Intelligent optimization method and device for distributed rural sewage treatment station

Abstract

This invention provides an intelligent optimization method and device for decentralized rural sewage treatment stations, relating to the field of sewage treatment technology. The invention acquires historical sewage measurement data and corresponding rational control decision data at the sewage treatment station. Static conventional sewage indicators are obtained from the static sewage measurement data. Dynamic sewage treatment data is analyzed to obtain dynamic impact sequence data. Using the static conventional sewage indicators and dynamic impact sequence data as inputs and the corresponding rational control decision data as outputs, a sewage treatment decision optimization model is constructed and trained. The current static conventional sewage indicators and dynamic impact sequence data of the sewage treatment station are acquired, and the current first decision data is obtained through the sewage treatment decision optimization model. Control decision data within the most recent control window is acquired, and the current first decision data is adaptively adjusted using the control decision data within the most recent control window to obtain the final decision data.

Description

Technical Field

[0001] This invention relates to the field of wastewater treatment technology, specifically to a method and apparatus for intelligent optimization of decentralized rural wastewater treatment sites. Background Technology

[0002] Activated sludge technologies used in rural wastewater treatment mainly include the A2 / O process, SBR process, and MBR process. The A2 / O process, also known as the AAO process (Anaerobic-Anoxic-Oxic), is a commonly used secondary biologically enhanced activated sludge treatment process with excellent nitrogen and phosphorus removal effects. Wastewater and returned sludge first enter the anaerobic zone and are completely mixed to form a mixed liquor. Through anaerobic biochemical processes, some BOD is removed, while polyphosphate-accumulating microorganisms in the returned sludge release phosphorus, meeting the microbial demand for phosphorus. The mixed liquor then flows into the anoxic zone, where denitrifying microorganisms use carbon-containing organic matter as a carbon source to reduce nitrate ions from the aerobic zone through internal circulation to form denitrifying nitrate ions. The wastewater is released to achieve denitrification; the mixed liquor then flows into the aerobic zone, where carbonaceous organic matter in the wastewater serves as a substrate for heterotrophic bacteria growth and is further degraded under aerobic conditions. Nitrification produces nitrate ions, while the organic matter in the water is oxidized and decomposed to provide energy for phosphorus-absorbing microorganisms. The microorganisms absorb excess phosphorus from the water, and after sedimentation and separation, it is discharged from the system in the form of phosphorus-rich sludge, thus achieving biological phosphorus removal.

[0003] Currently, decentralized rural wastewater treatment plants, due to their geographical dispersion, small scale, and complex operating conditions, generally suffer from problems such as incomplete monitoring data, reliance on experience for control strategies, and delayed regulation. Traditional AAO process wastewater treatment ponds often rely solely on static water quality indicators or simple operational condition monitoring for aeration, recirculation, and dosage adjustments. This fails to fully reflect the dynamic imbalance between dissolved oxygen supply and demand and the changing trends of the sluggish response of the microbial community, leading to delayed operational decisions or accumulated biases, easily causing short-term fluctuations and even long-term declines in treatment effectiveness. Furthermore, the lack of professional personnel for real-time monitoring of decentralized plants makes it difficult for existing technologies to achieve both precise and stable intelligent optimization control.

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

[0005] The purpose of this invention is to provide an intelligent optimization method and device for decentralized rural sewage treatment sites to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] A method for intelligent optimization of decentralized rural wastewater treatment plants, comprising the following steps:

[0008] Step 1: Obtain historical reasonable control decision data and corresponding wastewater measurement data at the wastewater treatment site. The wastewater measurement data includes static wastewater measurement data and dynamic wastewater treatment data. The control decision data includes aeration power, reflux ratio, and carbon source dosage.

[0009] Step 2: Analyze the static wastewater measurement data to obtain static conventional wastewater indicators, and analyze the dynamic wastewater treatment data to obtain dynamic impact sequence data. The dynamic impact sequence data includes oxygen transfer rate sequence data reflecting oxygen transfer during wastewater treatment and microbial hysteresis response sequence data reflecting the inhibition of nitrifying aerobic microorganisms.

[0010] Step 3: Using static conventional wastewater indicators and dynamic impact sequence data as inputs and corresponding rational control decision data as outputs, construct and train a wastewater treatment decision optimization model; obtain the current static conventional wastewater indicators and dynamic impact sequence data of the wastewater treatment station and input them into the wastewater treatment decision optimization model to obtain the current first decision data;

[0011] Step 4: Obtain the control decision data in the most recent control window, and use the control decision data in the most recent control window to perform trend contraction adaptive adjustment on the current first decision data to obtain the final decision data.

[0012] Furthermore, control decision data and corresponding wastewater measurement data for each moment within the historical treatment period are obtained. A time length threshold is preset. If there is a time period within the historical treatment period with a time length equal to the time length threshold, and this time period meets the decision rationality rule, then this time period is taken as the rational interval. The control decision data within the rational interval is the rational control decision data. The rational control decision data and corresponding wastewater measurement data are retrieved.

[0013] The reasonable range of ammonia nitrogen content, pH value and heavy metal content is preset. If the ammonia nitrogen content, pH value and heavy metal content at all times within a time period equal to the time length threshold are within the corresponding reasonable range, it is considered to meet the decision rationality rule.

[0014] The sewage storage tank of the sewage treatment plant is divided into N areas of equal area. The static sewage measurement data are the biochemical oxygen demand, ammonia content, phosphorus content, nitrogen content, organic carbon content, pH value and heavy metal content of each area in the sewage storage tank.

[0015] The dynamic wastewater treatment data includes time-series data of dissolved oxygen content in the treatment tank, time-series data of aeration power, time-series data of dissolved oxygen content at the inlet of the treatment tank, time-series data of ammonia nitrogen content at the inlet of the treatment tank, and time-series data of ammonia nitrogen change rate in the treatment tank.

[0016] Furthermore, the static wastewater measurement data are normalized, and the mean of the normalized static wastewater measurement data is analyzed to obtain static conventional wastewater indicators.

[0017] Furthermore, within a reasonable range, the time-series data of dissolved oxygen content and aeration power in the treatment tank were analyzed to obtain oxygen transfer rate sequence data, with the following logic:

[0018] The dissolved oxygen content time series data of the treatment tank is used as the distance from the saturated dissolved oxygen concentration as the dissolved oxygen difference; power functions are constructed based on the aeration power time series data and the volume of the treatment tank to obtain the aeration power time series influence data and the volume influence data; the oxygen mass transfer coefficient data is obtained by weighted fitting of the product of the aeration power time series influence data and the volume influence data; the oxygen transfer rate sequence data of the treatment tank is obtained by weighting the dissolved oxygen difference based on the oxygen mass transfer coefficient data.

[0019] Furthermore, hysteresis response sequence data is obtained by using time-series data of dissolved oxygen content, ammonia nitrogen content, and ammonia nitrogen change rate at the inlet of the treatment tank. Specifically, for each moment within a reasonable interval, the time window is used as the end point of the time window. The time-series data of dissolved oxygen content, ammonia nitrogen content, and ammonia nitrogen change rate at the inlet of the treatment tank are extracted using the time window. The extracted time-series data of dissolved oxygen content and ammonia nitrogen content at the inlet of the treatment tank are weighted and summed to obtain the extracted input time data. The extracted time-series data of ammonia nitrogen change rate at the inlet of the treatment tank is used as the output time data. For each moment within a reasonable interval, the hysteresis response at which the Pearson correlation coefficient between the extracted input time data and the extracted output time data is maximized is obtained. The microbial hysteresis response at each moment constitutes the microbial hysteresis response sequence data.

[0020] Furthermore, using static conventional wastewater indicators, oxygen transfer rate sequence data, and microbial hysteresis response sequence data within a reasonable range as inputs, and reasonable control decision data at the end of the reasonable range as outputs, a wastewater treatment decision optimization model is constructed and trained.

[0021] Obtain the current interval defined with the current time as the endpoint, and the time length of the current interval is the same as that of the reasonable interval. Obtain the static conventional wastewater indicators, oxygen transfer rate sequence data and microbial hysteresis response sequence data of the current interval, and input them into the wastewater treatment decision optimization model to obtain the current first decision data.

[0022] Furthermore, the most recent control window is a control window that traces back a certain time length from the current time. The decision control data within the most recent control window is sampled at equal time intervals to obtain H sets of historical decision control data. The average value of the first decision data at the current time and the H sets of historical decision control data is used as the final decision data.

[0023] The present invention further provides an intelligent optimization device for decentralized rural sewage treatment plants, the device being used in the aforementioned intelligent optimization method for decentralized rural sewage treatment plants, specifically comprising:

[0024] The data acquisition module is used to acquire historical reasonable control decision data and corresponding wastewater measurement data at the wastewater treatment station. The wastewater measurement data includes static wastewater measurement data and dynamic wastewater treatment data, and the control decision data includes aeration power, reflux ratio, and carbon source dosage.

[0025] The feature extraction module is used to analyze static wastewater measurement data to obtain static conventional wastewater indicators and to analyze dynamic wastewater treatment data to obtain dynamic impact sequence data. The dynamic impact sequence data includes oxygen transfer rate sequence data reflecting oxygen transfer during wastewater treatment and microbial hysteresis response sequence data reflecting the inhibition of nitrifying aerobic microorganisms.

[0026] The initial decision-making module is used to construct and train a wastewater treatment decision optimization model by taking static conventional wastewater indicators and dynamic impact sequence data as inputs and corresponding reasonable control decision data as outputs; it obtains the current static conventional wastewater indicators and dynamic impact sequence data of the wastewater treatment station and inputs them into the wastewater treatment decision optimization model to obtain the current first decision data;

[0027] The decision optimization module is used to obtain the control decision data in the most recent control window, and use the control decision data in the most recent control window to perform trend contraction adaptive adjustment on the current first decision data to obtain the final decision data.

[0028] Compared with the prior art, the beneficial effects of the present invention are:

[0029] This invention combines historical wastewater measurement data with control decision data to extract dynamic sequence features reflecting oxygen transfer efficiency and microbial lag effects. These features are then fused with static conventional indicators to form a model, thus simultaneously considering the wastewater quality baseline state and dynamic evolution patterns in the control model. Based on this, an adaptive adjustment mechanism with trend contraction is employed to account for both the lag inertia of biochemical reactions and to smooth the decision curve, effectively avoiding in-pool fluctuations caused by environmental disturbances and model lag. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the overall method flow of the present invention;

[0031] Figure 2 A comparison chart of ammonia nitrogen content in treated water with and without the present invention's adaptive adjustment for trend contraction.

[0032] Figure 3 A comparison chart of lead content in treated water with and without trend contraction adaptive adjustment according to the present invention;

[0033] Figure 4 This is a schematic diagram of the overall device structure of the present invention. Detailed Implementation

[0034] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0035] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0036] Example:

[0037] Please see Figure 1 The present invention provides a technical solution:

[0038] A method for intelligent optimization of decentralized rural wastewater treatment plants, comprising the following steps:

[0039] Step 1: Obtain historical reasonable control decision data and corresponding wastewater measurement data at the wastewater treatment site. The wastewater measurement data includes static wastewater measurement data and dynamic wastewater treatment data. The control decision data includes aeration power, reflux ratio, and carbon source dosage.

[0040] The wastewater treatment station described in this embodiment includes a storage tank, a treatment tank, and an effluent tank. The storage tank is used to store wastewater for orderly discharge into the treatment tank. The treatment tank is used to treat wastewater to output treated water. If the treated water quality meets the standards, it is discharged into the effluent tank; otherwise, it is discharged into the inlet of the treatment tank through a pipeline.

[0041] The system acquires control decision data and corresponding wastewater measurement data for each moment within a historical treatment period. A preset time length threshold is used. If a time period equal to the threshold exists within the historical treatment period and meets the decision rationality rules, this time period is designated as a reasonable interval. The control decision data within this reasonable interval is then considered the reasonable control decision data. Wastewater measurement data and corresponding reasonable decision data are retrieved from each reasonable interval. The length of the reasonable interval is determined based on the performance of the wastewater treatment station. Specifically, the system acquires the time from the start of wastewater treatment at the historical wastewater treatment station to the first time the treated water meets the standards, and uses the average of this time as the length of the reasonable interval.

[0042] The reasonable range of ammonia nitrogen content, pH value and heavy metal content is preset. If the ammonia nitrogen content, pH value and heavy metal content at all times within a time period equal to the time length threshold are within the corresponding reasonable range, it is considered to meet the decision rationality rule.

[0043] Based on the relevant requirements for rural decentralized sewage treatment stations in the "Pollutant Discharge Standards for Urban Sewage Treatment Plants" or similar regulations promulgated by the state or local governments, determine the reasonable range of values ​​for various effluent water quality indicators.

[0044] The sewage storage tank of the sewage treatment plant is divided into N areas of equal area. The static sewage measurement data are the biochemical oxygen demand, ammonia content, phosphorus content, nitrogen content, organic carbon content, pH value and heavy metal content of each area in the sewage storage tank.

[0045] The dynamic wastewater treatment data includes time-series data of dissolved oxygen content in the treatment tank, time-series data of aeration power, time-series data of dissolved oxygen content at the inlet of the treatment tank, time-series data of ammonia nitrogen content at the inlet of the treatment tank, and time-series data of ammonia nitrogen change rate in the treatment tank.

[0046] In wastewater treatment, the aeration system is the largest energy-consuming unit, and any optimization method must take aeration power as the core monitoring and optimization object.

[0047] In activated sludge processes, especially the AAO process with denitrification capabilities, one of the core tasks is removing ammonia nitrogen from wastewater. This process mainly relies on a relay race between two types of microbial communities:

[0048] Nitrification: Under aerobic conditions, autotrophic nitrifying bacteria convert ammonia nitrogen into nitrate. This is an aerobic process. Nitrifying bacteria must be in the presence of oxygen to function. The dissolved oxygen content directly determines the rate of nitrification. Insufficient dissolved oxygen will slow down or even stop the nitrification process, resulting in ineffective removal of ammonia nitrogen.

[0049] Denitrification: Under anaerobic conditions, heterotrophic denitrifying bacteria convert nitrates into nitrogen gas, which escapes from the water. This process requires a carbon source as food but does not require oxygen. The level of ammonia nitrogen concentration determines the nitrification task the system needs to complete. Furthermore, the rate of change in ammonia nitrogen concentration directly reflects the current intensity of nitrification.

[0050] Step 2: Analyze the static wastewater measurement data to obtain static conventional wastewater indicators, and analyze the dynamic wastewater treatment data to obtain dynamic impact sequence data. The dynamic impact sequence data includes oxygen transfer rate sequence data reflecting oxygen transfer during wastewater treatment and microbial hysteresis response sequence data reflecting the inhibition of nitrifying aerobic microorganisms.

[0051] Furthermore, the static wastewater measurement data are normalized using the following formula:

[0052]

[0053] in, To indicate the first Block region Normalized data of static wastewater measurement data, For the first Block region The raw data of static wastewater measurement data, For the first The maximum value of static wastewater measurement data, For the first The minimum value of static wastewater measurement data, An index for the types of static wastewater measurements. Index for the wastewater storage tank area;

[0054] The static conventional wastewater index is obtained by mean analysis of the normalized static wastewater measurement data, using the following formula:

[0055]

[0056] in, For the first A type of static, conventional wastewater indicator.

[0057] Static conventional wastewater indicators reflect the inherent characteristics and initial pollution load of the wastewater when it enters the treatment tank. It depicts a stable state of wastewater before it enters the dynamic biochemical reaction. Due to the decentralized nature of rural treatment plants, wastewater may not be completely mixed instantaneously and requires standardization and averaging to reflect the state of the wastewater before it enters the treatment tank.

[0058] Furthermore, within a reasonable range, the time-series data of dissolved oxygen content and aeration power in the treatment tank were analyzed to obtain oxygen transfer rate sequence data, with the following logic:

[0059] The dissolved oxygen content time series data of the treatment tank is used as the distance from the saturated dissolved oxygen concentration as the dissolved oxygen difference; power functions are constructed based on the aeration power time series data and the volume of the treatment tank to obtain the aeration power time series influence data and the volume influence data; the oxygen mass transfer coefficient data is obtained by weighted fitting of the product of the aeration power time series influence data and the volume influence data; the oxygen transfer rate sequence data of the treatment tank is obtained by weighting the dissolved oxygen difference based on the oxygen mass transfer coefficient data.

[0060] The formula is:

[0061]

[0062]

[0063] in, For the oxygen transfer rate sequence data of the treatment tank, For time variables within a reasonable range, This is a sequence of oxygen mass transfer coefficients. For time-series data of dissolved oxygen content in the treatment pond, This is time-series data on aeration power. This represents the saturated dissolved oxygen concentration. For the volume of the treatment tank, , , The constant is an empirical constant, fitted using historical data. As the starting point of the reasonable range, This is the endpoint of the reasonable range.

[0064] in, The power function weights used to represent the time series data of aeration power The power function weights used to represent the volume of the processing pool. A combined linear weight used to represent the time-series data of aeration power and the volume of the treatment tank.

[0065] Oxygen transfer rate (OTR) measures the speed at which oxygen dissolves from the gas phase into the liquid phase. Its sequence data reflects the overall efficiency of the aeration system and the actual oxygen consumption environment of the treatment tank at a specific point in time. A higher OTR indicates stronger "oxygen supply power," better system operation, sufficient contact between bubbles and water, rapid oxygen dissolution, and active microbial communities that are rapidly consuming pollutants and depleting large amounts of oxygen. This maintains a high oxygen concentration gradient while keeping aeration power constant, driving oxygen transfer. The system is capable of providing sufficient oxygen for high-intensity biochemical reactions (such as nitrification), indicating good treatment results. However, it can also lead to over-aeration. If the OTR is high, but the dissolved oxygen in the tank is already high, and the concentrations of pollutants such as ammonia nitrogen are low, this usually means energy waste. The system is "idling," providing far more oxygen than needed.

[0066] Traditional methods typically control dissolved oxygen at a fixed setpoint (e.g., 2 mg / L). This is a lagging and passive control method because it only cares about "how much oxygen is left in the tank," not "how fast the oxygen dissolves." This embodiment, by monitoring the oxygen transfer rate sequence, focuses on "the efficiency of oxygen supply." If the system predicts an upcoming high ammonia nitrogen load, it will increase the aeration power in advance to ensure that the oxygen transfer rate matches the upcoming high oxygen demand, thus achieving proactive control.

[0067] Furthermore, hysteresis response sequence data is obtained by using time-series data of dissolved oxygen content, ammonia nitrogen content, and ammonia nitrogen change rate at the treatment tank inlet. Specifically, for each moment within a reasonable interval, using that moment as the end of the time window, the time-series data of dissolved oxygen content, ammonia nitrogen content, and ammonia nitrogen change rate at the treatment tank inlet are extracted using the time window. The extracted time-series data of dissolved oxygen content and ammonia nitrogen content at the treatment tank inlet are then weighted and summed to obtain the extracted input time data, represented as follows:

[0068]

[0069] To extract the input time data, The data represents the time-series dissolved oxygen content at the inlet of the treatment tank. The data represents the time series of ammonia nitrogen content at the inlet of the treatment tank. , The weighting coefficients are fitted using historical data. The length of the time window, This refers to the time variable within the corresponding time window, with time t as the endpoint within a reasonable interval. The purpose of using a window to extract data is to analyze the instantaneous delay characteristics at the corresponding moment. However, since the delay characteristics need to be analyzed over a time period, the delay characteristics of the time period extracted by the time window are used to characterize the instantaneous delay characteristics. Therefore, the length of the time window should not be too large, and should not exceed 5 seconds; the shorter the window, the more accurate it is.

[0070] The time-series data of ammonia nitrogen change rate from the extracted treatment tank is used as the output time data, represented as follows:

[0071]

[0072] in, To extract the output time data, The data represents the time-series data of ammonia nitrogen change rate in the treatment tank.

[0073] The maximum time-shift cross-correlation between the truncated input time data and the truncated output time data is calculated and expressed as:

[0074]

[0075] in, For time t within a reasonable interval, the hysteresis is... Pearson correlation coefficient at time As a lagged variable; the lag with the largest absolute value of the Pearson correlation coefficient at time t within a reasonable interval is selected as the lagged variable. The delayed response of microorganisms at any given moment To calculate the Pearson correlation coefficient, the microbial hysteresis response at each time step is used to construct a microbial hysteresis response sequence. ,in, Let t be the microbial hysteresis response at time t.

[0076] Microbial lag response reflects the "health status" and "responsiveness" of the microbial community (especially nitrifying bacteria) in a wastewater treatment system. Specifically, it quantifies the time required for the microbial community to "sense" changes in influent water quality and then "mobilize" its enzyme system to accelerate metabolism. This time lag includes the time consumed by a series of complex physiological processes such as microbial adsorption and transport of the substrate and the initiation of internal biochemical processes. A longer lag response time is a clear "negative signal," indicating a decline in the performance of the microbial system and low or unhealthy microbial activity: nitrifying bacteria may be inhibited and sluggish in their response. This embodiment transforms the lag response from a concept into quantifiable sequence data, allowing the system to directly "diagnose" the microorganisms through the lag response sequence. If the lag time is longer, even if dissolved oxygen and ammonia nitrogen are currently normal, the system can provide early warning, giving the control strategy a fundamentally forward-looking approach.

[0077] Step 3: Using static conventional wastewater indicators and dynamic impact sequence data as inputs and corresponding rational control decision data as outputs, construct and train a wastewater treatment decision optimization model; obtain the current static conventional wastewater indicators and dynamic impact sequence data of the wastewater treatment station and input them into the wastewater treatment decision optimization model to obtain the current first decision data;

[0078] Using static conventional wastewater indicators, oxygen transfer rate sequence data, and microbial hysteresis response sequence data within a reasonable range as inputs, and reasonable control decision data at the end of the reasonable range as outputs, a wastewater treatment decision optimization model is constructed and trained.

[0079] Obtain the current interval defined with the current time as the endpoint, and the time length of the current interval is the same as that of the reasonable interval. Obtain the static conventional wastewater indicators, oxygen transfer rate sequence data and microbial hysteresis response sequence data of the current interval, and input them into the wastewater treatment decision optimization model to obtain the current first decision data.

[0080] Extracting oxygen transfer rate sequence data and microbial hysteresis response sequence data as model inputs represents a qualitative leap from "perceiving phenomena" to "understanding mechanisms," offering a fundamental advantage compared to models that only use conventional raw data. Conventional models can only perceive the apparent state of the system (such as low dissolved oxygen or high effluent ammonia nitrogen) and thus implement delayed and blind reactive control. They cannot distinguish whether a low dissolved oxygen phenomenon stems from a decline in the efficiency of the aeration system or a sudden increase in microbial oxygen consumption, resulting in crude and risky decision-making. The two advanced features introduced in this embodiment are: the oxygen transfer rate sequence, which directly quantifies the overall working efficiency of the aeration system, enabling the model to accurately judge the health status and performance trend of the oxygen supply equipment, thereby achieving precise and forward-looking adjustment of aeration power; and the microbial hysteresis response sequence, which acts like a biosensor embedded in the process, dynamically capturing the activity and health of the microbial community (especially nitrifying bacteria), allowing the model to detect the "discomfort" of the microbial community before the effluent indicators deteriorate, and diagnose that the root cause of the problem is microbial inhibition rather than simply insufficient dissolved oxygen, thereby avoiding ineffective aeration energy consumption and turning to root-cause treatment such as adjusting sludge volume, pH, or investigating toxic substances.

[0081] Acquire static routine wastewater indicators, oxygen transfer rate sequence data, microbial hysteresis response sequence data, and corresponding reasonable control decision data within a reasonable range during historical treatment periods; and construct a dataset, using 80% of the dataset as the training set and 20% as the validation set.

[0082] The wastewater treatment decision optimization model employs a feedforward neural network. This model is trained using static conventional wastewater indicators, oxygen transfer rate sequence data, and microbial delayed response sequence data within a reasonable timeframe to determine reasonable control decision data. Specifically, the model uses these data as inputs, along with the reasonable control decision data as labels. The model training, utilizing existing technology, includes an input layer, hidden layers, an output layer, and an activation function. The input layer receives these data within the reasonable timeframe. The hidden layers process this data. Each layer contains multiple time points, and the time points of each hidden layer are connected to the previous layer via weights, serving to process the static conventional wastewater indicators within the reasonable timeframe. The model performs feature abstraction and nonlinear transformation on indicators, oxygen transfer rate sequence data, and microbial hysteresis response sequence data; learns the complex mapping relationship between input and output; introduces nonlinear relationships by using the ReLU activation function, enabling the model to fit complex feature relationships; sets an independent neuron in the output layer to transform the local and high-level feature representations extracted from the hidden layer for outputting reasonable control decision data; adopts the root mean square error loss function; performs one computation on the input data through the network to obtain the output result, calculates the loss function based on the predicted and true values, calculates the gradient of the loss function with respect to each weight and bias using the chain rule, and updates the network weights and biases using the gradient descent algorithm to minimize the loss function.

[0083] Step 4: Obtain the control decision data in the most recent control window, and use the control decision data in the most recent control window to perform trend contraction adaptive adjustment on the current first decision data to obtain the final decision data.

[0084] The most recent control window is a control window that traces back a certain time length from the current time (the tracing time length is usually greater than a reasonable interval but less than twice the reasonable interval); the decision control data within the most recent control window is sampled at equal time intervals to obtain H sets of historical decision control data, and the formula for obtaining the final decision data is:

[0085]

[0086]

[0087] in, for The final decision data at any given moment. for The first decision data at any given moment For the current moment, Within the most recently controlled window, the first Decision data collected during the second sampling. Within the most recently controlled window, the first The aeration power collected during the second sampling. Within the most recently controlled window, the first The reflux ratio collected during the second sampling Within the most recently controlled window, the first The amount of carbon source added during the second sampling.

[0088] Because wastewater treatment involves biochemical reactions related to nitrifying bacteria, these reactions exhibit a certain lag (delay) characteristic. Therefore, when controlling wastewater, it's insufficient to consider only the current control parameters; it's also necessary to analyze past control trends (the average value of decision data collected within the most recent control window), combining past decision-making inertia with current needs. This is crucial for achieving more accurate and effective control. Furthermore, drastic fluctuations in decision-making lead to unstable and inefficient energy utilization, often resulting in significant energy waste. By introducing feedback based on historical control trends, adjustment mechanisms can effectively smooth out sharp jumps in the decision curve.

[0089] Please refer to Figures 2-3 Comparison charts of ammonia nitrogen content and the highest heavy metal content in the wastewater treatment pond (lead in this example) were collected at multiple times, with and without trend contraction adaptive adjustment.

[0090] Please see Figure 4 The present invention further provides an intelligent optimization device for decentralized rural sewage treatment stations. The device is used in the aforementioned intelligent optimization method for decentralized rural sewage treatment stations, and specifically includes:

[0091] The data acquisition module is used to acquire historical reasonable control decision data and corresponding wastewater measurement data at the wastewater treatment station. The wastewater measurement data includes static wastewater measurement data and dynamic wastewater treatment data, and the control decision data includes aeration power, reflux ratio, and carbon source dosage.

[0092] The feature extraction module is used to analyze static wastewater measurement data to obtain static conventional wastewater indicators and to analyze dynamic wastewater treatment data to obtain dynamic impact sequence data. The dynamic impact sequence data includes oxygen transfer rate sequence data reflecting oxygen transfer during wastewater treatment and microbial hysteresis response sequence data reflecting the inhibition of nitrifying aerobic microorganisms.

[0093] The initial decision-making module is used to construct and train a wastewater treatment decision optimization model by taking static conventional wastewater indicators and dynamic impact sequence data as inputs and corresponding reasonable control decision data as outputs; it obtains the current static conventional wastewater indicators and dynamic impact sequence data of the wastewater treatment station and inputs them into the wastewater treatment decision optimization model to obtain the current first decision data;

[0094] The decision optimization module is used to obtain the control decision data in the most recent control window, and use the control decision data in the most recent control window to perform trend contraction adaptive adjustment on the current first decision data to obtain the final decision data.

[0095] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0096] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0097] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0098] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that cannot be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for intelligent optimization of decentralized rural sewage treatment plants, characterized in that, The specific steps include: Step 1: Obtain historical reasonable control decision data and corresponding wastewater measurement data at the wastewater treatment site. The wastewater measurement data includes static wastewater measurement data and dynamic wastewater treatment data. The control decision data includes aeration power, reflux ratio, and carbon source dosage. Acquire control decision data and corresponding wastewater measurement data for each moment within the historical treatment period. Set a time length threshold. If there is a time period within the historical treatment period with a time length equal to the time length threshold, and this time period meets the decision rationality rule, then this time period is taken as the rational interval. The control decision data within the rational interval is the rational control decision data. Retrieve the rational control decision data and corresponding wastewater measurement data within each rational interval. The reasonable range of ammonia nitrogen content, pH value and heavy metal content is preset. If the ammonia nitrogen content, pH value and heavy metal content at all times within a time period equal to the time length threshold are within the corresponding reasonable range, it is considered to meet the decision reasonableness rule. Step 2: Analyze the static wastewater measurement data to obtain static conventional wastewater indicators, and analyze the dynamic wastewater treatment data to obtain dynamic impact sequence data. The dynamic impact sequence data includes oxygen transfer rate sequence data reflecting oxygen transfer during wastewater treatment and microbial hysteresis response sequence data reflecting the inhibition of nitrifying aerobic microorganisms. Step 3: Using static conventional wastewater indicators and dynamic impact sequence data as inputs and corresponding rational control decision data as outputs, construct and train a wastewater treatment decision optimization model; obtain the current static conventional wastewater indicators and dynamic impact sequence data of the wastewater treatment station and input them into the wastewater treatment decision optimization model to obtain the current first decision data; Step 4: Obtain the control decision data in the most recent control window, and use the control decision data in the most recent control window to perform trend contraction adaptive adjustment on the current first decision data to obtain the final decision data; The most recent control window is a control window that traces back a certain time length from the current time. The decision control data within the most recent control window is sampled at equal time intervals to obtain H sets of historical decision control data. The average of the first decision data at the current time and the H sets of historical decision control data is used as the final decision data.

2. The intelligent optimization method for decentralized rural sewage treatment stations according to claim 1, characterized in that: The sewage storage tank of the sewage treatment plant is divided into N areas of equal area. The static sewage measurement data are the biochemical oxygen demand, ammonia content, phosphorus content, nitrogen content, organic carbon content, pH value and heavy metal content of each area in the sewage storage tank. The dynamic wastewater treatment data includes time-series data of dissolved oxygen content in the treatment tank, time-series data of aeration power, time-series data of dissolved oxygen content at the inlet of the treatment tank, time-series data of ammonia nitrogen content at the inlet of the treatment tank, and time-series data of ammonia nitrogen change rate in the treatment tank.

3. The intelligent optimization method for decentralized rural sewage treatment stations according to claim 2, characterized in that: The static wastewater measurement data are normalized, and the mean value of the normalized static wastewater measurement data is analyzed to obtain the static conventional wastewater indicators.

4. The intelligent optimization method for decentralized rural sewage treatment stations according to claim 2, characterized in that: Within a reasonable range, the oxygen transfer rate sequence data is obtained by analyzing the time-series data of dissolved oxygen content and aeration power in the treatment tank. The logic is as follows: The dissolved oxygen content time series data of the treatment tank is used as the distance from the saturated dissolved oxygen concentration as the dissolved oxygen difference; power functions are constructed based on the aeration power time series data and the volume of the treatment tank to obtain the aeration power time series influence data and the volume influence data; the oxygen mass transfer coefficient data is obtained by weighted fitting of the product of the aeration power time series influence data and the volume influence data; the oxygen transfer rate sequence data of the treatment tank is obtained by weighting the dissolved oxygen difference based on the oxygen mass transfer coefficient data.

5. The intelligent optimization method for decentralized rural sewage treatment stations according to claim 2, characterized in that: Hysteresis response sequence data was obtained by using time-series data of dissolved oxygen content, ammonia nitrogen content, and ammonia nitrogen change rate at the inlet of the treatment tank. Specifically, for each moment within a reasonable interval, the time window was used as the end point of the time window to extract the time-series data of dissolved oxygen content, ammonia nitrogen content, and ammonia nitrogen change rate at the inlet of the treatment tank. The extracted time-series data of dissolved oxygen content and ammonia nitrogen content at the inlet of the treatment tank were weighted and summed to obtain the extracted input time data. The extracted time-series data of ammonia nitrogen change rate at the inlet of the treatment tank was used as the output time data. For each moment within the reasonable interval, the hysteresis response at which the Pearson correlation coefficient between the extracted input time data and the extracted output time data was maximized was obtained. The microbial hysteresis response at each moment was used to construct the microbial hysteresis response sequence data.

6. The intelligent optimization method for decentralized rural sewage treatment stations according to claim 5, characterized in that: Using static conventional wastewater indicators, oxygen transfer rate sequence data, and microbial hysteresis response sequence data within a reasonable range as inputs, and reasonable control decision data at the end of the reasonable range as outputs, a wastewater treatment decision optimization model is constructed and trained. Obtain the current interval defined with the current time as the endpoint, and the time length of the current interval is the same as that of the reasonable interval. Obtain the static conventional wastewater indicators, oxygen transfer rate sequence data and microbial hysteresis response sequence data of the current interval, and input them into the wastewater treatment decision optimization model to obtain the current first decision data.

7. A smart optimization device for decentralized rural sewage treatment stations, characterized in that: The device is used to implement the intelligent optimization method for decentralized rural sewage treatment stations as described in any one of claims 1-6, specifically including: The data acquisition module is used to acquire historical reasonable control decision data and corresponding wastewater measurement data at the wastewater treatment station. The wastewater measurement data includes static wastewater measurement data and dynamic wastewater treatment data, and the control decision data includes aeration power, reflux ratio, and carbon source dosage. The feature extraction module is used to analyze static wastewater measurement data to obtain static conventional wastewater indicators and to analyze dynamic wastewater treatment data to obtain dynamic impact sequence data. The dynamic impact sequence data includes oxygen transfer rate sequence data reflecting oxygen transfer during wastewater treatment and microbial hysteresis response sequence data reflecting the inhibition of nitrifying aerobic microorganisms. The initial decision-making module is used to construct and train a wastewater treatment decision optimization model by taking static conventional wastewater indicators and dynamic impact sequence data as inputs and corresponding reasonable control decision data as outputs; it obtains the current static conventional wastewater indicators and dynamic impact sequence data of the wastewater treatment station and inputs them into the wastewater treatment decision optimization model to obtain the current first decision data; The decision optimization module is used to obtain the control decision data in the most recent control window, and use the control decision data in the most recent control window to perform trend contraction adaptive adjustment on the current first decision data to obtain the final decision data.

Images