage according to preset thresholds or empirical rules to maintain the basic stability of the treatment system.

[0004] However, with the increasing complexity of toxic substances in wastewater, relying solely on monitoring traditional indicators and instantaneous data is no longer sufficient to comprehensively assess the actual impact of wastewater on treatment systems. Existing control methods are gradually proving inadequate when faced with the synergistic effects of multiple pollutants. Specifically, current methods for treating multi-component chemical wastewater have two main shortcomings: First, most existing technologies only focus on water quality or toxicity indicators at a specific moment, lacking a systematic analysis of the evolution of toxicity over time. This makes it impossible for the system to identify the cumulative trend of toxicity or potential sudden risks over a longer period. When wastewater toxicity appears in a "slow rise" or "insidious enhancement" manner, it is often difficult to provide timely warnings, and the treatment system usually only makes passive adjustments after the toxic shock has already occurred, which is too late.

[0005] Second, existing treatment systems often employ fixed process paths or simple adjustment rules in their operation control. Even when anomalies are detected, a "one-size-fits-all" approach to parameter adjustments is frequently used, lacking the ability to select targeted treatment paths based on toxicity risk characteristics. The system struggles to dynamically match the most suitable operating scheme for different toxicity risk states, thus weakening its buffering capacity for highly toxic wastewater and impacting overall treatment efficiency. Summary of the Invention

[0006] In view of the aforementioned existing problems, the present invention is proposed.

[0007] Therefore, this invention provides a method and system for treating chemical multi-element wastewater, which solves the prominent problems of existing chemical multi-element wastewater treatment technologies in terms of lagging toxicity evolution identification and lack of targeted treatment path scheduling.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, the present invention provides a method for treating chemical multi-element wastewater, comprising, At the wastewater collection node, the multi-source wastewater operation data entering the treatment system are uniformly accessed, the influent flow rate and online toxicity detection results are obtained synchronously, and the data are time-aligned to form influent information reflecting the current wastewater status. The water ingress information at continuous moments is organized into fixed time windows to obtain a state sequence that describes the trend of toxicity changes and to generate a toxicity time-series fingerprint. Based on the toxicity time-series fingerprint, a toxicity load prediction value is generated. The current toxicity time-series fingerprint and the prediction result are used together as the judgment basis to identify the risk pattern of the current operating status and determine whether there is a high-risk toxicity state. After completing risk identification, based on the high-risk toxicity status determined, an appropriate operating path is selected from the preset processing paths, and the selected operating path is converted into a scheduling control command for execution.

[0009] As a preferred embodiment of the chemical multi-source wastewater treatment method of the present invention, the step of uniformly accessing the multi-source wastewater operation data entering the treatment system at the wastewater collection node, synchronously acquiring the influent flow rate and online toxicity detection results, and aligning the data by time to form influent information reflecting the current wastewater status includes: establishing a unified data access interface at the wastewater collection node, centrally collecting wastewater operation data from different chemical production units, synchronously reading the influent flow rate value and online toxicity detection value at the current moment with a fixed sampling period, performing timestamp correction on the acquired influent flow rate data and toxicity detection data, uniformly mapping the two to the current analysis time t, and simultaneously verifying the validity of the collected data, removing missing data points, and retaining only the valid flow rate value and toxicity detection value within the current sampling period; After the data is aligned and verified by time, the influent flow rate Q(t) at the current time t is combined with the corresponding online toxicity detection value TI(t) to construct the influent state vector X(t) that reflects the overall operating status of the wastewater and write it into the real-time buffer in chronological order.

[0010] As a preferred embodiment of the chemical multi-element wastewater treatment method of the present invention, the step of organizing the influent information at continuous time moments into a fixed time window to obtain a state sequence for describing the trend of toxicity change and generating a toxicity time fingerprint includes: establishing a sliding time window container with a fixed capacity of W in memory, storing the influent state vectors of the most recent W time moments, and forming a toxicity state sequence. Each vector is read sequentially from the toxicity state sequence to obtain two aligned time series, including the flow rate sequence and the toxicity sequence; The absolute value of the toxicity difference between adjacent sampling points is calculated in chronological order, and the average of all adjacent differences within the window is calculated to obtain the rate of toxicity change. The maximum and minimum values ​​of the toxicity index are extracted within the same time window, and their difference is calculated to obtain the toxicity impact amplitude. The toxicity index at each moment within the window is weighted according to the corresponding flow rate and normalized with the total flow rate within the window to obtain the equivalent toxicity load. The three features are combined in a fixed order to form the toxicity time fingerprint F(t).

[0011] As a preferred embodiment of the chemical multi-element wastewater treatment method of the present invention, the step of generating toxicity load prediction values ​​based on toxicity time-series fingerprints includes: Using the current time t as a reference, slide forward to construct a toxic fingerprint time series of length H. ; The fingerprint vector at each time point is standardized using Z-score. Load the BiGRU that has been trained offline. θ Model, input of the currently constructed The sequence outputs real-time predicted toxicity load values.

[0012] As a preferred embodiment of the chemical multi-element wastewater treatment method of the present invention, the step of using the current toxicity time-series fingerprint and the prediction result together as the judgment basis to perform risk pattern identification on the current operating state and determine whether a high-risk toxicity state exists includes: The toxicity pattern template library is retrieved from the historical operational database. Each template corresponds to a verified high-risk toxicity evolution pattern. The current toxicity time series fingerprint F(t) is calculated and compared with each template F in the template library. k The normalized Euclidean distance is used to select the template with the smallest distance as the most similar reference object for the current toxic form; The obtained reference object is jointly judged with the toxicity load prediction value. Two deterministic criteria are set, including morphological similarity criterion and predicted load increase criterion. Only when both criteria are met at the same time is the current operating state determined to be a predictive high-risk toxicity state. When it is determined to be a predictive high-risk state, a risk label is output. The risk label also retains the index of the most similar historical template. If either criterion is not met, the normal operating path is entered.

[0013] As a preferred embodiment of the chemical multi-element wastewater treatment method of the present invention, the step of determining a suitable operating path from the preset treatment paths based on the high-risk toxicity state after risk identification refers to using a risk label as a trigger signal for path decision-making. First, the set of candidate operating paths corresponding to the high-risk toxicity state is retrieved from the preset treatment path configuration table. The current influent flow data and the predicted future toxicity load change trend are read. The adaptability of each candidate operating path under the current operating conditions is calculated and compared one by one. The operating path with the highest comprehensive adaptability is selected as the only execution path in the current operating cycle and marked as active, while other paths are frozen.

[0014] As a preferred embodiment of the chemical multi-element wastewater treatment method of the present invention, wherein: the step of converting the determined operating path into a scheduling control instruction for execution refers to generating a corresponding scheduling control instruction based on the determined operating path, uniformly encoding the parameters of load allocation ratio, execution sequence and priority involved in the operating path, and sending the scheduling control instruction to the execution layer.

[0015] Secondly, the present invention provides a chemical multi-element wastewater treatment system, comprising, The water status construction module is used to uniformly access multi-source wastewater operation data at the wastewater collection node, synchronously collect influent flow and online toxicity detection results and complete time alignment, and construct an influent status vector that reflects the current wastewater status. The toxicity time-series fingerprint generation module is used to organize the influent state vector at continuous time points according to a fixed time window, extract the toxicity change rate, toxicity impact amplitude and flow-weighted toxicity load, and generate a toxicity time-series fingerprint to characterize the toxicity change trend. The toxicity load prediction module is used to predict the equivalent toxicity load in the future time period based on the current and historical toxicity time series fingerprint sequences, and output the prediction results for judging the trend of toxicity impact in advance. The risk pattern recognition module is used to use the current toxicity time-series fingerprint and toxicity load prediction results as a joint judgment basis, match them with historical toxicity pattern templates, identify whether the current operating state belongs to the predictive high-risk toxicity state, and output a risk label. The path decision and scheduling execution module is used to determine the appropriate running path from the preset processing path configuration table based on the risk label after the state is determined to be high-risk toxicity, and to convert the determined running path into scheduling control instructions and send them to the execution layer to complete the operation adjustment.

[0016] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the chemical multi-element wastewater treatment method as described in the first aspect of the present invention.

[0017] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the chemical multi-element wastewater treatment method as described in the first aspect of the present invention.

[0018] The beneficial effects of this invention are as follows: By introducing a time-window-based toxicity time-series fingerprint construction and toxicity load prediction mechanism, a systematic characterization and forward-looking judgment of the toxicity change trend of wastewater is realized. This enables the treatment system to complete early warning and response before the high-risk toxicity state is fully manifested, effectively avoiding the passive adjustment problem caused by traditional methods that rely solely on instantaneous indicators. This invention uses the risk identification result as the direct basis for operation path scheduling, realizing accurate matching between the treatment path and the toxicity risk state. This allows the treatment system to select a more suitable operation path for different toxicity risk types, thereby significantly improving the buffering capacity and operational stability of chemical multi-element wastewater treatment to complex toxicity impacts, reducing the risk of system instability and improving overall treatment efficiency. Attached Figure Description

[0019] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 This is a flowchart of the chemical multi-element wastewater treatment method in Example 1.

[0021] Figure 2 This is a structural diagram of the chemical multi-element wastewater treatment system in Example 1. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Example 1, referring to Figure 1 and Figure 2 This is the first embodiment of the present invention, which provides a method for treating chemical multi-element wastewater, including the following steps: S1. At the wastewater collection node, the multi-source wastewater operation data entering the treatment system is uniformly accessed, the influent flow rate and online toxicity detection results are obtained synchronously, and the data is time-aligned to form influent information reflecting the current wastewater status. Specifically, a unified data access interface is established at the wastewater collection point to centrally collect wastewater operation data from different chemical production units. This access interface does not distinguish between wastewater source units, but treats all wastewater entering the treatment system as a whole influent object, thereby avoiding the data fragmentation problem caused by unit-based collection. After unified access is completed, the influent flow rate and online toxicity detection value are synchronously read at a fixed sampling period (set to 1 minute). The influent flow rate is obtained by the flow meter installed at the collection node, reflecting the volume of wastewater entering the treatment system per unit time. The online toxicity detection value is output by the toxicity detection unit deployed at the same node, used to characterize the comprehensive toxicity level of multiple pollutants in the current influent. By using a unified sampling triggering mechanism, we ensure that flow data and toxicity detection data are collected simultaneously within the same sampling period, thus guaranteeing the temporal consistency of the two types of data from the source. The toxicity detection unit preferably employs an online integrated toxicity detection technology based on the inhibitory effect of luminescent bacteria, or an online toxicity detection technology based on changes in bioelectrical signals. Its output is standardized to form a toxicity index that reflects the comprehensive toxicity level of multiple pollutants, which is used for subsequent toxicity time series modeling and predictive analysis. After completing the synchronous acquisition, the acquired influent flow data and toxicity detection data are timestamped and mapped to the current analysis time t. At the same time, the validity of the acquired data is verified, missing data points are removed, and only the flow value and toxicity detection value valid within the current sampling period are retained. After the data is aligned and verified by time, the influent flow rate Q(t) at the current time t is combined with the corresponding online toxicity detection value TI(t) to construct the influent state vector X(t) that reflects the overall operating status of the wastewater. This vector is then written into the real-time buffer in chronological order and marked as the input data for the current analysis period.

[0026] At the wastewater collection node, the system uniformly accesses the operational data of multi-source wastewater entering the treatment system and simultaneously acquires influent flow rate and online toxicity detection results. This allows the system to uniformly characterize multi-source chemical wastewater from the perspective of the overall influent, avoiding the data asynchrony and information fragmentation problems caused by decentralized collection by production unit. By implementing unified sampling triggering and time alignment processing for flow rate data and toxicity detection data, the system ensures that the two types of key operating parameters have a strict correspondence at the same time, so that the coupling relationship between changes in influent load and changes in toxicity can be accurately reflected. The influent status information constructed on this basis can stably and accurately depict the current comprehensive operating status of wastewater, providing highly consistent and reliable data input for subsequent toxicity evolution analysis, risk prediction, and treatment path scheduling. This significantly improves the perception capability and operational decision reliability of the chemical multi-source wastewater treatment system in response to complex influent conditions.

[0027] S2. Organize the water inflow information at continuous time points into fixed time windows to obtain a state sequence that describes the trend of toxicity changes and generate a toxicity time-series fingerprint. Specifically, a sliding time window container with a fixed capacity of W=60 is created in memory to store the influent state vectors of the most recent W moments. The reason for choosing W=60 for the window capacity is that the start-up, shutdown, switching and discharge shocks of chemical plants often manifest as water quality change cycles of "tens of minutes to one hour". This length can cover a typical shock evolution process, which is convenient for building a stable toxicity change trend view later. Once the system enters a stable sampling phase, the control program reads the 60 most recent water ingress state vectors from the buffer in chronological order, fills them sequentially into the sliding time window container, and arranges them in order from newest to oldest to form a toxicity state sequence. In subsequent real-time operation, the system performs a deterministic window sliding update every time a new water ingress state vector is generated: first insert the latest X(t) into the head of the sequence, and then remove the oldest X(tW) from the tail of the sequence, so as to keep the window length constant and always reflect the toxicity evolution process of the "most recent hour". Each vector is read sequentially from the toxicity state sequence to obtain two aligned time series, including a flow sequence and a toxicity sequence. The toxicity sequence is used to describe the "toxicity change trajectory", and the flow sequence is used to convert the toxicity trajectory into an "equivalent toxicity load". The absolute value of the toxicity difference between adjacent sampling points is calculated in chronological order, and the average of all adjacent differences within the window is taken to obtain the toxicity change rate v(t):

[0028] In the formula, TI(t−i) is the toxicity index at the i-th historical moment in the sequence, and W−1 is the number of adjacent difference terms; Within the same time window, the maximum and minimum values ​​of the toxicity index are extracted, and their difference is calculated to obtain the toxicity impact magnitude a(t):

[0029] The toxicity index at each moment within the window is weighted by the corresponding flow rate and normalized using the total flow rate within the window to obtain the equivalent toxicity load c(t):

[0030] In the formula, Q(t−i) is the influent flow rate into the wastewater treatment system during the i-th sampling period before the current analysis time t; The three features are combined in a fixed order to form the toxicity time fingerprint F(t): .

[0031] By introducing a sliding time window to uniformly organize influent information at multiple moments and constructing a toxicity time-series fingerprint, a systematic characterization of the toxicity evolution process of chemical multi-element wastewater is achieved. This avoids the risk identification lag problem caused by traditional methods that rely solely on instantaneous indicators. By simultaneously considering the rate of toxicity change, fluctuation amplitude, and equivalent toxicity load after flow weighting, the toxicity assessment results are closer to actual treatment conditions, effectively improving the ability to identify hidden and cumulative toxicity risks. Based on this, the formed toxicity time-series fingerprint provides stable and discriminative feature inputs for subsequent toxicity load prediction and risk pattern recognition, enabling the treatment system to take control measures in advance before the toxicity impact fully manifests. This significantly enhances the operational stability, risk control capabilities, and overall treatment efficiency of the chemical multi-element wastewater treatment system under complex conditions.

[0032] S3. Generate a toxicity load prediction value based on the toxicity time-series fingerprint, and use the current toxicity time-series fingerprint and the prediction result together as the judgment basis to identify the risk pattern of the current operating status and determine whether there is a high-risk toxicity state. Specifically, the toxicity load prediction values ​​generated based on toxicity time-series fingerprints include: Using the current time t as a reference, slide forward to construct a toxic fingerprint time series of length H=10. ; Z-score normalization is performed on the fingerprint vector at each time point to ensure that the numerical scale of features in different dimensions is consistent, so as to accelerate model convergence. Define the target output Δ = 10 minutes in the future as the predicted toxic load value at that time point. :

[0033] In the formula, It is a bidirectional GRU network trained based on parameter θ, and its output layer is a real number used for regression prediction of future toxicity load; The model structure is set as follows: Input layer: 10 time steps, with each step having an input dimension of d; Bidirectional GRU layer: Single layer, hidden state dimension set to 64, tanh activation used; Fully connected layer: The output is a single scalar, representing ; The loss function is the mean squared error (MSE). The model training process involves extracting toxicity time-series fingerprint data from a historical operational database of chemical multi-element wastewater within continuous and stable operating cycles as training samples. Each sample consists of a toxicity time-series fingerprint sequence spanning H=10 consecutive time steps as input, along with the measured equivalent toxicity load value corresponding to the future Δ=10 minutes later as a supervision label, forming an input-output sample pair. After sample construction, all toxicity time-series fingerprint features are standardized to eliminate the impact of differences in feature dimensions and numerical ranges on model training. Subsequently, a bidirectional gated recurrent unit network is used as the main structure of the prediction model, simultaneously modeling toxicity changes through forward and backward time dependencies. The dynamic trend of the model is analyzed, and the mean squared error is used as the loss function to measure the deviation between the predicted value and the actual toxicity load. The backpropagation algorithm based on gradient descent is used to iteratively update the model parameters. During the training process, the samples are divided into training set and validation set according to a preset ratio. The error change of the validation set is monitored synchronously in each iteration. When the validation error no longer decreases after several consecutive iterations, an early stopping mechanism is automatically triggered to prevent the model from overfitting and to ensure generalization ability. Finally, the converged model parameters are solidified and stored as a prediction model for rapid inference of the latest toxicity time series fingerprint sequence during subsequent real-time operation, thereby realizing the prospective prediction of toxicity load. Load the offline-trained BiGRU into the cloud inference module. θ Model, input of the currently constructed Sequence, real-time output of predicted values .

[0034] By introducing a toxicity load prediction mechanism based on toxicity time-series fingerprints, a forward-looking characterization of the toxicity change trend of chemical multi-element wastewater is achieved. This enables the treatment system to identify potential risks in advance before the actual occurrence of toxicity shocks, avoiding the passive response problem caused by relying solely on instantaneous indicators in existing technologies. At the same time, by modeling the toxicity evolution process through a bidirectional GRU model, the dynamic correlation of toxicity characteristics in the time dimension is fully utilized, improving the prediction accuracy and stability under complex, nonlinear, and multi-pollutant superposition conditions. The prediction results can be directly used to guide the scheduling of treatment paths and the adjustment of operating parameters, thereby enhancing the buffering capacity and operational robustness of the chemical multi-element wastewater treatment system to high-risk toxic conditions, reducing the risk of system instability, and improving overall treatment efficiency.

[0035] Furthermore, the current toxicity time-series fingerprint and the prediction results are used together as the basis for judgment to identify risk patterns in the current operating state and determine whether a high-risk toxicity state exists, including: Retrieve the toxicity pattern template library L={F from the historical operation database. k |k=1,2,…,K}, each template F k =[v k ,a k ,c k This corresponds to a verified high-risk toxicity evolution model. To ensure that the current fingerprint and the template fingerprint are still comparable under different dimensions, the standard deviation parameter σ of each fingerprint dimension is retrieved. j (Obtained from template library statistics), used for subsequent normalized distance calculation, where j=1,2,3 correspond to the three dimensions v, a, and c respectively; Calculate the current toxicity time-series fingerprint F(t) and each template F in the template library one by one. k The normalized Euclidean distance is used to perform a minimum search on the distance sequence, and the template with the smallest distance is selected as the most similar reference object of the current toxicity pattern (the minimum morphological distance between the current toxicity time series fingerprint and historical high-risk patterns). ); The obtained reference objects and the predicted toxicity load values ​​are jointly judged, and two deterministic criteria are set, including: Morphological similarity criteria:

[0036] in, The morphological similarity threshold is determined by balancing historical false positive and false negative rates. Criteria for predicting load increase:

[0037] Where c(t) is the current equivalent toxicity load, and γ is the minimum significant increase, which is used to distinguish between "normal fluctuations" and "imminent toxic shock". The minimum significant increase is calculated by statistically analyzing the equivalent toxicity load change data of the system during the stable operation phase, based on the mean and standard deviation of its historical changes, using the fluctuation threshold setting method commonly used in statistical process control, and is used to distinguish between normal fluctuations and abnormal toxicity increases. The current running state is determined to be a predictive high-risk toxicity state only when both criteria are met. When it is determined to be a predictive high-risk state, a risk label is output, and the index of the most similar historical template is retained in the risk label. If either criterion is not met, the normal running path is entered.

[0038] By combining toxicity time-series fingerprints with toxicity load prediction results for risk pattern identification, this invention enables forward-looking judgment of the toxicity evolution process of chemical multi-element wastewater. It can identify potential toxic shocks before high-risk toxicity states are fully formed, effectively overcoming the adjustment lag problem caused by existing technologies that rely solely on instantaneous indicators or post-event responses. This invention uses historical high-risk toxicity patterns as a reference and makes judgments based on both morphological similarity and predicted trends, significantly reducing the probability of misjudgment caused by random fluctuations and improving the accuracy and stability of risk identification. This provides a reliable basis for the precise scheduling of subsequent treatment paths, and overall improves the operational safety, shock resistance, and long-term stable operation level of chemical multi-element wastewater treatment systems under complex operating conditions.

[0039] S4. After completing risk identification, based on the high-risk toxicity status obtained from the determination, determine the appropriate operation path from the preset processing path, and convert the determined operation path into a scheduling control command for execution. Specifically, after risk identification is completed, based on the high-risk toxicity status determined, a suitable operating path is determined from the preset treatment paths. The risk label is used as the trigger signal for path decision. First, the set of candidate operating paths corresponding to the high-risk toxicity status is retrieved from the preset treatment path configuration table. The current influent flow data and the predicted future toxicity load change trend are read. The adaptability of each candidate operating path under the current operating conditions is calculated and compared one by one. Specifically, it includes judging whether the load carrying capacity of each path under the current flow conditions meets the requirements, and whether it has the ability to buffer and reduce the toxicity impact in advance during the predicted toxicity load rise stage. After the comparison is completed, the operating path with the highest comprehensive adaptability is selected as the only execution path in the current operating cycle and marked as active. At the same time, other paths are frozen to avoid frequent switching during operation. The preset processing path configuration table is stored in the control system in the form of a software data structure. Its overall architecture is constructed in a hierarchical manner of "path definition - applicable conditions - control parameters - switching constraints", which is used to realize the rapid and deterministic selection and scheduling of processing paths during operation. The configuration table first includes path identifiers and path structure descriptions, which are used to uniquely identify each processing path defined in the system and record the corresponding processing unit combination order and operating mode. Secondly, a risk adaptation field is set to clarify the toxicity risk type or toxicity action mode corresponding to each path, so that after risk identification, the available path can be directly indexed through the risk label. The configuration table further includes an applicable operating condition field to limit the applicability of each path within different influent flow ranges and toxicity load intervals, thereby avoiding the misselection of operating paths when the system's carrying capacity is exceeded. The configuration table pre-stores control parameter templates for each path, which are used to directly generate scheduling control instructions when the path is activated. The parameter templates include at least key control information such as load allocation ratio, execution order, and effective time.

[0040] By establishing a mapping relationship between risk labels and preset treatment paths, the system achieves precise selection and stable execution of operating paths. This enables the treatment system to dynamically match the most suitable operating path for different toxicity risk states, thereby effectively improving the buffering capacity and operational stability of the chemical multi-element wastewater treatment process against complex toxicity impacts. It also reduces system fluctuations and energy consumption increases caused by frequent path switching, thus improving the overall safety and operational efficiency of the treatment system.

[0041] Furthermore, the determined operating path is transformed into a scheduling control instruction for execution. Based on the determined operating path, a corresponding scheduling control instruction is generated, and the parameters of load allocation ratio, execution sequence and priority involved in the operating path are uniformly encoded. The scheduling control instruction is then sent to the execution layer, enabling the processing system to proactively adjust before the toxicity risk is fully manifested.

[0042] This embodiment also provides a chemical multi-element wastewater treatment system, including: The water status construction module is used to uniformly access multi-source wastewater operation data at the wastewater collection node, synchronously collect influent flow and online toxicity detection results and complete time alignment, and construct an influent status vector that reflects the current wastewater status. The toxicity time-series fingerprint generation module is used to organize the influent state vector at continuous time points according to a fixed time window, extract the toxicity change rate, toxicity impact amplitude and flow-weighted toxicity load, and generate a toxicity time-series fingerprint to characterize the toxicity change trend. The toxicity load prediction module is used to predict the equivalent toxicity load in the future time period based on the current and historical toxicity time series fingerprint sequences, and output the prediction results for judging the trend of toxicity impact in advance. The risk pattern recognition module is used to use the current toxicity time-series fingerprint and toxicity load prediction results as a joint judgment basis, match them with historical toxicity pattern templates, identify whether the current operating state belongs to the predictive high-risk toxicity state, and output a risk label. The path decision and scheduling execution module is used to determine the appropriate running path from the preset processing path configuration table based on the risk label after the state is determined to be high-risk toxicity, and to convert the determined running path into scheduling control instructions and send them to the execution layer to complete the operation adjustment.

[0043] This embodiment also provides a computer device applicable to the treatment method of chemical multi-element wastewater, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the chemical multi-element wastewater treatment method proposed in the above embodiment.

[0044] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0045] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the chemical multi-element wastewater treatment method proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0046] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for treating chemical multi-element wastewater, characterized in that: include, At the wastewater collection node, the multi-source wastewater operation data entering the treatment system are uniformly accessed, the influent flow rate and online toxicity detection results are obtained synchronously, and the data are time-aligned to form influent information reflecting the current wastewater status. The water ingress information at continuous moments is organized into fixed time windows to obtain a state sequence that describes the trend of toxicity changes and to generate a toxicity time-series fingerprint. Based on the toxicity time-series fingerprint, a toxicity load prediction value is generated. The current toxicity time-series fingerprint and the prediction result are used together as the judgment basis to identify the risk pattern of the current operating status and determine whether there is a high-risk toxicity state. After completing risk identification, based on the high-risk toxicity status determined, an appropriate operating path is selected from the preset processing paths, and the selected operating path is converted into a scheduling control command for execution.

2. The method for treating chemical multi-element wastewater as described in claim 1, characterized in that: The process of unifying the multi-source wastewater operation data entering the treatment system at the wastewater collection node, synchronously acquiring influent flow rate and online toxicity detection results, and aligning the data with time to form influent information reflecting the current wastewater status includes: establishing a unified data access interface at the wastewater collection node, centrally collecting wastewater operation data from different chemical production units, synchronously reading the influent flow rate value and online toxicity detection value at the current moment with a fixed sampling period, performing timestamp correction on the acquired influent flow rate data and toxicity detection data, mapping both to the current analysis time t, and simultaneously verifying the validity of the collected data, removing missing data points, and retaining only the flow rate value and toxicity detection value valid within the current sampling period; After the data is aligned and verified by time, the influent flow rate Q(t) at the current time t is combined with the corresponding online toxicity detection value TI(t) to construct the influent state vector X(t) that reflects the overall operating status of the wastewater and write it into the real-time buffer in chronological order.

3. The method for treating chemical multi-element wastewater as described in claim 2, characterized in that: The step of organizing the water inflow information at continuous moments into a fixed time window to obtain a state sequence describing the trend of toxicity change and generating a toxicity time fingerprint includes: establishing a sliding time window container with a fixed capacity of W in memory to store the water inflow state vectors of the most recent W moments and forming a toxicity state sequence. Each vector is read sequentially from the toxicity state sequence to obtain two aligned time series, including the flow rate sequence and the toxicity sequence; The absolute value of the toxicity difference between adjacent sampling points is calculated in chronological order, and the average of all adjacent differences within the window is calculated to obtain the rate of toxicity change. The maximum and minimum values ​​of the toxicity index are extracted within the same time window, and their difference is calculated to obtain the toxicity impact amplitude. The toxicity index at each moment within the window is weighted according to the corresponding flow rate and normalized with the total flow rate within the window to obtain the equivalent toxicity load. The three features are combined in a fixed order to form the toxicity time fingerprint F(t).

4. The method for treating chemical multi-element wastewater as described in claim 3, characterized in that: The toxicity load prediction value generated based on the toxicity time-series fingerprint includes: Using the current time t as a reference, slide forward to construct a toxic fingerprint time series of length H. ; The fingerprint vector at each time point is standardized using Z-score. Load the BiGRU that has been trained offline. θ Model, input of the currently constructed The sequence outputs real-time predicted toxicity load values.

5. The method for treating chemical multi-element wastewater as described in claim 4, characterized in that: The step of using the current toxicity time-series fingerprint and the prediction result together as the judgment basis to perform risk pattern identification on the current operating state and determine whether a high-risk toxicity state exists includes: The toxicity pattern template library is retrieved from the historical operational database. Each template corresponds to a verified high-risk toxicity evolution pattern. The current toxicity time series fingerprint F(t) is calculated and compared with each template F in the template library. k The normalized Euclidean distance is used to select the template with the smallest distance as the most similar reference object for the current toxic form; The obtained reference object is jointly judged with the toxicity load prediction value. Two deterministic criteria are set, including morphological similarity criterion and predicted load increase criterion. Only when both criteria are met at the same time is the current operating state determined to be a predictive high-risk toxicity state. When it is determined to be a predictive high-risk state, a risk label is output. The risk label also retains the index of the most similar historical template. If either criterion is not met, the normal operating path is entered.

6. The method for treating chemical multi-element wastewater as described in claim 5, characterized in that: After completing risk identification, based on the high-risk toxicity status determined, a suitable operating path is determined from the preset processing paths. This means that the risk label is used as the trigger signal for path decision-making. First, the set of candidate operating paths corresponding to the high-risk toxicity status is retrieved from the preset processing path configuration table. The current influent flow data and the predicted future toxicity load change trend are read. The adaptability of each candidate operating path under the current operating conditions is calculated and compared one by one. The operating path with the highest comprehensive adaptability is selected as the only execution path in the current operating cycle and marked as active, while other paths are frozen.

7. The method for treating chemical multi-element wastewater as described in claim 6, characterized in that: The step of converting the determined running path into a scheduling control instruction for execution refers to generating a corresponding scheduling control instruction based on the determined running path, uniformly encoding the parameters of load allocation ratio, execution sequence and priority involved in the running path, and sending the scheduling control instruction to the execution layer.

8. A chemical multi-element wastewater treatment system, based on the chemical multi-element wastewater treatment method according to any one of claims 1 to 7, characterized in that: include, The influent status construction module is used to uniformly access multi-source wastewater operation data at the wastewater collection node, synchronously collect influent flow and online toxicity detection results and complete time alignment, and construct an influent status vector reflecting the current wastewater status. The toxicity time-series fingerprint generation module is used to organize the influent state vector at continuous time points according to a fixed time window, extract the toxicity change rate, toxicity impact amplitude and flow-weighted toxicity load, and generate a toxicity time-series fingerprint to characterize the toxicity change trend. The toxicity load prediction module is used to predict the equivalent toxicity load in the future time period based on the current and historical toxicity time series fingerprint sequences, and output the prediction results for judging the trend of toxicity impact in advance. The risk pattern recognition module is used to use the current toxicity time-series fingerprint and toxicity load prediction results as a joint judgment basis, match them with historical toxicity pattern templates, identify whether the current operating state belongs to the predictive high-risk toxicity state, and output a risk label. The path decision and scheduling execution module is used to determine the appropriate running path from the preset processing path configuration table based on the risk label after the state is determined to be high-risk toxicity, and to convert the determined running path into scheduling control instructions and send them to the execution layer to complete the operation adjustment.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the chemical multi-element wastewater treatment method according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the chemical multi-element wastewater treatment method according to any one of claims 1 to 7.

Images