Intelligent diversion and resource utilization decision support system for rural domestic sewage

Abstract

The application discloses a rural domestic sewage intelligent diversion and resource utilization decision support system and particularly relates to the technical field of sewage treatment, comprising five modules of sensing collection, preliminary identification, data quantification, path prediction and path execution, the sensing collection module acquires sewage quality information to generate a structured data set, the preliminary identification module judges whether there is abnormal discharge and generates a quantitative signal, the data quantification module calculates resource potential index and treatment risk index, the path prediction module outputs predicted path values based on the indexes, and the path execution module matches the path values with diversion operation types and calls corresponding sewage treatment paths; the application constructs a closed-loop control mechanism from data to execution through structured collection of water quality data, an abnormal triggering quantification process and double-index driven path prediction, realizes dynamic and accurate scheduling of sewage treatment paths, and improves data utilization efficiency and the responsiveness and intelligence of treatment decisions.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of sewage treatment, more particularly, the present application relates to an intelligent diversion and resource utilization decision support system for rural domestic sewage. BACKGROUND

[0002] Under the background of the increasingly perfect urban and rural community governance system, efficient treatment and resource utilization of rural domestic sewage has become one of the key tasks for the development of ecological agriculture. However, unlike the centralized sewage treatment system in cities, there are obvious differences in spatial layout, infrastructure and operation and maintenance capacity in rural areas, which leads to a series of inherent problems in the actual promotion of domestic sewage treatment.

[0003] On the one hand, rural settlements are scattered, water use habits differ significantly, and sewage discharge intensity is intermittent and unstable. The sewage generated in daily life often includes kitchen drainage, washing wastewater and part of the mixed livestock and poultry breeding wastewater, and the water quality parameters such as chemical oxygen demand, ammonia nitrogen concentration and suspended solids content fluctuate dramatically at different times. The composition of pollutants in sewage is complex and the proportion is variable, which is easy to cause local high pollution incidents. Traditional rural sewage treatment technology generally adopts fixed treatment path (such as constructed wetland, oxidation pond or small integrated equipment), and the treatment strategy lacks real-time identification and dynamic adaptation ability to sewage state, often resulting in redundant treatment capacity or treatment failure.

[0004] On the other hand, current rural sewage treatment schemes generally focus on "treatment compliance" orientation, lack of technical means for resource value assessment of recyclable components in sewage, resulting in the loss of resourceable substances, and failing to include sewage as a renewable resource into the village resource recycling system. In addition, even if some areas introduce water quality monitoring devices, their main function still stays at the data collection level, and a complete "collection-identification-analysis-prediction-execution" technical closed loop cannot be built, resulting in inefficient use of data, automatic triggering of decision-making behavior, and still requiring manual intervention for path control. The level of intelligence is seriously insufficient, and there is a lack of scientific and efficient decision support mechanism. Therefore, the present application proposes an intelligent diversion and resource utilization decision support system for rural domestic sewage to solve the above problems. SUMMARY

[0005] To achieve the above purpose, the present application provides the following technical scheme:

[0006] The intelligent diversion and resource utilization decision support system for rural domestic sewage comprises a perception and collection module, a preliminary identification module, a data quantification module, a path prediction module and a path execution module, and their composition and cooperative relationship are as follows:

[0007] The perception acquisition module obtains real-time water quality information of the sewage discharge point, and generates a structured data set in time sequence according to the real-time water quality information;

[0008] The preliminary judgment identification module identifies the composition of the sewage and its change trend based on the structured data set, judges whether there is an abnormal discharge condition, and generates a quantitative signal when the judgment condition corresponding to the abnormal discharge condition is met;

[0009] The data quantification module calculates the resource potential index and the treatment risk index of the sewage according to the independently constructed quantification model when receiving the quantitative signal, the resource potential index is used to represent the degree of the sewage that can be converted into reusable substances, and the treatment risk index is used to represent the degree of the sewage that produces unstable state risk in the treatment process;

[0010] The path prediction module receives the resource potential index and the treatment risk index of the sewage as input, calls the trained prediction model, and outputs a prediction path value, which is used to express the preferred direction of the treatment path that the sewage should match at present;

[0011] The path execution module maps the corresponding shunt operation type according to the numerical interval of the prediction path value and calls the corresponding sewage treatment path.

[0012] In a preferred embodiment, the real-time water quality information obtained in the perception acquisition module is:

[0013] The multiple water quality parameter data obtained after the sewage is collected at the discharge point through the perception device include three indicators of chemical oxygen demand, ammonia nitrogen concentration and suspended matter content, each indicator is collected every ten minutes, and six sampling data points are formed for each indicator in a sampling period. Three indicators form eighteen time sequence data units in total, which are arranged in time sequence to form a structured data set, the structured data set is organized in a two-dimensional matrix manner, each row represents a time section of a sampling behavior, and each column corresponds to the value of a water quality parameter at the time point.

[0014] In a preferred embodiment, when the preliminary judgment identification module judges whether there is an abnormal discharge condition, the following calculation process is performed based on the water quality information in the structured data set:

[0015] In the current sampling period, all sampling data points of the three water quality parameters of chemical oxygen demand, ammonia nitrogen concentration and suspended matter content are extracted to form three parameter sequences respectively;

[0016] The standard deviations of the three parameter sequences are calculated respectively, denoted as σ1, σ2, σ3, and then normalized to obtain normalized parameters σ1', σ2', σ3', which form a vector V=(σ1', σ2', σ3').

[0017] The change intensity value L1 is obtained by the vector length calculation formula, and the change intensity value L1 is compared with the abnormality determination threshold L0, when L1 is greater than L0, it is determined that there is an abnormal discharge situation, and the initial judgment identification module outputs the quantization signal as the calling trigger basis of the data quantization module under the condition.

[0018] In a preferred embodiment, the data quantization module calculates the resource potential index based on the chemical oxygen demand, ammonia nitrogen concentration and total phosphorus content in the current sampling period as follows:

[0019] Six sampling data points of each parameter form a time series vector, respectively defined as vectors P1, P2 and P3;

[0020] A parameter vector set is constructed in a three-dimensional vector space, and the direction with the largest mean square deviation from the average vector is determined by comparing the mean square deviations of the three vectors, and is recorded as the principal direction vector D;

[0021] The cosine value of the angle between each parameter vector and the principal direction vector D is calculated as the parameter contribution coefficient;

[0022] The contribution coefficient is multiplied by the corresponding parameter mean value and normalized to form the resource potential index, which is output in real number form and recalculated after each data update.

[0023] In a preferred embodiment, the data quantization module calculates the processing risk index using the time series data of ammonia nitrogen concentration and suspended solids content in the current two consecutive sampling periods in the structured data set, and performs the following steps:

[0024] Twelve data points of two parameters in two consecutive sampling periods are extracted to form two independent one-dimensional time series, respectively represented as sequences A and B;

[0025] For each set of adjacent data points in sequences A and B, calculate the first-order difference to obtain two difference sequences with a length of eleven;

[0026] Determine whether the positive and negative change direction of each difference value is reversed, if there are two times of positive and negative change reversal in the continuous three points, mark it as a disturbance point;

[0027] The number of all disturbance points in the two difference sequences is counted and recorded as the disturbance frequency T;

[0028] The absolute value of each difference value in the two difference sequences is averaged as the average disturbance amplitude M; then the disturbance frequency T and the disturbance amplitude M are substituted into the formula FRI=ln(1+T×M) to obtain the processing risk index FRI.

[0029] In a preferred embodiment, the prediction model invoked in the path prediction module is a multivariate regression model constructed based on historical wastewater treatment data, the model training input contains the resource potential index and the treatment risk index, the model training output is the predicted path value, the predicted path value is an integer between 0 and 100, used to represent the most matched treatment path direction of the current wastewater, wherein the larger the value, the more priority to match the resource conversion path, and the smaller the value, the more priority to match the risk avoidance path, the path prediction module recalculates the predicted path value in real time after each new index is generated, maintaining the timeliness of decision-making.

[0030] In a preferred embodiment, the path execution module completes the mapping process when executing the corresponding shunt operation type according to the predicted path value by the following steps:

[0031] The predicted path value generated by the path prediction module is numerically matched with the preset shunt value interval, the shunt value interval is composed of four boundary values, which are sequentially divided into a first shunt interval, a second shunt interval and a third shunt interval, each interval corresponds to a unique shunt operation type;

[0032] When the predicted path value is less than the first boundary value, it automatically falls into the first shunt interval, and the corresponding shunt operation type is to guide to the risk avoidance path;

[0033] When the predicted path value is between the first boundary value and the second boundary value, it falls into the second shunt interval, and the corresponding shunt operation type is to guide to the neutral treatment path;

[0034] When the predicted path value is greater than the second boundary value, it falls into the third shunt interval, and the corresponding shunt operation type is to guide to the resource utilization path.

[0035] In a preferred embodiment, the path execution module includes a path selection unit, a path switching unit and a path execution unit, which constitute a treatment process in sequence and cooperatively complete the actual calling operation of the predicted path value to the corresponding treatment path;

[0036] The path selection unit receives the predicted path value output by the path prediction module, and compares it with the preset multiple shunt value intervals to determine the target shunt operation type corresponding to the current path prediction value;

[0037] The path switching unit retrieves the matched target path identifier from the registered treatment path set according to the shunt operation type output by the path selection unit, and generates a path switching instruction;

[0038] The path execution unit receives the path switching instruction, controls the sewage flow to the preset treatment path entrance according to the target path identifier contained in the path switching instruction, completes the physical channel conversion process, and generates a confirmation signal to return to the path switching unit to end the current path execution period after the path switching is completed.

[0039] Technical effects and advantages of the present application:

[0040] The present application structures the real-time water quality information of the sewage discharge point through the perception acquisition module, so that the original water quality data has time series characteristics that can be used for judgment and modeling, thereby providing a standardized input basis for subsequent identification, quantification and prediction processes, and enhancing the uniformity and processing continuity of data analysis. The formation of the structured data set breaks the limitations of fragmented, discontinuous and difficult-to-process water quality data in the past, so that the information can be stably analyzed and called in the entire system, and the change trend and historical state evolution law of the sewage can be reflected, thereby improving the reliability of sewage analysis and processing decision from the source.

[0041] The present application sets the preliminary judgment identification module to identify the components and analyze the trends of the structured data, and triggers the generation mechanism of the quantification signal when the discharge abnormal condition is identified, thereby realizing the pre-response to the abnormal change of water quality. By judging whether there is a discharge abnormal condition and issuing a quantification signal, the present application establishes a closed-loop linkage logic from "perception data change" to "triggering subsequent calculation process", ensuring that only when the key data characteristics change significantly, the next processing stage is entered, avoiding invalid calculation and resource waste, effectively improving the system processing efficiency and reducing data processing redundancy.

[0042] The present application realizes the intelligent dynamic selection mechanism of the sewage treatment path through the sequential cooperation of the data quantification module, the path prediction module and the path execution module, so that the treatment path can be dynamically adjusted according to the real-time evaluation result of the sewage state. The dual-input structure of the resource potential index and the treatment risk index ensures that the path prediction process takes into account the reuse value and the processing difficulty of the sewage; and the matching mechanism of the path execution module to the predicted path value and the preset shunt value interval ensures that the prediction result can be directly converted into specific shunt behavior, realizing seamless connection from data-driven judgment to physical path scheduling, and improving the scientificity and real-time of the treatment path selection. BRIEF DESCRIPTION OF DRAWINGS

[0043] In order to facilitate the understanding of those skilled in the art, the present application will be further described below with reference to the accompanying drawings;

[0044] Fig. 1 The principle diagram of the rural domestic sewage intelligent shunting and resource decision support system in the present application.

[0045] Fig. 2 The principle diagram of the path execution module in the present application. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0047] Reference Figs. 1-2 The following examples were obtained:

[0048] Example 1:

[0049] The intelligent diversion and resource utilization decision support system for rural domestic sewage includes a sensing and data acquisition module, a preliminary judgment and identification module, a data quantification module, a path prediction module, and a path execution module. Their composition and collaborative relationships are as follows:

[0050] The sensing and acquisition module acquires real-time water quality information from wastewater discharge points and generates a structured dataset based on this information in chronological order. Deployed at the front end of the domestic wastewater discharge point, the module collects representative and variable water quality parameters reflecting the wastewater's current state and pollution load. The data collection process is cyclical, ensuring data continuity and real-time performance. The generated structured dataset is arranged chronologically, forming a multi-dimensional and time-series data set that serves as the foundation for subsequent modules' dynamic analysis and judgment.

[0051] The initial identification module, based on a structured dataset, identifies the composition and changing trends of wastewater components, determines the presence of any abnormal discharges, and generates a quantitative signal when the corresponding criteria for abnormal discharges are met. The module reads the structured dataset from the sensing and acquisition module, calculates the numerical changes in wastewater components across multiple time points, extracts key trend information and rate of change parameters, and thus identifies the distribution and evolution patterns of major pollutants in the wastewater. This module uses a built-in algorithm to comprehensively evaluate the magnitude of component changes, fluctuation frequency, and abnormal gradients within the current detection period. When the detection results meet the system's predefined criteria for abnormal discharges, such as a sudden increase in components, a reversal in composition, or drastic concentration unevenness, the module immediately outputs a standardized quantitative signal as the basis for initiating subsequent analysis procedures.

[0052] Upon receiving a quantification signal, the data quantification module calculates the wastewater's resource potential index and treatment risk index based on independently constructed quantification models. The resource potential index represents the degree to which wastewater can be converted into reusable substances, while the treatment risk index represents the degree of instability risk during wastewater treatment. Using the quantification signal output by the initial identification module as an activation condition, the data quantification module initiates in-depth analysis of the pollutant indicators contained in the current structured dataset, and calls two non-interfering quantification models constructed based on different technical approaches. The calculation of the resource potential index focuses on assessing the quantity and conversion potential of organic matter, nutrients, and calorific value factors in wastewater that can be recycled, reused, or used for ecological compensation. The calculation of the treatment risk index focuses on assessing whether wastewater entering the treatment system may cause overload of the treatment unit, reaction instability, or effluent exceeding standards. The numerical outputs of both indices will serve as important variables for subsequent decision-making.

[0053] The path prediction module receives the wastewater's resource potential index and treatment risk index as input, calls the trained prediction model, and outputs a predicted path value. This predicted path value expresses the preferred treatment path direction for the current wastewater. After acquiring the two indices, the path prediction module inputs them as variables into the internally embedded prediction model. This model has been pre-trained on historical wastewater treatment data samples to establish a mapping mechanism between wastewater characteristics and treatment path matching. The predicted path value output by the model is a single numerical result, representing the current treatment demand tendency of wastewater between resource utilization and risk control. This value will fluctuate within a set range and has guiding significance, providing a quantitative basis for subsequent treatment path selection and execution.

[0054] The path execution module maps the predicted path value to the corresponding diversion value range based on the range it falls into and invokes the corresponding wastewater treatment path. The module parses the predicted path value generated by the prediction module, locating it within a pre-defined range of values ​​to a specific treatment area. Different ranges correspond to different diversion operation types, which are preset to instruction sets corresponding to the physical treatment paths. Once the diversion operation type is determined, the path execution module locates and activates the corresponding wastewater treatment process, completing the specific wastewater flow control. This achieves multi-path treatment guidance driven by automatic system decision-making, ensuring dynamic optimization of treatment effectiveness and resource utilization efficiency.

[0055] The real-time water quality information acquired by the sensing and acquisition module consists of multiple water quality parameter data collected by sensing devices at the wastewater discharge point. In this system, the "discharge point" refers to the location where domestic sewage is discharged from a household or small drainage unit, typically located at the kitchen drain, bathroom outlet, or inlet node of a centralized drainage network in rural residential areas. The "sensing device" refers to water quality detection and sensing equipment deployed at the discharge point for real-time monitoring of wastewater status; its operation involves quantitative detection and data transmission at fixed intervals. The collected "multiple water quality parameter data" refers to various water quality indicators that quantify the characteristics and concentration changes of pollutants in the wastewater.

[0056] In the current implementation, the sensing device collects three parameters: chemical oxygen demand (COD), ammonia nitrogen concentration, and suspended solids content. COD is used to characterize the content of oxidizable organic matter in wastewater and is a key indicator reflecting the degree of organic pollution. Ammonia nitrogen concentration is used to measure the concentration of nitrogen-containing pollutants in wastewater and has a strong correlation with treatment difficulty and biotoxicity. Suspended solids content reflects the density of non-sedimentary solid particles in wastewater and is an important parameter for assessing the difficulty of sedimentation and filtration.

[0057] During the data collection process, each indicator is sampled every ten minutes, resulting in six equally spaced sampling data points per hour, which constitutes a complete "sampling cycle." Within one sampling cycle, each indicator generates six sampling data points, resulting in a total of eighteen independent data points across the three indicators. These eighteen sampling data points are strictly synchronized in time and arranged according to the sampling order, forming a data set with time attributes, referred to as a "structured dataset." This structured dataset is organized and stored using a two-dimensional matrix, mapping the time dimension and parameter dimension to rows and columns of the matrix, respectively: each row represents a time segment of a sampling action, indicating the time point in time at which the sampling occurred; each column represents the value of the corresponding water quality parameter at that time point, indicating the attribute dimension of the sampled content. Therefore, this structure achieves a dual-ordered data structure of "time-parameter," which is beneficial for subsequent modules to analyze and calculate data trends, fluctuations, and characteristic distributions. It should be noted that although the current implementation only lists three indicators—chemical oxygen demand, ammonia nitrogen concentration, and suspended solids content—the system's structure is not limited to these three types of parameters. Depending on changes in wastewater treatment objectives or differences in application environments, the sensing and acquisition module can also be extended to detect other water quality indicators, such as total phosphorus concentration, total nitrogen content, conductivity, dissolved oxygen, pH value, temperature, and biological oxygen demand. These extended parameters can be individually or in combination integrated into the sensing device's acquisition process and added to a structured dataset through a unified data format and time window mechanism, enabling the system to flexibly adapt to the wastewater characteristic monitoring requirements of different regions and objectives.

[0058] When the initial identification module determines whether there is an abnormal discharge, it performs the following calculation process based on the water quality information in the structured dataset: Within the current sampling period, all sampling data points for each of the three water quality parameters—chemical oxygen demand (COD), ammonia nitrogen concentration, and suspended solids content—are extracted, forming three parameter sequences. The "sampling period" here refers to a period of six consecutive samplings conducted by the sensing and acquisition module at ten-minute intervals within one hour. The three indicators—"chemical oxygen demand," "ammonia nitrogen concentration," and "suspended solids content"—have been defined in the sensing and acquisition module section and are the three basic parameters constituting the structured dataset in the current implementation plan. Each parameter forms six data points within one sampling period; therefore, the three extracted parameter sequences each contain six real values ​​arranged in chronological order, used to characterize the temporal variation characteristics of the corresponding parameter within that sampling period. The standard deviations of the three parameter sequences are calculated and denoted as σ1, σ2, and σ3.

[0059] The standard deviation is calculated as follows: for each parameter sequence, the squared difference between the six data points and the mean of the sequence is calculated, then the sums are taken and the square root is taken. The result is the standard deviation of the sequence, representing the degree of numerical fluctuation of the parameter within the sampling period. The larger the standard deviation, the more drastic the change of the parameter between different sampling time points, reflecting higher instantaneous instability. Normalization is then performed to obtain normalized parameters σ1′, σ2′, and σ3′. Normalization refers to mapping the original standard deviation values ​​to a unified dimensional range through a linear scaling function, facilitating subsequent joint analysis between different parameters. In the current implementation, the normalization method is min-max normalization, which linearly maps the standard deviation values ​​to the real number range between 0 and 1 according to set upper and lower limits. The three components form a vector V = (σ1′, σ2′, σ3′). This vector is a three-dimensional vector, where each component represents the relative intensity of fluctuation in the chemical oxygen demand, ammonia nitrogen concentration, and suspended solids content within the current sampling period.

[0060] The change intensity value L1 is obtained through the vector magnitude calculation formula. The formula involves squaring the three normalized standard deviations, adding them together, and then taking the square root to obtain a non-negative real number. The change intensity value L1 is used to comprehensively measure the overall volatility of the three parameters within the sampling period, serving as a quantitative indicator of the degree of change in emission status. This change intensity value L1 is then compared with the anomaly judgment threshold L0. The threshold L0 is a preset static constant in the system, used to determine whether an abnormal state exists. Its value is determined by historical data fitting and risk assessment parameters and does not change with real-time data. In the current implementation, the default value of L0 is 1.5, representing the upper limit of change intensity based on system experience. When L1 is greater than L0, an abnormal emission situation is determined, indicating that the wastewater exhibits abnormal fluctuations in multiple parameters simultaneously within the sampling period, which may lead to mismatches or overload of subsequent treatment paths. When this condition is met, the initial judgment identification module outputs a quantization signal as the trigger for calling the data quantization module. The quantization signal is a logic control signal used to activate the subsequent calculation process of the data quantization module, ensuring that the system only starts the analysis process of the resource potential index and the processing risk index when an emission anomaly is confirmed, thereby realizing the on-demand call mechanism of system resources and response behavior.

[0061] When calculating the resource potential index, the data quantification module performs the following steps based on three parameters acquired within the current sampling period: chemical oxygen demand (COD), ammonia nitrogen concentration, and total phosphorus content. COD and ammonia nitrogen concentration are derived from a structured dataset, while total phosphorus content is integrated into the system analysis process as an expandable acquisition parameter. In system deployment, total phosphorus content can be acquired through the parameter extension interface of the sensing acquisition module and combined with other parameters in the form of additional columns or parallel structures when generating the structured dataset, ensuring temporal synchronization with other parameters and serving as a complete input for the resource potential index calculation in this step. The six sampling data points for each parameter are used to construct a time series vector, defined as vectors P1, P2, and P3. The "sampling period" refers to the time interval formed by the sensing acquisition module sampling water quality six times at ten-minute intervals within a continuous one-hour period. COD and ammonia nitrogen concentration are already introduced as default system parameters; "total phosphorus content" is a newly added water quality parameter in this step, used to characterize the concentration of phosphorus nutrients in wastewater and is an important indicator for resource reuse. The time series vector formed by each parameter consists of the measured values ​​of that parameter at six time points, forming three numerical vectors with time order, denoted as P1, P2, and P3, corresponding to the three indicators mentioned above. For example, if the six sampled values ​​of chemical oxygen demand in a certain sampling period are [31, 30, 32, 34, 33, 35], then P1 = (31, 30, 32, 34, 33, 35).

[0062] A set of parameter vectors is constructed in a three-dimensional vector space. By comparing the root mean square deviations of the three vectors from the average vector, the direction with the largest root mean square deviation is determined and denoted as the principal direction vector D. P1, P2, and P3 are considered as three-dimensional parameter vectors in the same coordinate system, with each dimension representing the value at the corresponding time point in its time series. The deviation of each of these three vectors from the average vector μ is analyzed. The average vector μ is the average vector of the values ​​of P1, P2, and P3 at each time point. The degree of deviation is calculated using the root mean square deviation, i.e., by normalizing and averaging the sum of the squares of the differences between each vector and μ, thus determining which direction among the three vectors shows the greatest change. The corresponding vector direction is defined as the principal direction vector D. The principal direction vector D reflects the dominant trend direction of wastewater characteristic changes within the sampling period. The cosine of the angle between each parameter vector and the principal direction vector D is calculated as the parameter contribution coefficient. The cosine of the angle is defined as the dot product of two vectors divided by the product of their magnitudes; this value is between -1 and 1, with a value closer to 1 indicating a more consistent direction between the two vectors. Therefore, the cosine values ​​of the angles between P1, P2, P3 and D are calculated and denoted as C1, C2, and C3, respectively, representing the contribution of each parameter to the current dominant trend. This calculation process does not change the original parameter values, but only uses them as weighting criteria in subsequent calculations. The contribution coefficients are multiplied by the corresponding parameter mean, summed, and then normalized to form the resource potential index. This index is output in real number form and recalculated after each round of data updates. That is: Resource Potential Index RPI = Normalize(C1×mean(P1)+C2×mean(P2)+C3×mean(P3)), where mean(Pn) is the arithmetic mean of the six data points in vector Pn, and Normalize represents the normalization function. The normalization function is based on existing technology and will not be elaborated further here. The calculation results are mapped to the real number range between 0 and 10 to unify the dimensions of the results in each round. This index is used to comprehensively characterize the value potential of wastewater in resource conversion within the current sampling period. Since the index is recalculated after each new sampling period, it can dynamically reflect the status of wastewater resource availability over time.

[0063] When calculating the risk index, the data quantification module uses time-series data of ammonia nitrogen concentration and suspended solids content from the structured dataset within the current two consecutive sampling periods, and performs the following steps: It extracts twelve data points for each parameter from the two consecutive sampling periods, forming two independent one-dimensional time series, denoted as Sequence A and Sequence B, respectively. Here, "sampling period" refers to the periodic unit formed by the sensing and acquisition module completing six water quality samples within one hour at a fixed time interval of ten minutes. Two consecutive sampling periods represent two adjacent one-hour periods, covering a total time interval of two hours. "Ammonia nitrogen concentration" and "suspended solids content" are core water quality parameters defined in the structured dataset, used to measure the concentration of soluble nitrogen pollutants and the content of non-dissolved solid particulate matter in wastewater, respectively. Each parameter has six data points in each period; therefore, a total of twelve sampling values ​​are extracted from the two periods, forming one-dimensional time series A and Sequence B, used to express the continuous numerical change trend of the parameter within two hours.

[0064] For each pair of adjacent data points in sequence A and sequence B, calculate their first-order difference to obtain two difference sequences of length eleven. The "first-order difference" refers to the difference between two sampled values ​​at each adjacent position in the sequence, calculated as the latter value minus the former. Specifically, sequence A has twelve data points, denoted as A1 to A12, so the first-order difference results are A2-A1, A3-A2, ..., A12-A11, totaling eleven differences, denoted as the difference sequence ΔA. Similarly, sequence B generates the difference sequence ΔB. Determine whether the positive and negative directions of each difference value have reversed. If there are two reversals of positive and negative changes in three consecutive points, it is marked as a disturbance point. Analyze each difference sequence to examine its sign change behavior. If there are three consecutive differences Δn-1, Δn, and Δn+1, such that Δn-1 and Δn have opposite signs, and Δn+1 has an opposite sign to Δn, forming a "repeated positive and negative" fluctuation structure, then this position is considered to have experienced a disturbance. Each occurrence of this inversion structure is counted as a perturbation point. The total number of perturbation points in the two difference sequences is counted and denoted as the perturbation number T.

[0065] The positions in sequences ΔA and ΔB that satisfy the above definition of disturbance points are counted, and the sum of the results is the disturbance frequency T, representing the instantaneous fluctuation frequency of the two parameters during the monitoring period, indirectly reflecting the degree of instability of wastewater quality. The average of the absolute values ​​of each difference in the two difference sequences is taken as the average disturbance amplitude M. That is, after taking the absolute values ​​of all differences in the difference sequences ΔA and ΔB, the sum of the two is taken, and then divided by the total number of difference data (22) to obtain the average value M, which represents the average level of the fluctuation intensity of the two parameters in the current period. Then, the disturbance frequency T and the disturbance amplitude M are substituted into the formula FRI=ln(1+T×M) to obtain the treatment risk index FRI. In this formula, "ln" represents the natural logarithm function, and T×M constitutes a product term used to simultaneously measure the frequency and amplitude of the disturbance. The overall function structure is exponentially growing, which is more sensitive to high-frequency, high-amplitude disturbances. The calculated result FRI is in real number form, and its range is limited to the non-negative interval. The larger the value, the higher the treatment risk level under the current water quality condition. The risk index, as an important evaluation indicator of the system's stability in the current cycle, is directly used by the path prediction module.

[0066] The prediction model used in the path prediction module is a multivariate regression model built based on historical wastewater treatment data. The "path prediction module" refers to the analysis component in the system responsible for intelligently deciding the processing path direction based on the current water quality status; its core function relies on the output of the prediction model. The "prediction model" is a mathematical model established during the initial deployment or operation of the system by collecting, organizing, and training a large number of historical wastewater treatment behavior samples. This model uses regression modeling techniques to establish a numerical mapping relationship between multiple input variables and a single output result, used for trend prediction and numerical output of future data.

[0067] A "multivariate regression model" refers to a model that relies not only on a single input but also on multiple related variables as independent variables to fit the dependent variable. Compared to linear or univariate models, multivariate regression models can simultaneously consider the interactive effects of multiple factors, improving the accuracy of path prediction. The model's training input includes a resource potential index and a treatment risk index. The resource potential index and treatment risk index are calculated by the data quantification module based on real-time water quality information within the current sampling period. The former reflects the resource reuse value of wastewater, while the latter assesses the stability risk of the treatment system. During model training, the system uses data samples from past treatment periods that simultaneously contain the resource potential index, treatment risk index, and final treatment path selection results as training data. The two indices are input as independent variables, and the treatment path direction is output as the model's label. Through model training, a numerical mapping relationship between the three is established.

[0068] A multivariate regression model is a mathematical model that establishes the quantitative relationship between multiple input variables and one output variable. This model fits historical data to determine a set of fixed coefficients, allowing each input variable to influence the final prediction to varying degrees. The basic structure of the model can be represented as follows: The model receives two input variables: a resource potential index and a treatment risk index, denoted as the first and second input variables, respectively. These two input variables are assigned two corresponding coefficients, the first and second coefficients, representing their strength of influence on the predicted path value. During model calculation, the first input variable is first multiplied by the first coefficient to obtain the first part of the result; then the second input variable is multiplied by the second coefficient to obtain the second part of the result. These two parts are then summed to form an intermediate predicted value. Based on this summation, the model introduces a fixed offset term called the intercept term. This intercept term adjusts the base position of the predicted value within the overall numerical range, ensuring the model output has translational adjustment capabilities. Finally, the predicted path value is the sum of the three terms—the first part of the result, the second part of the result, and the intercept term. The sum of these values ​​represents the predicted path output by the model, and is limited to integers between zero and one hundred. To maintain an integer output, if the sum is a decimal, the integer closest to that decimal is taken as the final result. This regression model obtains the value of each coefficient through sample training. The training process is based on the known resource potential index, processing risk index, and corresponding optimal processing path direction from historical sampling periods. It automatically adjusts each coefficient using the minimum error principle to ensure that the model's predictions closely approximate actual path selection behavior.

[0069] The model training output is a predicted path value, which is an integer between 0 and 100. The "predicted path value" is the direct output of the model, used to quantify the degree of bias of the current wastewater state towards different treatment paths. The model output is an integer value, limited to the range of 0 to 100. This means the system distributes all treatment path directions within this numerical range at the quantification level, and uses this value as the basis for the path execution module to perform diversion mapping. The value represents the most suitable treatment path direction for the current wastewater, where a larger value indicates a priority for matching resource conversion paths, and a smaller value indicates a priority for matching risk avoidance paths. "Resource conversion paths" refer to treatment paths in the system that primarily aim to achieve the reuse of materials, energy, or water in wastewater, such as reclaimed water reuse, biogas recovery, or agricultural irrigation. "Risk avoidance paths" refer to path types that prioritize deep treatment or closed treatment processes to reduce system anomalies, aiming to ensure treatment safety and compliance with discharge standards, such as physicochemical sedimentation, high-intensity aeration, or emergency storage and discharge processes. The magnitude of the predicted path value forms the basis for path classification: the larger the value, the more it is biased towards resource reuse, and the smaller the value, the more it tends towards a steady-state emission strategy.

[0070] After each new index is generated, the path prediction module recalculates the predicted path value in real time to maintain the timeliness of decision-making. Real-time recalculation means that whenever the system completes the calculation of the resource potential index and the processing risk index, it immediately uses them as model inputs for path value prediction, so that the processing path decision can be highly synchronized with the changes in the sewage state. The timeliness of decision-making means that the system can respond in a timely manner according to the rapid fluctuation characteristics of the sewage quality and output the updated optimal path preference, avoiding path selection errors or system load imbalance caused by processing lag. For example, when the system finds that the resource potential index decreases and the processing risk index increases in two consecutive cycles, the predicted path value of the model will automatically approach the lower limit, and the path prediction module will then switch to the path direction biased towards risk control, ultimately affecting the operation behavior of the diversion and execution module.

[0071] When the path execution module performs the corresponding diversion operation type according to the predicted path value, it completes the mapping process through the following steps: matching the predicted path value generated by the path prediction module with the preset diversion numerical interval. The predicted path value is an integer value output by the path prediction module, which is used to express the degree of bias of the sewage treatment path, and its numerical range is between 0 and 100. To make a classification judgment on this value on the processing path, the system sets four boundary values during initialization, denoted as F1, F2, F3, and F4, satisfying F1 < F2 < F3 < F4. Among them, F1 is the starting value (fixed at 0), F4 is the maximum boundary value (fixed at 100), and F2 and F3 are adjustable intermediate demarcation values. According to the above four boundary values, the system divides the entire numerical interval into three continuous and non-overlapping diversion intervals: the first diversion interval is from F1 to F2 (excluding F2), the second diversion interval is from F2 to F3 (including F2, excluding F3), and the third diversion interval is from F3 to F4 (including F3).

[0072] Each diversion interval corresponds to a unique diversion operation type. The system determines the specific treatment direction based on the interval the predicted path value falls into. When the predicted path value is less than F2, it automatically falls into the first diversion interval, and the corresponding diversion operation type is guiding towards the risk-avoidance path. The risk-avoidance path prioritizes system operational safety and treatment compliance, suitable for situations with large fluctuations in water quality parameters and high risk indices. Typical processes include high-intensity oxidation, physicochemical treatment, or temporary storage control devices. When the predicted path value is between F2 and F3, it falls into the second diversion interval, and the corresponding diversion operation type is guiding towards the neutral treatment path. The neutral treatment path represents wastewater with both resource conversion potential and risk in the middle range, suitable for cost-effective standard treatment solutions, such as conventional biological treatment tanks, constructed wetlands, or oxidation ditch systems. When the predicted path value is greater than or equal to F3, it falls into the third diversion interval, and the corresponding diversion operation type is guiding resources towards the resource utilization path. The path utilization method is suitable for scenarios with low risk and a high proportion of convertible components in wastewater, such as path structures suitable for anaerobic fermentation, agricultural irrigation, wastewater reuse, or biomass extraction. This diversion mapping process is completed by the path selection unit in the path execution module. Through numerical comparison, it determines the location assignment of the predicted path value between F1 and F4, and then passes this assignment result to the path switching unit and the path execution unit for processing, ultimately achieving the wastewater flow direction control operation.

[0073] The path execution module comprises a path selection unit, a path switching unit, and a path execution unit. These three units sequentially constitute the processing flow and work together to complete the actual invocation of the processing path corresponding to the predicted path value. The "path execution module" is a collection of functional units responsible for performing traffic diversion operations based on the predicted path value and ultimately switching the processing path. It includes three sub-units, each undertaking a specific function and invoked sequentially to form a complete path control link. The path selection unit serves as the first-level entry logic, the path switching unit is responsible for generating intermediate decision instructions, and the path execution unit ultimately completes the physical channel operation. These three units maintain a strict information flow and control loop to ensure the responsiveness and consistency of the entire traffic diversion control behavior. The path selection unit receives the predicted path value output by the path prediction module and compares it with multiple preset traffic diversion value ranges to determine the target traffic diversion operation type corresponding to the current predicted path value. The "predicted path value" is a decision value generated in real time by the path prediction module based on the resource potential index and the processing risk index, with a value range limited to 0 to 100, as detailed above. The system internally pre-defines three diversion value intervals, consisting of four ordered boundary values ​​(e.g., F1 to F4), to map predicted path values ​​to specific diversion types. Upon receiving the predicted path value, the path selection unit immediately performs an interval judgment operation and converts the judgment result into a standardized diversion operation type, such as "guide to a risk avoidance path," "guide to a neutral processing path," or "guide to a resource utilization path." This operation type is transmitted to the path switching unit in the form of a logic code or control command.

[0074] The path switching unit retrieves the matching target path identifier from the registered treatment path set based on the diversion operation type output by the path selection unit, and generates a path switching instruction. The "treatment path set" consists of multiple pre-configured wastewater treatment channels in the system, each registered and coded according to its treatment objectives and technical characteristics, and mapped to corresponding operation types. For example, the treatment path set may include multiple options such as conventional biological treatment tanks, anaerobic reactors, high-load sedimentation tanks, and storage bypasses. Upon receiving the operation type, the path switching unit quickly searches the pre-established path dictionary table, identifies the treatment path entry matching the current operation type, and generates a control instruction structure, which is then transmitted to the path execution unit as a "path switching instruction." The path switching instruction explicitly includes the target path identifier, instruction triggering sequence, channel number, and optional redundant path configuration parameters.

[0075] Upon receiving a path switching command, the path execution unit controls the wastewater flow towards the preset treatment path inlet based on the target path identifier contained within, completing the physical channel conversion process. After the path switching is complete, it generates a confirmation signal and returns it to the path switching unit to end the current path execution cycle. The path execution unit interfaces with the actual wastewater treatment pipeline system and controls the fluid path through electric valves, level control devices, or programmable logic controllers (PLCs). This unit decodes the target path identifier provided in the path switching command, identifies its corresponding physical inlet location, and sends a drive signal to control the wastewater to flow along the selected path into the designated treatment section.

[0076] After the control action is completed and the valve position is confirmed to be closed, the path execution unit generates an acknowledgment signal and returns it to the path switching unit, notifying it that the current operation is complete and marking the end of this round of path scheduling. The acknowledgment signal can serve as the basis for system status recording, supporting subsequent path optimization algorithms or archived abnormal switching records. For example, in a certain operation, when the predicted path value is 82, the path selection unit determines that this value falls into the third diversion interval, corresponding to "guided to resource utilization path". The path switching unit retrieves "path R3" as the resource utilization channel, and the path execution unit controls the electric switching device to introduce sewage into the R3 inlet until the acknowledgment process is completed.

[0077] The above model formulas are all dimensionless and numerical calculations. The model formulas are derived from software simulations using a large amount of collected data to obtain the most recent real-world model formulas. The preset parameters in the model formulas are set by those skilled in the art according to the actual situation.

[0078] It should be understood that, in the various embodiments of this application, the order of the above-mentioned processes does not imply the order of execution. The execution order of each process should be determined by its module and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0079] 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 in electronic hardware, or a combination of computer software and electronic hardware. Whether these modules are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described modules for each specific application, but such implementation should not be considered beyond the scope of this application.

[0080] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0081] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can 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. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. An intelligent diversion and resource utilization decision support system for rural domestic sewage, characterized in that, The system comprises a perception acquisition module, a preliminary judgment identification module, a data quantification module, a path prediction module and a path execution module, and their structures and cooperative relationships are as follows: The perception acquisition module acquires real-time water quality information of the sewage discharge point, and generates a structured data set in time sequence according to the real-time water quality information; The preliminary judgment identification module identifies the composition of the sewage and its change trend based on the structured data set, judges whether there is an abnormal discharge condition, and generates a quantization signal when the judgment condition corresponding to the abnormal discharge condition is met; The data quantification module calculates the resource potential index and the processing risk index of the sewage according to the independently constructed quantization model when receiving the quantization signal, the resource potential index is used to represent the degree of the sewage that can be converted into reusable substances, and the processing risk index is used to represent the degree of the risk of unstable state of the sewage in the processing process; The path prediction module receives the resource potential index and the processing risk index of the sewage as input, calls the trained prediction model, and outputs a prediction path value, which is used to express the priority direction of the processing path that the sewage should match at present; The path execution module maps the corresponding shunt operation type according to the interval of the prediction path value, and calls the corresponding sewage treatment path; When calculating the resource potential index, the data quantification module executes the following steps based on the chemical oxygen demand, ammonia nitrogen concentration and total phosphorus content in the current sampling period: Six sampling data points of each parameter form a time series vector, respectively defined as vectors P1, P2 and P3; A parameter vector set is constructed in a three-dimensional vector space, and the direction with the largest mean square deviation from the average vector is determined by comparing the mean square deviations of the three vectors, which is recorded as the main direction vector D; The cosine value of the angle between each parameter vector and the main direction vector D is calculated as the parameter contribution coefficient; The contribution coefficient is multiplied by the corresponding parameter mean value, and then normalized to form the resource potential index; When calculating the processing risk index, the data quantification module uses the time series data of the ammonia nitrogen concentration and the suspended solids content in the current two consecutive sampling periods in the structured data set, and executes the following steps: Twelve data points of two parameters in two consecutive sampling periods are extracted to form two independent one-dimensional time series, respectively represented as sequence A and sequence B; For each group of adjacent data points in sequence A and sequence B, the first-order difference value is calculated to obtain two difference sequences with a length of eleven; Determine whether the positive and negative change directions of each difference value are reversed, if there are two times of positive and negative change reversal in the continuous three points, mark it as a disturbance point; The number of all disturbance points in the two difference sequences is counted, and is recorded as the disturbance frequency T; The absolute values of each difference value in the two difference sequences are averaged as the average disturbance amplitude M; Then, the disturbance frequency T and the disturbance amplitude M are substituted into the formula FRI = ln(1 + T × M) to obtain the processing risk index FRI.

2. The rural domestic sewage intelligent diversion and resource utilization decision support system according to claim 1, characterized in that, The real-time water quality information acquired in the perception acquisition module is: The multiple water quality parameter data obtained after sewage is collected by a sensing device at a discharge point include three indexes of chemical oxygen demand, ammonia nitrogen concentration and suspended matter content, each index is collected every ten minutes, six sampling data points of each index are formed in a sampling period, and eighteen time sequence data units are formed by the three indexes, the eighteen data units are arranged in time sequence to form a structured data set, the structured data set is organized in a two-dimensional matrix manner, each row represents a time section of a sampling behavior, and each column corresponds to a value of a water quality parameter at the time point. 3.The rural domestic sewage intelligent diversion and resource decision support system according to claim 2, characterized in that, When the preliminary judgment identification module judges whether there is an abnormal discharge condition, the following calculation process is performed based on the water quality information in the structured data set: In the current sampling period, all sampling data points of the three water quality parameters of chemical oxygen demand, ammonia nitrogen concentration and suspended matter content are extracted to form three parameter sequences respectively; The standard deviations of the three parameter sequences are calculated and denoted as σ1, σ2 and σ3 respectively, and then normalized to obtain normalized parameters σ1', σ2' and σ3', which form a vector V=(σ1', σ2', σ3'); The change intensity value L1 is obtained by the vector length calculation formula, and the change intensity value L1 is compared with the abnormal judgment threshold L0, when L1 is greater than L0, it is judged that there is an abnormal discharge condition, and the preliminary judgment identification module outputs a quantitative signal as a calling trigger basis for the data quantization module when the condition is met.

4. The rural domestic sewage intelligent diversion and resource utilization decision support system according to claim 1, characterized in that, The prediction model called in the path prediction module is a multivariate regression model constructed based on historical sewage treatment data, the model training input includes resource potential index and treatment risk index, and the model training output is a predicted path value, which represents the most matched treatment path direction of the current sewage.

5. The rural domestic sewage intelligent diversion and resource utilization decision support system according to claim 4, characterized in that, When the path execution module executes the corresponding shunt operation type according to the predicted path value, the mapping process is completed through the following steps: The predicted path value generated by the path prediction module is numerically matched with the preset shunt value interval, the shunt value interval is composed of four boundary values, which are sequentially divided into a first shunt interval, a second shunt interval and a third shunt interval, each interval corresponds to a unique shunt operation type; When the predicted path value is less than the first boundary value, it automatically falls into the first shunt interval, and the corresponding shunt operation type is to guide to the risk avoidance path; When the predicted path value is between the first boundary value and the second boundary value, it falls into the second shunt interval, and the corresponding shunt operation type is to guide to the neutral treatment path; When the predicted path value is greater than the second boundary value, it falls into the third shunt interval, and the corresponding shunt operation type is to guide to the resource utilization path.

6. The rural domestic sewage intelligent diversion and resource utilization decision support system according to claim 5, characterized in that, The path execution module includes a path selection unit, a path switching unit and a path execution unit, which constitute a treatment process in sequence and cooperatively complete the actual calling operation of the treatment path corresponding to the predicted path value; The path selection unit receives the predicted path value output by the path prediction module and compares it with the preset multiple shunt value intervals to determine the target shunt operation type corresponding to the current path prediction value; The path switching unit retrieves a matched target path identifier from the registered set of processing paths according to the shunting operation type output by the path selection unit, and generates a path switching instruction; After receiving the path switching instruction, the path execution unit controls the sewage to flow into the preset processing path entrance according to the target path identifier contained in the path switching instruction, completes the physical channel conversion process, and generates a confirmation signal to return to the path switching unit to end the current path execution period.

Images