An intelligent control method and device for sewage treatment with working condition self-adaptation and a medium

Abstract

The present application relates to sewage treatment technical field, disclose a kind of working condition self-adapting sewage treatment intelligent control method, device and medium, the present application is clustered to historical operation data using clustering algorithm, to identify several working conditions;Based on the several working conditions identified, determine the control strategy corresponding to each working condition, wherein the control strategy corresponding to different working conditions has different multi-objective optimization weight combination;Obtain the current real-time operation data of sewage treatment system, determine the current working condition according to current real-time operation data;According to the current working condition determined, match and execute the control strategy corresponding to the working condition, solve the technical problems that control strategy is single in prior art, difficult to adapt to working condition dramatic change, and lack of active identification and response mechanism for extreme working condition.

Description

Technical Field

[0001] This invention relates to the field of wastewater treatment technology, specifically to an adaptive intelligent control method, device, and medium for wastewater treatment. Background Technology

[0002] In recent years, a large number of rural domestic sewage treatment facilities have been built. Existing technologies include solutions that apply the Internet of Things (IoT) and new energy technologies to rural sewage treatment. For example, "A Smart Optimization Method and System for Decentralized Rural Sewage Treatment Stations" (CN119490241B) uses an IoT sensing architecture to achieve dynamic control through water quality prediction and parameter optimization; "A Method for Treating Village Domestic Sewage by Using Non-point Source Pollution Purified Water for Power Generation in Hilly Areas" (CN114409205A) utilizes the terrain difference in hilly areas to generate electricity for sewage treatment, achieving partial energy self-sufficiency.

[0003] However, the aforementioned existing technology has the following drawbacks: The current control strategies are too simplistic and unable to adapt to drastic changes in operating conditions. Wastewater treatment in hilly rural areas faces drastically different operating conditions, including high-water seasons, low-water seasons, periods of accelerated filter clogging, and peak electricity prices. Control objectives should be tailored to these different conditions. Existing technologies use fixed optimization objectives or fixed weights, failing to proactively switch control strategies based on changing conditions. For example, during periods of accelerated filter clogging, backwashing should be prioritized to ensure effluent quality, but fixed-weight optimization may delay backwashing due to an excessive focus on energy conservation, leading to equipment damage. Similarly, during peak electricity prices, energy storage should be prioritized for peak shaving, but fixed-weight optimization may still operate in the conventional mode, resulting in wasted electricity.

[0004] There is a lack of proactive identification and response mechanisms for extreme operating conditions. Although existing technologies incorporate data prediction, they lack operating condition classification models and cannot identify abnormal states such as "accelerated blockage" or "impending heavy rain" in advance, resulting in a lack of foresight in control decisions. CN114409205A only describes basic electrical control and has no adaptive capability for operating conditions; while CN119490241B and others use machine learning for water quality prediction, the prediction results are only used for parameter fine-tuning and do not change the fundamental orientation of the control strategy. Summary of the Invention

[0005] This invention provides an intelligent control method, device, and medium for wastewater treatment that adapts to different operating conditions, thereby solving the technical problems of existing technologies, such as the single control strategy, difficulty in adapting to drastic changes in operating conditions, and lack of active identification and response mechanisms for extreme operating conditions.

[0006] In a first aspect, the present invention provides an intelligent control method for wastewater treatment that adapts to different operating conditions, comprising: acquiring historical operating data of a wastewater treatment system; clustering the historical operating data using a clustering algorithm to identify several operating conditions; determining a control strategy corresponding to each operating condition based on the identified operating conditions, wherein the control strategies corresponding to different operating conditions have different multi-objective optimization weight combinations; acquiring current real-time operating data of the wastewater treatment system and determining the current operating condition based on the current real-time operating data; and matching and executing the control strategy corresponding to the determined current operating condition.

[0007] This invention uses a clustering algorithm to cluster historical operating data to automatically identify operating conditions, establishing an operating condition classification model that can identify abnormal states such as "accelerated congestion" and "coming rainstorms" in advance, solving the problem of existing technologies lacking an active identification and response mechanism for extreme operating conditions. Simultaneously, based on the identified operating conditions, corresponding control strategies are determined, with different multi-objective optimization weight combinations corresponding to different operating conditions. In real-time operation, the corresponding strategies are matched and executed according to the current operating conditions, solving the problem of existing technologies having single control strategies that are difficult to adapt to drastic changes in operating conditions. This achieves adaptive adjustment of the priority of optimization objectives under different operating conditions, improving the responsiveness to changes in operating conditions.

[0008] In one optional implementation, historical operational data includes a first type of feature data related to the material flow state of wastewater treatment, and a second type of feature data related to the energy flow state and / or economic state of the wastewater treatment system. This implementation incorporates both material flow state features and energy flow / economic state features into the basis of cluster analysis, enabling the clustered operating conditions to reflect not only the system's hydraulic load and pollutant treatment status, but also its energy consumption level and economic status. This provides a more comprehensive data foundation for the adaptive adjustment of multi-objective optimization weights under different operating conditions, facilitating the synergistic optimization of energy flow and material flow, and further improving the overall operational efficiency of the wastewater treatment system.

[0009] In one optional implementation, the first type of feature data includes at least water level, flow rate, filter pressure differential, and pressure differential change rate; the second type of feature data includes at least energy storage status and electricity price. This implementation, by introducing the pressure differential change rate, can capture the development trend of filter clogging, enabling early prediction of the clogging acceleration period, thereby timely and proactively executing backwashing, extending filter life, and ensuring effluent water quality. By introducing electricity price and energy storage status, priority can be given to energy storage during peak electricity price periods to reduce electricity purchase costs, and priority can be given to power generation during high-water seasons to improve energy self-sufficiency. These features collectively serve as input to the clustering algorithm, allowing the identified operating conditions to simultaneously reflect the equipment's health and economic status, providing a data foundation for adaptive adjustment of multi-objective optimization weights, achieving coordinated optimization of energy flow and material flow, and improving the system's economy and reliability.

[0010] In one optional implementation, historical operating data also includes rainfall probability. By introducing rainfall probability, this implementation enables operating condition identification not only based on the current operating status but also by incorporating future weather trends. This allows for the prediction of water volume growth trends or the risk of heavy rainfall during the high-water season, thus providing foresight in control strategy formulation and enabling advance adjustment of operating modes to cope with water volume fluctuations. This further enhances the system's adaptability and operational stability under complex climatic conditions.

[0011] In one optional implementation, the clustering algorithm is the DBSCAN algorithm, and the neighborhood radius and minimum neighborhood sample number of the DBSCAN algorithm are optimized and determined by the silhouette coefficient method. In this implementation, the DBSCAN algorithm does not require a preset number of clusters, can automatically discover clusters of arbitrary shapes, and effectively identify noisy data. Compared with partitioned clustering methods that require a preset number of clusters, it is more suitable for wastewater treatment scenarios where the types and quantities of operating conditions are unknown. At the same time, the silhouette coefficient method objectively evaluates the clustering effect to optimize and determine the neighborhood radius and minimum neighborhood sample number, avoiding the subjectivity and blindness of human experience setting, further improving the accuracy and rationality of operating condition classification, and providing a reliable data foundation for the accurate matching of subsequent multi-objective optimization weights.

[0012] In one optional implementation, determining the control strategy corresponding to each operating condition includes: for each operating condition, pre-setting a weight combination of a multi-objective optimization problem to form a mapping relationship between the operating condition and the weight combination, using the weight combination to characterize the control objective orientation under that operating condition, and serving as the parameterized expression of the control strategy corresponding to that operating condition; wherein, the optimization objectives of the multi-objective optimization problem include energy self-sufficiency rate, filter lifespan, and electricity purchase cost. This implementation expresses the control strategy as a multi-objective optimization weight combination under different operating conditions, so that the same set of optimization objectives (energy self-sufficiency rate, filter lifespan, and electricity purchase cost) have different priorities under different operating conditions, thereby realizing the adaptive adjustment of the optimization objective weights according to the operating conditions. For example, during the accelerated filter clogging period, the focus is on equipment maintenance to ensure safe operation; during peak electricity price periods, the focus is on reducing electricity purchase cost; and during the high-water season, the focus is on improving energy self-sufficiency rate. This avoids the problem in traditional schemes where fixed weights cannot take into account the needs of multiple operating conditions, and maximizes the overall benefits of the system.

[0013] In one optional implementation, determining the current operating condition based on real-time operating data includes: using a K-nearest neighbor classifier to calculate the distance between the current real-time operating data and the cluster centers of each operating condition, and determining the current operating condition based on the distance. In this implementation, the K-nearest neighbor classifier, as a distance-based non-parametric classification method, has low computational cost and fast decision-making speed. It does not require a complex online training process and can quickly match the operating condition category to which the current operating state belongs based on offline clustering results. It is suitable for deployment on edge devices with limited computing resources, ensuring the real-time performance and reliability of operating condition identification, and providing an accurate decision basis for timely switching of subsequent control strategies.

[0014] In one optional implementation, the control strategy corresponding to the determined current operating condition is matched and executed, including: for the current operating condition, calling the corresponding weight combination from the mapping relationship; solving a multi-objective optimization problem based on the current real-time operating data and the called weight combination to obtain the optimal equipment control parameters under the current operating condition; generating a corresponding control command sequence based on the optimal equipment control parameters, and using the control command sequence as the control strategy for the current operating condition. This implementation, by calling the weight combination corresponding to the current operating condition and combining it with the current real-time operating data to solve the multi-objective optimization problem online, can generate the optimal control command sequence matching the current operating condition in real time based on the correct identification of the operating condition. This transforms the operating condition identification result into specific equipment control actions, achieving complete closed-loop control from operating condition identification to strategy execution. This avoids the problem of strategy execution being disconnected from operating condition identification in traditional solutions, achieving precise matching and dynamic adaptive adjustment of the control strategy, and further improving the operational efficiency and intelligence level of the wastewater treatment system under complex operating conditions.

[0015] Secondly, the present invention provides an intelligent control device for wastewater treatment with adaptive operating conditions, comprising: a historical data acquisition module for acquiring historical operating data of the wastewater treatment system; an operating condition classification module for clustering the historical operating data using a clustering algorithm to identify several operating conditions; a strategy preset module for determining the control strategy corresponding to each operating condition based on the identified operating conditions, wherein the control strategies corresponding to different operating conditions have different multi-objective optimization weight combinations; a real-time judgment module for acquiring the current real-time operating data of the wastewater treatment system and determining the current operating condition based on the current real-time operating data; and a strategy execution module for matching and executing the control strategy corresponding to the determined current operating condition.

[0016] Thirdly, the present invention provides a computer-readable storage medium storing computer instructions for causing a computer to execute the adaptive intelligent control method for wastewater treatment described in the first aspect or any corresponding embodiment. Attached Figure Description

[0017] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is a schematic diagram of the first process of the intelligent control method for wastewater treatment with adaptive operating conditions according to an embodiment of the present invention. Figure 2 This is a schematic diagram of a second process of an intelligent control method for wastewater treatment that adapts to operating conditions, according to an embodiment of the present invention. Figure 3 This is a structural block diagram of an intelligent control device for wastewater treatment that adapts to working conditions according to an embodiment of the present invention. Figure 4 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means.

[0021] The terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0022] Existing technologies suffer from the following shortcomings: First, they employ a single control strategy, using fixed optimization targets or weights, and cannot proactively switch strategies based on changes in operating conditions such as high-water seasons, low-water seasons, accelerated filter clogging periods, and peak electricity prices. Second, they lack operating condition classification models, making it impossible to identify abnormal states such as "accelerated clogging" or "impending heavy rain" in advance, resulting in a lack of foresight in control decisions. Based on these shortcomings, this invention provides an adaptive intelligent control method, device, and medium for wastewater treatment, addressing the technical problems of existing technologies, such as a single control strategy, difficulty in adapting to drastic changes in operating conditions, and a lack of proactive identification and response mechanisms for extreme operating conditions.

[0023] According to an embodiment of the present invention, an embodiment of an intelligent control method for wastewater treatment with adaptive operating conditions is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0024] This embodiment provides an intelligent control method for wastewater treatment that adapts to different operating conditions. Figure 1 This is a flowchart of an intelligent control method for wastewater treatment with adaptive operating conditions according to an embodiment of the present invention, such as... Figure 1 As shown, the process includes the following steps: Step S101: Obtain historical operating data of the wastewater treatment system.

[0025] Historical operational data refers to the time-series data of various operating parameters collected and recorded by sensors and monitoring instruments during the past operation of the wastewater treatment system. This data reflects the operating status and changing patterns of the system under different time periods and external conditions, and serves as the data foundation for subsequent cluster analysis.

[0026] This step involves acquiring historical operational data covering a sufficiently long time span, enabling the clustering algorithm to extract the distribution characteristics and pattern differences of system operational data under different operating conditions, thus providing data support for the automatic classification of operating condition categories. The acquired historical operational data can be structured data that has undergone preprocessing (such as cleaning, normalization, and missing value imputation) to facilitate subsequent clustering algorithm processing.

[0027] Step S102: Clustering algorithm is used to cluster historical operating data to identify several operating conditions.

[0028] This step employs a clustering algorithm to cluster historical operational data. The algorithm automatically analyzes the similarity between operational parameter vectors from different time periods within the historical data, grouping periods with similar characteristics into the same category, thus dividing the historical data into several clusters. Each cluster represents a set of data with similar operational patterns, corresponding to a specific operating condition, such as high load during the wet season, low load during the dry season, accelerated filter clogging, and priority given to energy storage during peak electricity prices.

[0029] This step automatically discovers and classifies operating conditions from historical data using clustering algorithms. It eliminates the need for manual pre-definition or labeling of sample categories, avoiding the subjectivity and limitations of experience-based classification. This allows operating condition identification to adapt to the actual operating characteristics of different wastewater treatment systems and to continuously optimize as historical data accumulates. Furthermore, after clustering, each cluster generates a cluster center, which characterizes the typical features of that type of operating condition and serves as a reference benchmark for determining the current operating condition in subsequent online real-time identification stages.

[0030] Clustering algorithms can employ partitioning clustering methods (such as K-means clustering, fuzzy C-means clustering, etc.) or density-based clustering methods (such as DBSCAN, etc.).

[0031] Preferably, this step uses density clustering, which does not require a preset number of clusters, can automatically discover clusters of arbitrary shapes and effectively identify noisy data, and is more suitable for wastewater treatment scenarios where the type and quantity of operating conditions are unknown.

[0032] Step S103: Based on the identified operating conditions, determine the control strategy corresponding to each operating condition, wherein the control strategies corresponding to different operating conditions have different multi-objective optimization weight combinations.

[0033] A control strategy refers to a decision-making scheme formulated for the operational objectives and control methods of a wastewater treatment system under specific operating conditions. In this embodiment, the control strategy is represented as a weight combination in a multi-objective optimization problem. Multi-objective optimization refers to simultaneously considering multiple conflicting optimization objectives (such as reducing energy consumption, extending equipment life, and reducing electricity purchase costs) in wastewater treatment control, and balancing the priority relationship between different objectives by adjusting the weight coefficients of each objective. A weight combination is the set of weight coefficients assigned to each optimization objective. Different weight combinations reflect the different degrees of emphasis on each optimization objective under different operating conditions. The larger the weight coefficient of a certain objective in a weight combination, the higher the priority of that objective under that operating condition, and the more the optimization decision will favor the achievement of that objective.

[0034] In this step, for each operating condition identified by clustering, a set of weight combinations matching that operating condition is pre-defined, forming a mapping relationship of "operating condition - weight combination". Different operating conditions correspond to different weight combinations, meaning that the same system has different control objectives under different operating conditions. For example, during peak electricity price periods, the primary objective is to reduce electricity purchase costs, and the weight corresponding to electricity purchase costs is the highest in its weight combination; during the wet season, the primary objective is to improve energy self-sufficiency, and the weight corresponding to energy self-sufficiency is the highest in its weight combination.

[0035] This step quantifies the abstract control strategy into specific multi-objective optimization weight combinations, enabling the system to call different weight combinations to solve the optimization problem under different operating conditions. This achieves adaptive adjustment of the control objective according to the operating conditions, solving the problem in traditional schemes where fixed weights under different operating conditions lead to multiple optimization objectives being unable to be taken into account and some objectives deviating significantly from expectations.

[0036] Step S104: Obtain the current real-time operating data of the sewage treatment system and determine the current operating condition based on the current real-time operating data.

[0037] Current real-time operating data refers to the real-time values ​​of various operating parameters of the wastewater treatment system collected by sensors and monitoring instruments at the current moment or within the current time window. Corresponding to the historical operating data in step S101, the current real-time operating data reflects the operating status of the system at the current moment and serves as the input basis for determining the current operating condition.

[0038] After completing cluster analysis and determining the cluster centers for various operating conditions in the offline phase, the real-time operating data is mapped to the same feature space as the historical operating data. By calculating the similarity or distance between the current real-time operating data and the cluster centers of each operating condition, the operating condition category at the current moment can be identified.

[0039] This step continuously collects current operating data during real-time system operation and automatically determines the current operating condition category based on the classification criteria obtained from clustering in step S102. This provides a basis for subsequent matching and execution of corresponding control strategies, enabling the system to promptly perceive changes in operating conditions and avoid mismatches between control strategies and the operating environment due to misjudgments or delays in operating conditions.

[0040] Step S105: Based on the determined current operating condition, match and execute the control strategy corresponding to that operating condition.

[0041] After determining the current operating condition in step S104, this step searches for and calls the weight combination that matches the current operating condition from the "operating condition-weight combination" mapping relationship established in step S103 based on the determination result. Then, based on the weight combination and the current real-time operating data, it generates specific control commands and drives the equipment to execute.

[0042] This step links the operating condition identification results with control execution, forming a complete closed-loop control link from identification to decision-making to execution. This ensures that the system can execute the control strategy that matches the current operating condition in a timely and accurate manner after identifying it, avoiding a decrease in control effectiveness or equipment damage due to strategy execution lag or mismatch.

[0043] The adaptive intelligent control method for wastewater treatment provided in this embodiment uses a clustering algorithm to cluster historical operating data to automatically identify operating conditions and establish an operating condition classification model. This model can identify abnormal states such as "accelerated blockage" and "impending rainstorm" in advance, solving the problem that existing technologies lack an active identification and response mechanism for extreme operating conditions. At the same time, based on the identified operating conditions, corresponding control strategies are determined. Different operating conditions correspond to different multi-objective optimization weight combinations, and the corresponding strategies are matched and executed according to the current operating conditions during real-time operation. This solves the problem that existing technologies have single control strategies that are difficult to adapt to drastic changes in operating conditions, and realizes adaptive adjustment of the priority of optimization objectives under different operating conditions, thereby improving the responsiveness to changes in operating conditions.

[0044] This embodiment also provides another intelligent control method for wastewater treatment that adapts to different operating conditions. Figure 2 This is a flowchart of an intelligent control method for wastewater treatment with adaptive operating conditions according to an embodiment of the present invention, such as... Figure 2 As shown, the process includes the following steps: Step S201: Obtain historical operating data of the wastewater treatment system.

[0045] In this embodiment, historical operational data includes a first type of characteristic data related to the material flow state of wastewater treatment, and a second type of characteristic data related to the energy flow state and / or economic state of the wastewater treatment system. In an optional implementation, the first type of characteristic data includes at least water level, flow rate, filter pressure differential, and pressure differential change rate; the second type of characteristic data includes at least energy storage status and electricity price. In an optional implementation, historical operational data also includes rainfall probability.

[0046] Material flow state refers to the distribution, flow, and transformation of wastewater, pollutants, and other substances within the various units of a wastewater treatment system. It reflects the system's hydraulic load conditions and pollutant treatment status, including parameters such as water level, flow rate, filter pressure differential, and pressure differential change rate. Energy flow state refers to the generation, storage, and consumption of electrical energy within a wastewater treatment system, including parameters such as energy storage status (i.e., the state of charge of batteries, used to characterize the current energy storage level) and power generation. Economic state refers to factors related to system operating costs, including real-time electricity prices.

[0047] The first category of characteristic data is selected from material flow state-related parameters of the wastewater treatment system, including water level, flow rate, filter pressure differential, and pressure differential change rate. Water level refers to the water level in the storage tank or equalization tank of the wastewater treatment system, reflecting the system's water storage capacity and influent conditions; flow rate refers to the influent flow rate entering the wastewater treatment system, reflecting the system's hydraulic load; filter pressure differential refers to the pressure difference between the influent and effluent sides of the filter, reflecting the degree of filter clogging; and pressure differential change rate refers to the change in filter pressure differential per unit time, reflecting the development trend of filter clogging. This can be used to predict when the filter will enter the clogging acceleration period, thereby identifying the accelerated clogging condition in advance and providing a basis for timely and proactive backwashing.

[0048] The second category of feature data is selected from parameters related to the energy flow status and / or economic status of the wastewater treatment system, including energy storage status and electricity price. Energy storage status refers to the current state of charge of energy storage devices, reflecting the system's current energy storage level. This can be used to determine whether the system has sufficient energy storage to meet electricity demand during peak electricity price periods, thus supporting the identification of priority energy storage operation during peak electricity price periods. Electricity price refers to the real-time time-of-use electricity price information of the power grid, reflecting the current electricity purchase cost. This can be used to determine whether the current period is a peak electricity price period, thus supporting the identification of priority energy storage operation during peak electricity price periods and providing a basis for decision-making to reduce electricity purchase costs.

[0049] Rainfall probability refers to the predicted probability of rainfall within a future period (such as the next 24 hours), reflecting future weather trends. By incorporating rainfall probability, clustering algorithms can not only rely on the current operating status of the system when identifying operating conditions, but also integrate future weather trends. This allows for the early prediction of water volume growth trends during the high-water season or the risk of heavy rain impacts, making the identified operating conditions more forward-looking. This provides a basis for the pre-adjustment of control strategies and avoids the system only responding passively after sudden changes in operating conditions occur.

[0050] The aforementioned feature data serve as input to the clustering algorithm, enabling the clustered operating conditions to simultaneously reflect the system's hydraulic load state, energy state, and economic state. This incorporates equipment maintenance and economic characteristics into the basis for operating condition classification, providing a more comprehensive data foundation for the adaptive adjustment of multi-objective optimization weights under different operating conditions. This facilitates the coordinated optimization of energy flow and material flow, thereby improving the overall operational efficiency of the wastewater treatment system.

[0051] The specific values ​​or levels of the aforementioned "water level," "flow rate," "filter pressure difference," "pressure difference change rate," "energy storage status," "electricity price," and "rainfall probability" can be set by those skilled in the art according to actual operating conditions, and this embodiment does not impose specific limitations. All the above-mentioned characteristic data are illustrative examples; in practical applications, the types of characteristic data can be added, removed, or adjusted according to the specific configuration and operational requirements of the system.

[0052] In one alternative implementation, data acquisition and feature extraction are performed on a local controller, which collects sensor data and extracts features. Compared to uploading raw data directly to the cloud for processing or using distributed smart sensors to extract features locally, this implementation performs data acquisition and feature extraction on a local controller, eliminating reliance on network transmission, resulting in low latency, controllable costs, and no need for additional distributed computing devices. It balances real-time performance, reliability, and economy, making it particularly suitable for application scenarios in rural areas with limited network conditions and weak equipment maintenance capabilities.

[0053] Step S202: Clustering algorithm is used to cluster historical operating data to identify several operating conditions.

[0054] In one alternative implementation, the clustering algorithm is the DBSCAN algorithm, and the neighborhood radius and minimum number of neighborhood samples of the DBSCAN algorithm are determined by optimization using the silhouette coefficient method.

[0055] Specifically, an unsupervised clustering algorithm is used to cluster the historical running data obtained in step S201. In this embodiment, the clustering algorithm is the DBSCAN algorithm. The DBSCAN algorithm does not require a preset number of clusters, can automatically discover clusters of arbitrary shapes based on the density characteristics of the data distribution, and effectively identify noisy data that does not belong to any cluster.

[0056] The clustering performance of the DBSCAN algorithm is influenced by two hyperparameters: neighborhood radius and minimum number of neighborhood samples. In this embodiment, the silhouette coefficient method is used to optimize and determine these two hyperparameters. The silhouette coefficient is an evaluation index for clustering performance. It is calculated by taking the ratio of the similarity (cohesion) of a sample with samples in the same cluster to the difference (separation) with samples in different clusters to assess the rationality and effectiveness of the clustering results. Specifically, multiple sets of candidate parameter combinations are selected in the hyperparameter search space, and the DBSCAN algorithm is run to cluster historical data for each set of parameters. The silhouette coefficient is calculated for each set of parameters, and the parameter combination with the largest silhouette coefficient is selected as the optimal neighborhood radius and minimum number of neighborhood samples. Optimizing and determining hyperparameters using the silhouette coefficient method avoids the subjectivity and blindness of human experience-based settings, further improving the accuracy and rationality of the work condition division.

[0057] After clustering using DBSCAN, historical operating data is automatically divided into several clusters, each representing a typical operating condition. Taking a hilly rural wastewater treatment system as an example, the operating conditions automatically identified by the system include, but are not limited to, high load during the wet season, low load during the dry season, accelerated filter clogging period, and peak electricity price energy storage. After clustering, a corresponding cluster center is generated for each cluster. This cluster center is used to characterize the typical features of that type of operating condition and serves as a reference benchmark for determining the current operating condition in the subsequent online real-time identification stage.

[0058] This process requires no manual labeling and can autonomously discover known and unknown operating condition structures. Furthermore, as historical operational data accumulates, clustering can be periodically re-executed to update the operating condition database, allowing the operating condition classification to adapt to the long-term evolution of system operating characteristics and avoiding the subjectivity and limitations of manual experience-based classification. Compared to manual operating condition classification or using supervised learning (such as neural networks or support vector machines) for operating condition classification, this implementation method does not require a large number of labeled samples. The model can still function normally when encountering unknown operating conditions, exhibiting stronger objectivity and adaptability. It is particularly suitable for rural wastewater treatment scenarios where the types and number of operating conditions dynamically change with seasons, climate, and village population.

[0059] Step S203: Based on the identified operating conditions, determine the control strategy corresponding to each operating condition, wherein the control strategies corresponding to different operating conditions have different multi-objective optimization weight combinations.

[0060] In one optional implementation, after clustering is completed, for each identified working condition, the control strategy to be executed is determined in advance to form a "working condition-control strategy" correspondence library. Here, the control strategy is reflected in the form of multi-objective optimization weight combination.

[0061] Specifically, for each operating condition, a weight combination of the multi-objective optimization problem is preset to form a mapping relationship between the operating condition and the weight combination. The weight combination represents the control objective orientation under the operating condition and serves as the parameterized expression of the control strategy corresponding to the operating condition. Among them, the optimization objectives of the multi-objective optimization problem include energy self-sufficiency rate, filter life, and electricity purchase cost.

[0062] Step S204: Obtain the current real-time operating data of the sewage treatment system and determine the current operating condition based on the current real-time operating data.

[0063] Specifically, during the real-time operation of the system, a group of sensors deployed at the field equipment layer continuously collects the current real-time operating data and extracts the feature vector at the current moment.

[0064] In one optional implementation, a K-nearest neighbor classifier is used to calculate the distance between the current real-time running data and the cluster centers of each working condition in step S202, and the current working condition is determined based on the distance.

[0065] K-Nearest Neighbors (KNN) classifier, as a distance-based nonparametric classification method, has low computational cost and fast decision-making speed. It does not require a complex online training process and can quickly match the current operating status to the working condition category based on offline clustering results. It is suitable for deployment on edge devices with limited computing resources, ensuring the real-time performance and reliability of working condition identification.

[0066] In one optional implementation, to prevent misjudgment due to data noise in a single determination, a sliding window voting mechanism can be used to determine the current operating condition. Specifically, operating condition determinations are performed within multiple consecutive time windows, and the determination result that occurs most frequently is taken as the final operating condition identification result. This can effectively avoid interference from noisy data on the determination result and further improve the stability and accuracy of operating condition identification.

[0067] Step S205: Based on the determined current operating condition, match and execute the control strategy corresponding to that operating condition.

[0068] In this embodiment, once the current operating condition is identified, the system immediately switches to the control strategy corresponding to that condition. For example, when it is determined that the current period is the accelerated clogging phase of the filter bed, the system actively performs backwashing, suspends other unnecessary operations, and ensures the safe operation of the filter bed; when it is determined that the current period is the peak electricity price period with priority given to energy storage, the system raises the target water level for energy storage to reduce the cost of purchasing electricity during peak hours; when it is determined that the current period is the high-load period during the wet season, the system increases power generation output, prioritizing self-consumption and feeding surplus electricity into the grid. This step directly changes the control objective based on the operating condition. The system calls the corresponding weight combination from the mapping relationship established in step S203 based on the determined current operating condition, rather than making a fixed weight trade-off among multiple objectives. This allows the system to quickly and accurately execute the most reasonable operation for the current operating condition when there are sudden changes in the operating condition.

[0069] Specifically, step S205 above includes: Step S2051: For the current working condition, retrieve the corresponding weight combination from the mapping relationship.

[0070] Step S2052: Based on the current real-time operating data and the weight combination of the call, solve the multi-objective optimization problem to obtain the optimal equipment control parameters under this operating condition.

[0071] In one alternative implementation, the equipment control parameters include target water level, inlet valve opening, booster pump start / stop status, backwash trigger flag, and aeration intensity.

[0072] Step S2053: Generate a corresponding control command sequence based on the optimal equipment control parameters, and use this control command sequence as the control strategy for the current operating condition.

[0073] Specifically, the optimal equipment control parameters obtained by solving a multi-objective optimization problem are converted into a sequence of executable control commands. These commands include, but are not limited to, valve opening adjustment commands, booster pump start / stop commands, backwash trigger commands, and aeration intensity adjustment commands. These control commands are then sent to the corresponding equipment and executed, enabling the system to operate according to the control strategy corresponding to the current operating condition. Sensors continuously transmit new real-time operating data for the next round of operating condition determination and optimization, forming a complete closed-loop control circuit. This closed-loop mechanism ensures the implementation of control actions and the continuous effectiveness of system control.

[0074] This implementation verifies the effectiveness of the previous strategy through feedback data, which serves as the input for the next round of optimization. Compared to open-loop control or timing control without feedback, open-loop control cannot correct execution deviations or environmental disturbances and is prone to deviating from the expected operating state. In contrast, the closed-loop mechanism of this implementation, through continuous data feedback and re-determination of operating conditions, enables the system to promptly perceive execution deviations and environmental changes, and to make corrections and adjustments in the next control cycle. This ensures the continuous effectiveness of the control strategy and improves the system's operational stability and adaptability in complex environments.

[0075] This implementation directly switches control strategies based on identified operating conditions, rather than seeking compromises between fixed rules or continuous multi-objective optimization. Compared to fixed-rule control (such as threshold triggering), fixed rules cannot automatically adjust when operating conditions change; for example, continuing to operate in the normal mode during a blockage period can lead to equipment damage. Compared to continuous multi-objective optimization (fixed weights or adaptive weights), continuous multi-objective optimization is slow to respond to sudden changes in operating conditions, and weight compromises may lead to the neglect of critical objectives (such as delaying backwashing for energy saving). The strategy switching mechanism of this implementation is fast-responding and logically clear, ensuring that the predetermined strategy is executed unambiguously under critical operating conditions. Its robustness is significantly better than compromise optimization, making it particularly suitable for hilly rural wastewater treatment scenarios with drastic changes in operating conditions.

[0076] In one alternative implementation, for contiguous villages, the alternating direction multiplier method is used to achieve distributed collaboration among multiple sites.

[0077] Specifically, each station exchanges only a limited amount of boundary information (such as planned drainage volume and electricity demand). When communication is interrupted, the station automatically switches to local independent operation and resumes coordination once communication is restored. This expansion scheme can improve the overall economic efficiency of the region without relying on a stable network.

[0078] This implementation uses the Alternating Directional Multiplier Method (ADMM) to achieve distributed collaboration among multiple sites, exchanging only a small amount of boundary information and automatically switching to local operation when the network is down. Compared to centralized optimization in the cloud or completely independent local control, centralized optimization in the cloud requires stable communication and is prone to failure when rural networks are unreliable; completely local control cannot achieve cross-site energy and material exchange. This implementation strikes a balance between global benefits and robustness, and is particularly suitable for hilly areas with unstable network conditions.

[0079] Using a specific natural village as an example, this invention implements an adaptive intelligent control method for wastewater treatment in the village's wastewater treatment system. The village has 128 permanent resident households with a population of approximately 460. The village covers an area of ​​approximately 0.32 square kilometers, with typical hilly terrain, an elevation ranging from 12.6m to 38.4m, and an average slope of approximately 4.7%. The village's existing wastewater treatment facilities include one elevated reservoir (effective volume 80m³), one biological contact oxidation pond, two constructed wetlands, and a supporting pipe network system. A micro-hydrogen generator set (rated power 1.2kW) and a lithium iron phosphate battery energy storage system (capacity 24kWh) are installed utilizing the village's natural elevation difference (approximately 8.5m). The village is located in a subtropical monsoon climate zone, with an average annual rainfall of 1073mm. The rainy season is concentrated from June to August (accounting for 48% of the annual rainfall), while the dry season is from November to February of the following year, exhibiting typical seasonal fluctuations in water volume.

[0080] Step a1: Data acquisition and feature extraction.

[0081] A sensor array deployed at the field equipment layer collects system operating parameters in real time at a sampling frequency of 1Hz, including the water level of the high-level reservoir (range 0-5m), the instantaneous influent flow rate (range 0-50m³ / d), the pressure difference before and after the filter (range 0-50kPa), the output power of the hydro-generator unit, and the state of charge (SOC) of the battery. The programmable logic controller (PLC) reads the sensor signals at 100ms intervals, performs filtering and engineering value conversion, and uploads the processed data to the edge computing nodes every minute. The edge nodes simultaneously acquire 72-hour rainfall forecasts and real-time time-of-use electricity price information (peak 1.124 yuan / kWh, peak 0.897 yuan / kWh, average 0.618 yuan / kWh, valley 0.328 yuan / kWh), constructing a feature vector x(t) = [L, Qin, ΔP, dΔP / dt, SOC, P] for time t. elec [R], where L is the normalized water level, Q in ΔP is the influent flow rate, dΔP / dt is the pressure difference in the filter bed, dΔP / dt is the pressure difference change rate over 6 hours, SOC is the state of charge, and P is the influent flow rate. elec The normalized electricity price is represented by R, which is the probability of rainfall in the next 24 hours.

[0082] Step a2: Offline working condition clustering and strategy library construction.

[0083] By collecting operational data from the village over the past 12 months, the number of minute-level samples is N=525,600. Unsupervised clustering was performed using the DBSCAN algorithm. After optimization using the silhouette coefficient method, the neighborhood radius ε=0.35 and the minimum neighborhood sample size MinPts=15 were set. The clustering results divided the system's operating status into four typical conditions, as shown in Table 1.

[0084] Table 1

[0085] For each operating condition, the corresponding control strategy is determined offline, forming a "operating condition - control strategy" mapping library: C1 (High-load during the wet season): Increase power generation output, prioritize self-consumption, and feed surplus power into the grid.

[0086] C2 (Low Load During Dry Season): Maintain basic operation and reduce unnecessary energy consumption.

[0087] C3 (Accelerated Filter Clogging Period): Immediately trigger backwashing, suspend other non-essential operations, and ensure filter safety.

[0088] C4 (Peak Electricity Price Priority for Energy Storage): Raise the target water level and utilize pumped storage to reduce peak electricity purchases.

[0089] This step requires no manual labeling; the algorithm autonomously discovers the working condition structure and pre-sets the most reasonable control strategy for each working condition, providing a basis for online real-time switching.

[0090] Step a3: Online operating condition identification and strategy switching.

[0091] During the real-time operation phase, the edge nodes extract the current feature vector x every minute. t The system calculates the Euclidean distance between the current working condition and the cluster centers of each working condition, uses a K-nearest neighbor classifier (K=5) to determine the current working condition, and employs a sliding window voting mechanism (window length W=15 minutes) to avoid single-point noise interference. After identifying the current working condition k, the system reads the corresponding weight combination from the working condition-weight mapping table and substitutes it into the multi-objective optimization objective function. ; Among them, E saved T represents the energy self-sufficiency rate. filter C represents the filter's lifespan (i.e., the expected lifespan of the filter, characterized by backwash cycles). power For electricity purchase costs; , , These are the weighting coefficients for energy self-sufficiency rate, filter life, and electricity purchase cost under the current operating condition k. Decision variables include: target water level (L). target ), Inlet valve opening (V)valve ), start / stop status of booster pump (P) pump ), backwash trigger flag (τ) backwash ), Aeration intensity (R) aeration ).

[0092] The constraints include: Water level limits: 0.5m ≤ L ≤ 4.5m; Equipment capacity: Q in ≤45m 3 / d, P pump,max =1.5kW; Effluent water quality: COD≤50mg / L, NH3-N≤5mg / L, TP≤0.5mg / L; Using a model predictive control framework, the time-domain N is predicted. p =6 hours, control time domain N c =2 hours, solve the following optimization problem within each control cycle (15 minutes): ; in, Represents the (t+i)th... Control the change in input at any given moment; This indicates the control increment penalty coefficient; This represents the objective function value predicted at time t+i from time t. During the solution process, the weights of the current operating condition are fixed at pre-stored values ​​and are not adjusted online. For example: At 14:30 on July 15, 2025, the system detected the real-time feature vector x. t =[0.78, 27.3, 3.5, 0.05, 0.62, 0.88, 75], KNN classification determines the current operating condition as C4 (peak electricity price, energy storage priority), and the weights (0.45, 0.15, 0.40) are substituted into the optimization model. Considering the 6-hour rainfall forecast (12mm), real-time electricity price (1.124 yuan / kWh), and current SOC (62%), the optimal control command is obtained: target water level L. target =3.8m (lifted 0.5m), inlet valve opening 65%, booster pump started, backwash not triggered, aeration intensity 18m³ / h. This instruction prioritizes pumping water for energy storage during peak electricity price periods, and is expected to increase energy storage by 45kWh in 6 hours, saving approximately 50.6 yuan in electricity purchase costs.

[0093] Step a4: Multi-site distributed collaborative optimization based on ADMM.

[0094] To address the hydraulic connectivity (upstream and downstream drainage) and electrical coupling (transformer sharing) between this natural village and its neighboring villages, the Alternating Directional Multiplier Method (ADMM) is further employed to achieve distributed collaborative optimization across multiple sites. Let the local decision variable for each site i be: planned drainage volume Q.out,i (Unit: m³ / h), Planned power generation P gen,i (Unit: kW) and target water level L target,i (Unit: m). The global system needs to satisfy two coupled constraints: first, upstream and downstream flow balance, that is, the upstream drainage plus natural inflow should equal the total inflow of each downstream station; second, microgrid power balance, that is, the total power generation of each station plus the power supply from the grid should equal the total power load plus line losses.

[0095] The ADMM algorithm iteration format is as follows: ; ; Among them, u i Let be the decision variable for station i, λ be the Lagrange multiplier, ρ be the penalty parameter (taken as 0.5), and b be the objective value of the global coupling constraint; U i For the feasible domain of site i, A i Let M be the coupling constraint matrix for station i, M be the total number of stations participating in the collaboration, and j be the other stations besides station i. Each edge node exchanges boundary variables (planned drainage volume, required power) every 30 minutes via the LoRa wireless communication module, and converges to the global optimum after about 15 iterations.

[0096] The optimization results show that the planned drainage capacity of this village is 4.8 m³ / h, which can be used for irrigation of downstream villages, replacing the energy consumption of drawing water from the river (approximately 1.2 kWh / m³). The planned power generation is 0.8 kW, with surplus electricity fed into the grid, alleviating peak electricity loads at adjacent sites. The overall electricity purchase cost of the three sites is reduced by 12.3%, which is better than the 8.7% reduction achieved when each site was optimized independently. When communication is interrupted at a site, that site automatically switches to local independent operation mode, continuing control based on locally cached operating condition policies. It rejoins the collaboration after communication is restored, ensuring the robustness of the system in complex network environments.

[0097] The adaptive intelligent control method for wastewater treatment provided in this embodiment has the following beneficial effects: 1. Adaptive switching of operating conditions to avoid control mismatch.

[0098] Traditional methods employ fixed weights or continuous multi-objective optimization, which struggles to balance all objectives under varying operating conditions, easily leading to significant deviations of some objectives from expectations. This invention directly switches strategies based on real-time assessments of operating conditions. For example, when entering the "accelerated filter clogging period," backwashing is immediately executed, avoiding equipment damage caused by delayed backwashing in pursuit of energy conservation; when entering the "peak electricity price priority for energy storage" period, the target water level is proactively raised for pumping and energy storage, reducing electricity purchase costs. Taking this village as an example, the annual electricity purchase cost decreased from approximately 4636 yuan to 3419 yuan, a reduction of 26.3%; the annual backwashing frequency of the filter decreased from 52 times to 38 times (a reduction of 27%), and the filter media replacement cycle was extended from 5 years to 6.5 years.

[0099] Accordingly, this invention constructs a control architecture of "offline clustering and database construction - online identification and judgment - instantaneous strategy switching", which directly switches the control strategy (such as active backwashing during blockage period, priority energy storage during peak electricity price, and priority power generation during wet season) based on the real-time judgment of the working conditions. This avoids the problems caused by fixed weight compromise and improves the robustness and economy of the system in sudden change environment.

[0100] 2. No complex online optimization is required, resulting in a light computational burden.

[0101] Traditional multi-objective optimization requires solving a nonlinear programming problem in each control cycle, which places high demands on the computing power of edge nodes. This invention combines offline clustering with online table lookup. The online phase only requires feature extraction and distance calculation, making it suitable for deployment on low-cost embedded devices and reducing hardware investment and maintenance costs.

[0102] 3. It is highly robust and adaptable to network fluctuations.

[0103] The system relies on local edge nodes to complete operational condition identification and policy switching, without depending on real-time cloud communication. Even during network outages, the system can still operate normally based on locally cached clustering models and historical data. Optionally, the multi-site distributed collaboration (ADMM) automatically switches to local independent operation mode when communication is interrupted, maintaining basic processing functions for up to 6 hours of network outage, and resynchronizing within 30 seconds after communication is restored, ensuring the reliability of the regional system.

[0104] Accordingly, for unstable network scenarios, this invention adopts the Alternating Directional Multiplier Method (ADMM) to achieve multi-site distributed collaboration, exchanging only a small amount of boundary information. When communication is interrupted, it automatically switches to local independent operation and re-collaborates after recovery, thus enhancing system robustness while maintaining global optimization capabilities.

[0105] 4. The model is evolvable and has strong long-term adaptability.

[0106] As runtime increases, the cloud can periodically incorporate new data into historical datasets, re-execute DBSCAN clustering, update the operating condition database and classifier, and then distribute the data to edge nodes. This mechanism enables the system to adapt to long-term changes such as village population shifts, industrial restructuring, and climate pattern evolution, achieving self-evolution of control strategies.

[0107] Accordingly, this invention forms a complete closed loop of "data acquisition - feature extraction - working condition identification - strategy switching - instruction execution - feedback verification - model update", and periodically re-executes unsupervised clustering to update the working condition library and classifier, enabling the system to continuously adapt to long-term evolution and realize the self-evolution of the strategy library.

[0108] 5. Unsupervised clustering automatically discovers operating conditions.

[0109] This invention is the first to apply the DBSCAN unsupervised clustering algorithm to a rural wastewater treatment system in hilly areas. Through feature design and parameter optimization (profile coefficient method), it automatically divides typical operating conditions from historical data, such as high load during the wet season, low load during the dry season, accelerated filter clogging period, and priority for energy storage during peak electricity prices, overcoming the subjectivity and limitations of manual division.

[0110] This embodiment also provides an intelligent control device for wastewater treatment with adaptive operating conditions. This device is used to implement the above embodiments and preferred embodiments, and details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0111] This embodiment provides an intelligent control device for wastewater treatment that adapts to changing operating conditions, such as... Figure 3 As shown, it includes: Historical data acquisition module 301 is used to acquire historical operating data of the wastewater treatment system; The working condition classification module 302 is used to cluster historical operating data using a clustering algorithm to identify several working conditions; The strategy preset module 303 is used to determine the control strategy corresponding to each working condition based on the identified working conditions. The control strategies corresponding to different working conditions have different multi-objective optimization weight combinations. The real-time judgment module 304 is used to acquire the current real-time operating data of the sewage treatment system and determine the current operating condition based on the current real-time operating data. The strategy execution module 305 is used to match and execute the control strategy corresponding to the current working condition based on the determined current working condition.

[0112] In one optional implementation, the policy execution module 305 includes: The strategy invocation unit is used to invoke the corresponding weight combination from the mapping relationship for the current working condition.

[0113] The multi-objective solution unit is used to solve multi-objective optimization problems based on the current real-time running data and the weight combination called, so as to obtain the optimal equipment control parameters under the current operating condition.

[0114] The control strategy execution unit is used to generate a corresponding control instruction sequence based on the optimal equipment control parameters, and to use the control instruction sequence as the control strategy for the current operating condition.

[0115] The adaptive wastewater treatment intelligent control device provided in this embodiment of the invention can execute the adaptive wastewater treatment intelligent control method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects for executing the method. Further functional descriptions of the above modules and units are the same as in the corresponding embodiments described above, and will not be repeated here.

[0116] Figure 4 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0117] The following is a detailed reference. Figure 4 This diagram illustrates a structural schematic suitable for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 401, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 402 or a program loaded from memory 408 into random access memory (RAM) 403. The RAM 403 also stores various programs and data required for the operation of the electronic device. The processor 401, ROM 402, and RAM 403 are interconnected via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0118] Typically, the following devices can be connected to I / O interface 405: input devices 406 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 407 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 408 including, for example, magnetic tapes, hard disks, etc.; and communication devices 409. Communication device 409 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 4 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0119] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 409, or installed from a memory 408, or installed from a ROM 402. When the computer program is executed by the processor 401, it performs the functions defined in the condition-adaptive intelligent control method for wastewater treatment according to embodiments of the present invention.

[0120] Figure 4 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of the present invention.

[0121] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that computers, processors, microprocessor controllers, or programmable hardware include storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the adaptive wastewater treatment intelligent control method shown in the above embodiments is implemented.

[0122] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0123] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A condition-adaptive intelligent control method for wastewater treatment, characterized in that, The method includes: Obtain historical operating data of the wastewater treatment system; Clustering algorithms are used to cluster the historical operating data to identify several operating conditions; Based on the identified operating conditions, a control strategy corresponding to each operating condition is determined, wherein the control strategies corresponding to different operating conditions have different combinations of multi-objective optimization weights. Obtain the current real-time operating data of the wastewater treatment system, and determine the current operating condition based on the current real-time operating data; Based on the determined current operating condition, match and execute the control strategy corresponding to that operating condition.

2. The method according to claim 1, characterized in that, The historical operational data includes a first type of characteristic data related to the material flow state of wastewater treatment, and a second type of characteristic data related to the energy flow state and / or economic state of the wastewater treatment system.

3. The method according to claim 2, characterized in that, The first type of characteristic data includes at least water level, flow rate, filter pressure difference, and pressure difference change rate; the second type of characteristic data includes at least energy storage status and electricity price.

4. The method according to claim 3, characterized in that, The historical operational data also includes the probability of rainfall.

5. The method according to claim 1, characterized in that, The clustering algorithm is the DBSCAN algorithm, and the neighborhood radius and minimum number of neighborhood samples of the DBSCAN algorithm are determined by optimization using the silhouette coefficient method.

6. The method according to claim 1, characterized in that, The determination of the control strategy corresponding to each operating condition includes: For each working condition, a weight combination of multi-objective optimization problems is preset to form a mapping relationship between the working condition and the weight combination. The weight combination represents the control objective orientation under the working condition and serves as the parameterized expression of the control strategy corresponding to the working condition. The optimization objectives of the multi-objective optimization problem include energy self-sufficiency rate, filter life, and electricity purchase cost.

7. The method according to claim 1, characterized in that, The step of determining the current operating condition based on the current real-time operating data includes: The K-nearest neighbor classifier is used to calculate the distance between the current real-time running data and the cluster centers of each operating condition, and the current operating condition is determined based on the distance.

8. The method according to claim 6, characterized in that, The step of matching and executing the control strategy corresponding to the determined current operating condition includes: For the current operating condition, the corresponding weight combination is retrieved from the mapping relationship; Based on the current real-time operating data and the weight combination of the calls, the multi-objective optimization problem is solved to obtain the optimal equipment control parameters under this operating condition; Based on the optimal equipment control parameters, a corresponding control command sequence is generated, and this control command sequence is used as the control strategy for the current operating condition.

9. A condition-adaptive intelligent control device for wastewater treatment, characterized in that, The device includes: The historical data acquisition module is used to acquire historical operating data of the wastewater treatment system; The working condition classification module is used to cluster the historical operating data using a clustering algorithm to identify several working conditions; The strategy preset module is used to determine the control strategy corresponding to each working condition based on the identified working conditions, wherein the control strategies corresponding to different working conditions have different multi-objective optimization weight combinations. The real-time judgment module is used to acquire the current real-time operating data of the wastewater treatment system and determine the current operating condition based on the current real-time operating data. The strategy execution module is used to match and execute the control strategy corresponding to the current operating condition based on the determined current operating condition.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions for causing the computer to execute the adaptive intelligent control method for wastewater treatment according to any one of claims 1 to 8.

Images