Big data-based septic tank simulation design efficiency evaluation method and system

Abstract

The present application belongs to the technical field of septic tank design, and discloses a septic tank simulation design performance evaluation method and system based on big data. The method comprises the following steps: establishing digital simulation models of each septic tank to be evaluated, and setting corresponding design parameters based on predetermined design schemes; simulating and running the digital simulation models to obtain key performance parameters; calculating the performance evaluation coefficients of each septic tank design scheme to determine whether the design scheme is qualified; constructing a correlation matrix and a node performance characteristic matrix set, inputting a pre-constructed neural network model to obtain a global performance prediction set; and selecting the optimal design scheme of the septic tank to be evaluated based on the global performance prediction set. The present application comprehensively covers various working condition characteristics of the actual operation of the septic tank from three dimensions of steady-state operation performance, variable working condition adaptability and long-term operation attenuation characteristics to evaluate each design scheme, which can significantly improve the evaluation accuracy.

Description

Technical Field

[0001] This invention relates to the field of septic tank design technology, and more specifically, to a method and system for evaluating the performance of septic tank simulation design based on big data. Background Technology

[0002] As a core facility in decentralized wastewater treatment systems, the rationality of septic tank design directly determines the wastewater pretreatment effect, pipeline operation stability, and long-term operation and maintenance costs. Currently, the design evaluation of septic tanks in rural and urban areas mostly adopts empirical formula verification or single-tank static operating condition simulation. However, with the increasingly complex characteristics of wastewater discharge (time-based fluctuations in water use, seasonal changes in water quality and quantity) and the increasing systematization of regional pipeline networks, this approach is becoming less effective.

[0003] Traditional septic tank designs often use a single steady-state condition with rated influent load as the benchmark to determine the rationality of the design. This approach fails to adequately consider the impact of fluctuating residential water usage on operating conditions, as well as the efficiency degradation caused by sludge deposition and bacterial aging during long-term operation. This can easily lead to a situation where the design meets the standards in simulation but experiences a sharp drop in efficiency during actual operation. Furthermore, existing technologies generally adopt an approach of designing and evaluating each septic tank independently, without considering all septic tanks in the target area as coupled and interconnected system nodes, or modeling the network topology dependencies and sewage hydraulic linkages between nodes. Since fluctuations in effluent quality and flow rate from upstream septic tanks are directly transmitted to downstream nodes through the network, optimizing only the efficiency of a single tank can easily lead to hydraulic incompatibility between nodes (such as pipe diameter mismatch or load mismatch), causing problems such as network blockage and a cascading decline in downstream treatment efficiency, making it difficult to achieve global collaborative optimization of the target area.

[0004] In view of this, the present invention proposes a method and system for evaluating the design efficiency of septic tanks based on big data simulation in order to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art and achieve the above objectives, the present invention provides the following technical solution: A big data-based method for evaluating the design effectiveness of septic tanks through simulation, including the following methods: A corresponding digital simulation model is established for each septic tank to be evaluated within the target area, and corresponding design parameters are set for each septic tank to be evaluated in the digital simulation model based on the predetermined design scheme. Simulate different steady-state, variable and long-term operating conditions, perform CFD and biochemical reaction coupled simulation on the digital simulation model based on design parameters, and obtain key performance parameters of each septic tank to be evaluated in real time during the simulation operation. Obtain key performance benchmark information, combine key performance parameters to generate performance evaluation coefficients for each septic tank to be evaluated under each corresponding design scheme, evaluate the qualification of each design scheme based on the performance evaluation coefficients, and construct a qualified design scheme pool for each septic tank to be evaluated based on all qualified design schemes. Each septic tank to be evaluated within the target area is taken as a node. The pipe network topology correlation parameters and sewage hydraulic coupling parameters between any two nodes are calculated and fused to obtain the edge weights. The correlation matrix is ​​constructed based on the edge weights. Traverse the pool of qualified solutions for each node, select effective solution combinations, and generate node performance feature matrices under different effective solution combinations. Construct a set of node performance feature matrices based on the node performance feature matrices. The correlation matrix and the set of node performance feature matrices are input into a pre-built neural network model. The neural network model outputs a global performance prediction set of different effective scheme combinations. Based on the global performance prediction set, the performance of all effective scheme combinations is evaluated, and the optimal design scheme for each septic tank to be evaluated is selected based on the evaluation results.

[0006] Furthermore, the method for evaluating the suitability of a design scheme based on performance evaluation coefficients is as follows: Under multiple steady-state operating conditions, the steady-state performance deviation coefficient of the septic tank to be evaluated is calculated according to different simulated influent loads. Under varying influent load conditions, the varying operating condition adaptability deviation coefficient of the septic tank to be evaluated is calculated. Under long-term operating conditions, the long-term degradation deviation coefficient of the septic tank to be evaluated is calculated. Weighting coefficients are set for the steady-state performance deviation coefficient, the varying operating condition adaptability deviation coefficient, and the long-term degradation deviation coefficient. The steady-state performance deviation coefficient, the varying operating condition adaptability deviation coefficient, and the long-term degradation deviation coefficient are multiplied by their respective weighting coefficients and then summed to obtain the comprehensive deviation coefficient. A comprehensive deviation coefficient threshold and a benchmark confidence level are preset. The comprehensive deviation coefficient is divided by the product of the comprehensive deviation coefficient threshold and the benchmark confidence level to obtain the performance evaluation coefficient. If the performance evaluation coefficient is less than 1, the evaluation design scheme is qualified; otherwise, the evaluation design scheme is unqualified.

[0007] Furthermore, the calculation methods for the steady-state performance deviation coefficient and the variable operating condition adaptability deviation coefficient are as follows: Based on the different influent loads under the simulated m steady-state conditions, the detection cycle for each steady-state condition is set. Based on the changes in the key performance parameters of the septic tank to be evaluated over time under each steady-state condition within the detection cycle, the time variation function of each key performance parameter is proposed. Based on key performance benchmark information, a standard time variation function is constructed for the time variation function of each key performance parameter. Within the detection period, the absolute value of the first integral between the time variation function of the key performance parameter and the corresponding standard time variation function is calculated. The weighted summation of the absolute values ​​of the first integral of all key performance parameters yields the performance deviation of each steady-state condition. The weighted summation of the performance deviations of all steady-state conditions yields the steady-state performance deviation coefficient. Set up a variable operating condition with the influent load from minimum to maximum value, and formulate the load variation function of each key performance parameter based on the changes of each key performance parameter of the septic tank to be evaluated with the influent load. Based on key performance benchmark information, a standard load change function is constructed for the load change function of each key performance parameter. Under varying operating conditions, the second integral absolute value of the load change function of the key performance parameter and the corresponding standard load change function is calculated. The second integral absolute values ​​of all key performance parameters are weighted and summed to obtain the variation deviation coefficient of the varying operating conditions.

[0008] Furthermore, the long-term attenuation deviation coefficient is calculated as follows: Under long-term operating conditions, n consecutive normal operating cycles are set, and the key performance parameters of the septic tank to be evaluated are obtained in each operating cycle. The average performance parameter of each key performance parameter in each operating cycle is calculated. A benchmark value of the average performance parameter is set for each average performance parameter in each operating cycle. The absolute value of the difference between the average performance parameter and the corresponding benchmark value of the average performance parameter in each operating cycle is calculated. The absolute values ​​of the differences of the key performance parameters in all operating cycles are weighted and summed to obtain the attenuation coefficient of each key performance parameter. The attenuation coefficients of all key performance parameters are weighted and summed to obtain the long-term attenuation deviation coefficient.

[0009] Furthermore, the method for obtaining the baseline confidence level is as follows: Extract the number of valid samples and the total number of samples from the key performance benchmark information, and divide the number of valid samples by the total number of samples to obtain the data sufficiency sub-coefficient. The acquisition time of each valid sample is obtained, the average acquisition time of the valid samples is calculated, the current time of the simulation is recorded as the current evaluation time, the interval between the average acquisition time and the current evaluation time is obtained, a time decay coefficient is introduced, and the data timeliness coefficient is calculated through a preset exponential decay model. The operating condition parameters corresponding to each effective sample are extracted from the key performance benchmark information. Each operating condition parameter is pre-divided into different levels to form a gridded target operating condition combination. The number of effective samples in each target operating condition combination is counted. Based on the number of effective samples corresponding to each target operating condition combination, the corresponding sample proportion probability is calculated. The Shannon information entropy formula is introduced to calculate the actual operating condition distribution entropy value. The theoretical maximum information entropy corresponding to all target operating condition combinations is calculated. The ratio of the actual operating condition distribution entropy value to the theoretical maximum information entropy is used as the coverage integrity sub-coefficient. The baseline confidence level is obtained by multiplying the data sufficiency sub-coefficient, the data timeliness sub-coefficient, and the coverage completeness sub-coefficient together and then taking the cube root.

[0010] Furthermore, the method for constructing the correlation matrix is ​​as follows: Let any two nodes be labeled as node i and node j. Obtain the shortest merge path length between nodes i and j in the pipeline topology. Based on the merge relationship between nodes i and j in the pipeline topology, set the merge level weight between nodes i and j. Obtain the pipe diameter of the respective merge pipes to which nodes i and j are connected, and take the ratio of the minimum pipe diameter to the maximum pipe diameter as the pipe diameter correction coefficient. Multiply the reciprocal of the shortest merge path length, the merge level weight, and the pipe diameter correction coefficient to obtain the pipeline topology association parameters. Obtain the key performance parameters of nodes i and j under different operating conditions, construct the key performance parameter sequences, calculate the Pearson correlation coefficient between nodes i and j corresponding to the same key performance parameter sequence, calculate the mean Pearson correlation coefficient of nodes i and j under all operating conditions, and obtain the mean correlation coefficient of the same key performance parameter sequence of nodes i and j; weighted summation of the mean correlation coefficient of all key performance parameter sequences of nodes i and j to obtain the wastewater hydraulic coupling parameters. The pipeline topology association parameters and the sewage hydraulic coupling parameters are weighted and added together to obtain the edge weights between node i and node j; an initial node matrix is ​​constructed, and the edge weights between each node are sequentially mapped to the corresponding positions in the initial node matrix to obtain the association matrix.

[0011] Furthermore, the method for constructing the set of node performance feature matrices is as follows: Obtain the key performance parameters of each qualified solution at each node under different working conditions, and concatenate them according to the order of working conditions and the order of key performance parameters to obtain a one-dimensional feature vector of each qualified solution at each node. Traverse the pool of qualified solutions for each node, and select one design solution from each node's pool to form a combination of regional design solutions. Eliminate invalid regional design solution combinations using preset pruning rules to obtain a set of valid solution combinations. For any valid solution combination in the set of valid solution combinations, sort the one-dimensional feature vectors corresponding to the design solutions selected by each node in the corresponding valid solution combination according to the node order to generate the node performance feature matrix of the corresponding valid solution combination. Based on the node performance feature moments of all valid solution combinations, construct a set of node performance feature moments.

[0012] Furthermore, the method for the neural network model to output a global performance prediction set of different effective scheme combinations is as follows: The correlation matrix and the set of node performance feature matrices are input into the pre-built neural network model. The correlation matrix is ​​used to perform graph convolution feature transfer and aggregation operations on the set of node performance feature matrices, and the deep feature matrices of each node are output. Based on the deep feature matrix of each node, feature mapping and parsing are performed node by node through the built-in fully connected decoding layer to obtain the predicted values ​​of multiple key performance parameters of each node under multiple working conditions for each effective scheme combination. The predicted values ​​of multiple key performance parameters of all nodes under the same working condition are weighted and accumulated to obtain the local performance index value of each effective scheme combination under a single working condition. By iterating through all simulated operating conditions, calculating the local performance index value corresponding to each operating condition, and integrating and arranging the local performance index values ​​corresponding to all operating conditions in chronological order, a global performance prediction set of the corresponding effective scheme combination is obtained.

[0013] Furthermore, the method for selecting the optimal design scheme for the septic tank to be evaluated is as follows: The local efficiency index values ​​corresponding to each working condition within the global efficiency prediction set of the effective scheme combination are weighted and accumulated to obtain the comprehensive energy efficiency score of each effective scheme combination. Based on the comprehensive energy efficiency score, the effective scheme combinations are sorted in descending order to obtain the effective scheme combination ranking table. The design scheme corresponding to the node of the septic tank to be evaluated within the top-ranked effective scheme combination in the effective scheme combination ranking table is selected as the optimal design scheme.

[0014] A big data-based septic tank simulation design efficiency evaluation system, comprising: The digital simulation modeling module is used to establish a corresponding digital simulation model for each septic tank to be evaluated within the target area, and to set corresponding design parameters for each septic tank to be evaluated in the digital simulation model based on the predetermined design scheme. The simulation operation module is used to simulate different steady-state operating conditions, variable operating conditions and long-term operating conditions. Based on the design parameters, it performs CFD and biochemical reaction coupled simulation on the digital simulation model, and acquires the key performance parameters generated by each septic tank to be evaluated in real time during the simulation operation. The qualified design scheme evaluation module is used to obtain key performance benchmark information, combine key performance parameters to generate performance evaluation coefficients for each septic tank to be evaluated under each corresponding design scheme, evaluate the qualification of each design scheme in turn based on the performance evaluation coefficients, and build a qualified design scheme pool for each septic tank to be evaluated based on all qualified design schemes. The correlation matrix construction module is used to treat each septic tank to be evaluated in the target area as a node, calculate the pipe network topology correlation parameters and sewage hydraulic coupling parameters between any two nodes, and obtain the edge weights through fusion analysis, and construct the correlation matrix based on the edge weights; The feature matrix construction module is used to traverse the pool of qualified solutions for each node, filter out effective solution combinations, generate node performance feature matrices under different effective solution combinations, and construct a set of node performance feature matrices based on the node performance feature matrices.

[0015] The optimal solution selection module is used to input the correlation matrix and the set of node performance feature matrices into a pre-built neural network model. The neural network model outputs a global performance prediction set of different effective solution combinations. Based on the global performance prediction set, the performance of all effective solution combinations is evaluated, and the optimal design scheme for each septic tank to be evaluated is selected based on the evaluation results.

[0016] The technical effects and advantages of the big data-based septic tank simulation design efficiency evaluation method and system of this invention are as follows: This invention constructs a CFD and biochemical coupled simulation system covering three scenarios: steady-state, variable operating conditions, and long-term operation. It comprehensively simulates the performance of septic tanks under complex operating conditions such as actual water usage fluctuations, water quality changes, and long-term aging. At the same time, it constructs key performance benchmarks based on big data, calculates the benchmark confidence level through three-dimensional indicators of data sufficiency, timeliness, and coverage, and then generates quantitative performance evaluation coefficients by combining steady-state performance deviation coefficients, variable operating condition adaptability deviation coefficients, and long-term decay deviation coefficients to evaluate the design scheme. This avoids the limitations of traditional single-condition static verification and subjective setting of experience thresholds, and realizes an objective and accurate evaluation of the qualification of single-tank schemes, effectively reducing the engineering risk of design compliance but actual use failure.

[0017] This invention treats all septic tanks to be evaluated within the target area as coupled nodes. By constructing an association matrix that integrates the pipe network topology and sewage hydraulic coupling, it quantifies the physical confluence dependence and efficiency linkage effect between nodes. It models the physical coupling strength through confluence path, hierarchy, and pipe diameter matching degree, and captures the hydraulic efficiency correlation between nodes through Pearson correlation coefficient. This accurately reconstructs the dynamic process of upstream water quality and quantity fluctuations being transmitted downstream, avoiding the problems of hydraulic incompatibility between nodes, pipe network blockage, and downstream efficiency chain decline in traditional isolated designs. It achieves global collaborative optimization of all septic tanks to be evaluated within the target area.

[0018] This invention employs a graph convolutional neural network model adapted to the coupling characteristics of nodes. Based on the set of correlation matrices and node performance feature matrices, it achieves deep modeling of the coupling relationship between nodes and global performance prediction under multiple operating conditions. Compared with the traditional simple index weighting method, it can more accurately capture the nonlinear linkage effect between nodes. At the same time, by weighting and accumulating local performance indicators under multiple operating conditions to generate a comprehensive energy efficiency score, the optimal design scheme is selected. This can take into account the short-term stability, dynamic adaptability and long-term reliability of the scheme, ensuring that the selected optimal scheme not only meets the global performance optimization of the target area, but also meets the hydraulic compatibility and cost control requirements of the actual project. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method of the present invention; Figure 2 This is a system module block diagram of the present invention. Detailed Implementation

[0020] 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 skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] In one embodiment, a method for evaluating the design effectiveness of septic tanks based on big data simulation is disclosed, such as... Figure 1 As shown, the evaluation methods mainly include: A corresponding digital simulation model is established for each septic tank to be evaluated within the target area, and the corresponding design parameters are set for each septic tank to be evaluated in the digital simulation model based on the predetermined design scheme.

[0022] In this embodiment, multiple design schemes are first pre-defined for the septic tank to be evaluated. Corresponding design parameters can be set for the septic tank in the digital simulation model, including total tank volume, number of compartments, baffle placement, baffle outlet size, inlet and outlet pipe diameters, inlet design load, and nominal hydraulic retention time. Reasonable value ranges are set for each design parameter. Within these ranges, discrete values, uniform sampling, or orthogonal experiments are used to generate multiple pre-defined design schemes with different structural and operational parameters. Each design scheme corresponds to a septic tank structural layout and operational configuration, and these are individually modeled and simulated in the subsequent digital simulation model. Then, a corresponding digital simulation model is established for each septic tank to be evaluated within the target area. This digital simulation model is a digital twin model, and the establishment method is existing technology and will not be described in detail here. For example, based on the septic tank to be evaluated... The actual structural dimensions of the septic tank compartments, partitions, inlet channels, outlet channels, sludge settling zones, and anaerobic reaction zones were used to construct a three-dimensional geometric solid model on scale. Small structures that did not affect the flow field and biochemical reaction simulation were reasonably simplified. Next, the sewage flow and biochemical reaction areas inside the septic tank to be evaluated were selected as the simulation computation domain. The computation domain was meshed, and the mesh was locally refined at key locations such as the inlet, outlet, and partition flow outlet. Then, a fluid turbulence model, a solid-liquid multiphase flow model, and an organic matter biochemical degradation kinetic model were used to set the sewage medium properties and configure the inlet flow and water quality boundaries, the outlet free flow boundary, and the tank wall solid boundary. Finally, the structural and operational parameters corresponding to each predetermined design scheme were assigned to the model, completing the construction of the three-dimensional digital simulation model for each predetermined design scheme, and then conducting multi-condition simulations.

[0023] Simulate different steady-state, variable, and long-term operating conditions. Based on the design parameters, perform CFD and biochemical reaction coupled simulation on the digital simulation model, and obtain the key performance parameters generated by each septic tank to be evaluated in real time during the simulation operation.

[0024] In this embodiment, a CFD (Computational Fluid Dynamics) coupled simulation model with biochemical reaction is established for each septic tank to be evaluated within the target area; during the model construction process, simulation is performed separately according to the design parameters set by multiple preset design schemes.

[0025] Specifically, simulations were conducted for three types of operating conditions: steady-state conditions, variable conditions, and long-term operating conditions. For steady-state conditions, multiple fixed influent loads (e.g., 50%, 100%, 150% of rated load) were set to simulate a 24-hour stable operating state, and the changes of key performance parameters over time were collected during the simulation. For variable conditions, the influent load was continuously varied from a minimum (e.g., 30% of rated load) to a maximum (200% of rated load) to simulate the dynamic operating state under fluctuating influent, and the changes of key performance parameters with load were collected. For long-term operating conditions, the septic tank under evaluation was simulated to operate continuously for n cycles (e.g., 36 weeks, with each cycle as a week), and the changes of key performance parameters with each operating cycle were collected. During the simulation, key performance parameters of each septic tank under evaluation were collected in real time. These key performance parameters include, but are not limited to: COD removal rate, BOD removal rate, SS removal rate, hydraulic retention time distribution uniformity, dead zone ratio, sludge deposition rate, and influent / outfluent pressure difference, etc., to facilitate subsequent analysis.

[0026] Obtain key performance benchmark information, combine key performance parameters to generate performance evaluation coefficients for each septic tank to be evaluated under each corresponding design scheme, evaluate the qualification of each design scheme in turn based on the performance evaluation coefficients, and construct a qualified design scheme pool for each septic tank to be evaluated based on all qualified design schemes.

[0027] In this embodiment, key performance benchmark information is first extracted from the wastewater treatment engineering big data platform. This key performance benchmark information is obtained based on massive historical engineering data, simulation data, and industry standards and specifications. The key performance benchmark information consists of two parts: a standardized benchmark function layer and a bottom big data metadata layer. The bottom big data metadata layer includes engineering metadata (such as massive historical operation data of septic tanks already in operation, simulation verification data of similar projects, and performance benchmark data in industry standards and specifications), operating condition label data (such as steady-state load level, influent load fluctuation type, operating time level, etc.), and sample attribute data (such as sample collection time, sample size statistics, operating condition distribution information, etc.). The standardized benchmark function layer constructs standard performance functions (such as standard time variation functions and standard load variation functions of key performance parameters) for septic tanks under different operating conditions based on massive historical operating data of septic tanks already in operation, simulation verification data of similar projects, and industry standards and specifications. Polynomial fitting or nonlinear regression methods can be used to fit the standard performance functions of each key performance parameter. The underlying big data metadata layer in the key performance benchmark information is used to calculate the benchmark confidence level. Combined with each key performance parameter, performance evaluation coefficients are generated for each septic tank design scheme to be evaluated. Based on these performance evaluation coefficients, the suitability of each design scheme is assessed. This process includes the following sub-steps. Under multiple steady-state operating conditions, the steady-state performance deviation coefficient of the septic tank to be evaluated is calculated based on the different simulated influent loads. Based on the different influent loads under the simulated m steady-state operating conditions, a detection period is set for each steady-state operating condition. According to the changes in each key performance parameter of the septic tank to be evaluated over time under each steady-state operating condition within the detection period, a time-varying function for each key performance parameter is proposed. Based on the key performance benchmark information, a standard time-varying function for each key performance parameter's time-varying function is constructed. Within the detection period, the absolute value of the first integral between the time-varying function of the key performance parameter and the corresponding standard time-varying function is calculated. The weighted summation of the absolute values ​​of the first integrals of all key performance parameters yields the operating performance deviation for each steady-state operating condition. The weighted summation of the operating performance deviations for all steady-state operating conditions yields the steady-state performance deviation coefficient.

[0028] In this embodiment, m steady-state conditions are first set. Each steady-state condition refers to setting a fixed influent load and keeping it constant throughout the entire testing cycle. For example, the low-load steady-state condition is 50% of the rated influent load, the medium-load steady-state condition is 100% of the rated influent load, and the high-load steady-state condition is 150% of the rated influent load. The absolute value of the first integral represents the overall deviation between the time variation function of the key performance parameter and the corresponding standard time variation function throughout the entire testing cycle. The larger the value, the further the corresponding key performance parameter deviates from the standard benchmark and the worse the fit. This indicates that the corresponding design... The worse the effect achieved by the design scheme, the more severe the deviation of the operating condition efficiency is. The deviation of the operating condition efficiency is obtained by comprehensively analyzing the deviation of multiple key efficiency parameters (such as COD, dead zone, residence time, etc.). The larger the value, the worse the steady-state treatment performance of the design scheme. The steady-state efficiency deviation coefficient is obtained by weighted summation of the deviations under all steady-state operating conditions. The larger the value, the more serious the deviation of the overall steady-state treatment efficiency of the design scheme from the big data standard benchmark when the design scheme is running smoothly under low, medium and high constant influent loads. This indicates that the comprehensive performance of the design scheme under normal and stable operating conditions is worse and the rationality of the design scheme is lower.

[0029] Under varying influent load conditions, the adaptive deviation coefficient of the septic tank is calculated. A varying influent load from minimum to maximum is defined. Based on the changes in each key performance parameter of the septic tank under evaluation with the influent load, a load variation function for each key performance parameter is proposed. Based on the key performance benchmark information, a standard load variation function for each key performance parameter load variation function is constructed. Within the varying conditions, the absolute value of the second integral of the key performance parameter load variation function and the corresponding standard load variation function is calculated. The absolute values ​​of the second integrals of all key performance parameters are weighted and summed to obtain the adaptive deviation coefficient.

[0030] In this embodiment, a continuously varying influent load is set from a minimum (e.g., 30% of rated load) to a maximum (200% of rated load). Within this varying operating condition, based on the changes in each key performance parameter with the influent load, a load variation function for each key performance parameter is formulated. Based on the key performance benchmark information, a standard load variation function corresponding to each key performance parameter load variation function is constructed. The second absolute value of the integral is then calculated. The second absolute value of the integral characterizes the overall deviation between the key performance parameter load variation function and the corresponding standard load variation function within the varying operating condition range. The larger the value, the higher the degree of deviation of each key performance parameter from the standard benchmark in the actual operation of the septic tank, the worse the operational stability, and the worse the design scheme. Similarly, the variable operating condition adaptability deviation coefficient is used to quantify the overall deviation of each key performance parameter of the septic tank under fluctuating influent load. The larger the value, the weaker the adaptive ability of the septic tank to the variable operating condition, indicating a worse corresponding setting scheme and a lower rationality of the design scheme.

[0031] Under long-term operating conditions, the long-term attenuation deviation coefficient of the septic tank is calculated. Within n consecutive normal operating cycles, the key performance parameters of the septic tank under evaluation are obtained for each operating cycle. The average performance parameter of each key performance parameter in each operating cycle is calculated, and a benchmark value for the average performance parameter is set for each operating cycle. The absolute value of the difference between the average performance parameter and the corresponding benchmark value is calculated for each operating cycle. The attenuation coefficient of each key performance parameter is obtained by weighted summation of the absolute values ​​of the differences across all operating cycles. The long-term attenuation deviation coefficient is obtained by weighted summation of the attenuation coefficients of all key performance parameters.

[0032] In this implementation, a continuous normal operation cycle refers to the simulation timeline, starting from the initial operation time of the digital simulation model, progressively setting the first virtual operation week, the second virtual operation week, and so on, up to the nth virtual operation cycle corresponding to the preset simulation design life. Each operation cycle has a baseline value for the average performance parameter corresponding to each key performance parameter. This baseline value can be obtained by collecting a large amount of long-term operation monitoring data from septic tanks that have been actually put into operation, with the same tank size, the same influent conditions, and the same regional environment. The weekly average value of each key performance parameter is calculated week by week to form a historical cycle baseline sample, thereby obtaining the corresponding average performance parameter baseline value for each operation cycle. The larger the absolute value of the difference, the more serious the deviation of the performance of this design scheme from the standard level in the current cycle, and the potential risks of premature decay, substandard treatment efficiency, and abnormal sludge deposition. The decay of each key performance parameter is obtained by weighted summation of the absolute values ​​of the differences of the key performance parameters over all operation cycles. The attenuation coefficient indicates the severity of a single performance indicator's overall deviation from the standard attenuation pattern throughout its entire life cycle. The larger the value, the weaker the single indicator's ability to resist aging, attenuation, and clogging, indicating a less reasonable design scheme. The long-term attenuation deviation coefficient is obtained by weighting and summing the attenuation coefficients of all key performance parameters. The long-term attenuation deviation coefficient characterizes the overall deviation level of the septic tank's design scheme from the industry standard attenuation benchmark in terms of comprehensive treatment efficiency, hydraulic flow field characteristics, and sludge deposition characteristics during long-term operation throughout its entire life cycle. The larger the value, the more defective the design scheme is, indicating rapid attenuation of efficiency, deterioration of the flow field, severe siltation, and poor long-term service performance, indicating a lower degree of rationality in the design scheme.

[0033] It should be noted that the weighting coefficients mentioned above can be set independently based on the experience of people in this field and the professional knowledge of related industries, and will not be described in detail here.

[0034] Finally, corresponding weighting coefficients are assigned to the steady-state performance deviation coefficient, the variable operating condition adaptability deviation coefficient, and the long-term degradation deviation coefficient. These coefficients are then multiplied by their respective weighting coefficients and summed to obtain the comprehensive deviation coefficient. A comprehensive deviation coefficient threshold is preset, and a benchmark confidence level is calculated based on key performance benchmark information. The comprehensive deviation coefficient is divided by the product of the comprehensive deviation coefficient threshold and the benchmark confidence level to obtain the performance evaluation coefficient. If the performance evaluation coefficient is less than 1, the design scheme is considered qualified; otherwise, it is considered unqualified. The benchmark confidence level is calculated as follows: The data sufficiency sub-coefficient is obtained by extracting the number of valid samples and the total number of samples from the key performance benchmark information. The number of valid samples is divided by the total number of samples. The acquisition time of each valid sample is obtained, and the average acquisition time of the valid samples is calculated. The current simulation time is recorded as the current evaluation time. The interval between the average acquisition time and the current evaluation time is obtained. A time decay coefficient is introduced, and the data timeliness sub-coefficient is calculated through a preset exponential decay model. The operating condition parameters corresponding to each valid sample are extracted from the key performance benchmark information. The operating condition parameters are pre-divided into levels to form a gridded target operating condition combination. The number of valid samples in each target operating condition combination is counted. Based on the number of valid samples corresponding to each target operating condition combination, the corresponding sample proportion probability is calculated. The Shannon information entropy formula is introduced to calculate the actual operating condition distribution entropy value. The theoretical maximum information entropy corresponding to all target operating condition combinations is calculated. The ratio of the actual operating condition distribution entropy value to the theoretical maximum information entropy is used as the coverage integrity sub-coefficient. The data sufficiency sub-coefficient, the data timeliness sub-coefficient, and the coverage integrity sub-coefficient are multiplied together and the cube root is taken to obtain the benchmark confidence level.

[0035] In this embodiment, the weighting coefficients for the steady-state performance deviation coefficient, the variable operating condition adaptability deviation coefficient, and the long-term decay deviation coefficient can be set independently based on the importance ratio of the actual operating conditions of the septic tank to be evaluated and the statistical data of historical engineering samples. The comprehensive deviation coefficient threshold can be obtained by referring to industry design specifications and standards, and based on the statistical critical values ​​of a large number of verified and qualified septic tank simulation design samples. In order to correct the reliability of the big data performance benchmark, a multi-dimensional quality assessment model is constructed to calculate the benchmark confidence coefficient. Valid samples refer to septic tank performance-related data samples selected from all samples that can support the construction of standardized benchmark functions and the calculation of benchmark confidence. These include engineering metadata that has been removed from abnormal situations such as equipment failure and monitoring errors, operating condition label data containing clear operating parameters (such as influent load, temperature, and water quality), and sample attribute data with collection timestamps. The data sufficiency sub-coefficient reflects the sufficiency of the sample size, and the data timeliness sub-coefficient reflects the timeliness of the data. Here, the time decay coefficient... The coefficient can be set manually based on experience; the coverage integrity coefficient reflects the breadth and uniformity of the operating conditions, where operating conditions parameters include, but are not limited to, influent load parameters, ambient temperature parameters, and influent water quality pollutant concentration parameters; the preset exponential decay model, actual operating condition distribution entropy value, and theoretical maximum information entropy calculation method are all existing technologies and will not be described in detail here; the benchmark confidence level is used to quantitatively characterize the comprehensive credibility and engineering applicability of the key performance benchmark information itself, and the larger the value, the higher the reference value and assessment credibility; the smaller the performance evaluation coefficient, the smaller the overall deviation of the septic tank design scheme from the standard, the better the steady-state treatment effect, the stronger the resistance to shock under changing operating conditions, the controllable long-term sludge deposition and performance decay, the higher the design rationality, and the better the simulation design efficiency; therefore, when the performance evaluation coefficient is less than 1, it means that the comprehensive deviation is within the credibility threshold range, indicating that the septic tank simulation design scheme is qualified and meets the requirements of engineering design and long-term operation, and the evaluation design scheme is qualified; otherwise, the evaluation design scheme is unqualified.

[0036] This approach comprehensively covers the characteristics of various operating conditions of septic tanks from three dimensions: steady-state performance, adaptability to changing operating conditions, and long-term operational degradation characteristics. It overcomes the limitations of traditional methods that rely solely on single indicators or empirical thresholds. Furthermore, by fusing data sufficiency, completeness of operating condition coverage, and the geometric mean of data timeliness coefficients to obtain the baseline confidence level, it addresses the limitations of data quality and adaptively corrects for issues such as insufficient sample size, incomplete operating condition coverage, and outdated data in the big data performance benchmark itself. Then, through normalized performance evaluation coefficients, it quantifies the results, eliminating the arbitrariness of subjective human judgment. This allows for precise quantification of the deviation between the simulated design scheme and the optimal performance benchmark in actual engineering. It effectively avoids misjudgments and omissions caused by insufficient benchmark reliability and objectively identifies design flaws in the septic tank's structure, layout, and operating parameters. This provides a quantitative, rigorous, and practically accurate evaluation basis for the rational selection and parameter optimization of septic tank simulation design schemes, significantly improving the accuracy of septic tank design scheme evaluation.

[0037] Each septic tank to be evaluated within the target area is taken as a node. The pipe network topology correlation parameters and sewage hydraulic coupling parameters between any two nodes are calculated and fused to obtain the edge weights. The correlation matrix is ​​constructed based on the edge weights.

[0038] In this embodiment, each septic tank to be evaluated within the target area is treated as a node, and a node association matrix containing both pipe network topology association and hydraulic coupling association is constructed. Specifically: Let any two nodes be labeled as node i and node j. Obtain the shortest merge path length between nodes i and j in the pipeline topology. Based on the merge relationship between nodes i and j in the pipeline topology, set the merge level weight between nodes i and j. Obtain the pipe diameters of the respective merge pipes to which nodes i and j are connected, and take the ratio of the minimum pipe diameter to the maximum pipe diameter as the pipe diameter correction coefficient. Multiply the reciprocal of the shortest merge path length, the merge level weight, and the pipe diameter correction coefficient to obtain the pipeline topology association parameters. Obtain the key performance parameters of nodes i and j under different operating conditions, construct the sequence of key performance parameters, and calculate the performance parameters of node i. The Pearson correlation coefficient between nodes i and j corresponding to the same key performance parameter sequence is calculated. The mean Pearson correlation coefficient between nodes i and j under all operating conditions is obtained. The mean correlation coefficient between nodes i and j corresponding to the same key performance parameter sequence is then weighted and summed to obtain the wastewater hydraulic coupling parameter. The pipeline topology association parameter and the wastewater hydraulic coupling parameter are weighted and summed to obtain the edge weights between nodes i and j. An initial node matrix is ​​constructed, and the edge weights between each node are sequentially mapped to the corresponding positions in the initial node matrix to obtain the association matrix.

[0039] In the above scheme, the shortest confluence path length is the total length of the pipeline network through which sewage flows from the outlet of node i to the inlet of node j. If there is no direct or indirect confluence relationship between node i and node j, the pipeline topology association parameter is set to 0. Based on the confluence relationship between node i and node j in the pipeline network, a corresponding confluence level weight is set. For example, if node i is a direct upstream node of node j (sewage flows from i to j through only one pipeline segment), the confluence level weight is set to 1.0. If node i is a secondary upstream node of node j (sewage flows to j after being transferred through one intermediate node), the confluence level weight is set to 0.7, and so on. The more confluence relationships there are, the smaller the weight. If there is no confluence relationship between node i and node j, the confluence level weight is 0. The process involves obtaining the outlet pipe diameter of node i connected to the manifold network and the inlet pipe diameter of node j connected to the manifold network. The ratio of the minimum to the maximum value of the two pipe diameters is taken as the pipe diameter correction coefficient. This correction coefficient is used to correct the impact of pipe diameter mismatch on sewage transport efficiency: if the upstream outlet pipe diameter is much larger than the downstream inlet pipe diameter, sewage is prone to blockage in the pipe network, and the actual coupling between the two nodes will be significantly reduced; conversely, the higher the pipe diameter matching degree, the closer the correction coefficient is to 1, and the stronger the coupling. Finally, the reciprocal of the shortest confluence path length, the confluence level weight, and the pipe diameter correction coefficient are multiplied to obtain the pipe network topology association parameters. The pipe network topology association parameters quantify the coupling strength of the two septic tank nodes at the physical connection level of the sewage pipe network. The larger the value, the tighter the physical confluence dependency between the two nodes. Based on the key performance parameters (such as COD removal rate, BOD removal rate, effluent flow rate, sludge deposition rate, etc.) collected during the simulation of a single septic tank to be evaluated, time series sequences of each key performance parameter under all operating conditions (including steady-state, variable, and long-term operating conditions) are constructed for nodes i and j. The Pearson correlation coefficient between the sequences of the same key performance parameter for nodes i and j is calculated to quantify the linear correlation between the two nodes on that key performance parameter. The average Pearson correlation coefficient of the same key performance parameter for nodes i and j under all operating conditions is taken to eliminate the influence of random fluctuations under a single operating condition, thus obtaining the average correlation coefficient corresponding to that key performance parameter. Based on the engineering importance of different key performance parameters, corresponding weight coefficients are pre-set according to experience (e.g., COD removal rate weight 0.4, effluent flow rate weight 0.3, sludge deposition rate weight 0.2, BOD removal rate weight 0.1), and the weights of each parameter are... The correlation coefficients are weighted and summed to obtain the sewage hydraulic coupling parameters. These parameters characterize the correlation between the treatment efficiency of two septic tank nodes under different operating conditions. If the efficiency fluctuation of node i directly affects the efficiency of node j (e.g., fluctuations in upstream effluent quality directly affect downstream influent load), then the larger the hydraulic coupling parameter values ​​of the two nodes, the stronger the efficiency correlation. Finally, the pipeline topology correlation parameters and the sewage hydraulic coupling parameters are weighted and summed to obtain the edge weights between node i and node j. The weights of the pipeline topology correlation parameters and the sewage hydraulic coupling parameters can be set based on the experience and professional knowledge of those in the field. An initial node matrix is ​​constructed, with the dimension of the matrix consistent with the total number of septic tank nodes N in the target area. The edge weights between each node are sequentially mapped to the corresponding positions in the initial node matrix to obtain the correlation matrix. This correlation matrix can be directly used as the adjacency matrix input for the subsequent neural network model, providing structured node correlation information support for global efficiency prediction.

[0040] Traverse the pool of qualified solutions for each node, select effective solution combinations, and generate node performance feature matrices under different effective solution combinations. Construct a set of node performance feature matrices based on the node performance feature matrices.

[0041] In this embodiment, the method for constructing the node performance feature matrix set is as follows: Key performance parameters of each qualified scheme for each node under different operating conditions are obtained and concatenated according to the order of operating conditions and key performance parameters to obtain a one-dimensional feature vector for each qualified scheme of each node; the pool of qualified schemes for each node is traversed, and one design scheme is selected from the pool of qualified schemes for each node to form a regional design scheme combination; invalid regional design scheme combinations are removed through preset pruning rules to obtain a set of effective scheme combinations; for any effective scheme combination in the set of effective scheme combinations, the one-dimensional feature vectors corresponding to the design schemes selected by each node in the corresponding effective scheme combination are sorted according to the node order to generate the node performance feature matrix of the corresponding effective scheme combination; based on the node performance feature moments of all effective scheme combinations, a set of node performance feature moments is constructed.

[0042] In the above scheme, the order of operating conditions and key performance parameters can be predetermined in advance. For example, the first priority is from steady-state operating conditions to variable operating conditions to long-term operating conditions, and the second priority is from COD removal rate to BOD removal rate to SS removal rate to RTD uniformity to dead zone ratio to sludge deposition rate. All key performance parameter values ​​under each operating condition are concatenated in sequence to obtain a one-dimensional performance feature vector of each qualified scheme. Each element of the vector corresponds to the performance value under a specific operating condition and specific parameters. If the target region contains N nodes, and each node's pool of qualified solutions contains K design schemes, then the theoretical number of scheme combinations is: When the number of nodes is large, the combinatorial explosion problem occurs. Directly calculating the global performance of all combinations leads to an exponential increase in computational cost. Therefore, it is necessary to use preset pruning rules to eliminate invalid and redundant scheme combinations, retaining only effective combinations with engineering value, thus obtaining a set of effective scheme combinations. For any effective scheme combination in the set of effective scheme combinations, the nodes are arranged in the order of convergence in the pipeline network (upstream → midstream → downstream). The one-dimensional feature vectors of each node are used as row vectors and stacked sequentially to generate the node performance feature matrix corresponding to the effective scheme combination. In the node performance feature matrix, the row dimension represents the septic tank nodes in the area, and each row corresponds to the performance feature vector of a node. The column dimension represents the performance feature dimension of the node, corresponding to the splicing result of the working condition and key performance parameters. The node performance feature matrices corresponding to all effective scheme combinations are summarized to construct a set of node performance feature matrices. Each node performance feature matrix in this set corresponds one-to-one with the subsequent association matrix to ensure that the node features and node associations can be accurately matched in subsequent graph convolution operations.

[0043] It should be noted that the preset pruning rules can be combined with the engineering characteristics of the sewage pipe network system, and four types of pruning rules can be preset (hydraulic incompatibility pruning, water quality conflict pruning, global blockage risk pruning, and redundant combination pruning) to filter the initial scheme combinations and eliminate combinations without practical engineering value. Hydraulic incompatibility pruning can be: if the effluent flow rate of the upstream node is seriously mismatched with the influent load of the downstream node (such as the upstream effluent flow rate exceeding 200% of the downstream rated load, or falling below 30%), then it is determined that the combination will cause hydraulic shock or insufficient hydraulic retention time at the downstream node, and it is directly eliminated. Water quality conflict pruning can be: if the effluent COD / BOD concentration of the upstream node exceeds the influent tolerance threshold of the downstream node (such as if the downstream node's designed influent COD is 30%), then it is determined that the combination will cause hydraulic shock or insufficient hydraulic retention time at the downstream node, and it is directly eliminated. If the COD of the upstream effluent exceeds 400 mg / L, the combination is deemed unable to meet the treatment requirements of the downstream nodes and is removed. Global clogging risk pruning can be implemented as follows: if more than a preset proportion (e.g., 30%) of the node schemes in the combination have a sludge deposition rate exceeding the pipeline clogging warning threshold under long-term operating conditions, the combination is deemed to have a regional-level pipeline clogging risk and is removed. Redundant combination pruning can be implemented as follows: if the difference between the edge weights (i.e., the edge weights in the correlation matrix) of nodes in multiple scheme combinations is less than a preset threshold (e.g., 0.05), it is determined to be a highly similar combination, and only one group is retained, eliminating redundant combinations to reduce computational load. This can be set through the experience and professional knowledge of those in the field, and will not be described in detail here.

[0044] The correlation matrix and the set of node performance feature matrices are input into a pre-built neural network model. The neural network model outputs a global performance prediction set of different effective scheme combinations. Based on the global performance prediction set, the performance of all effective scheme combinations is evaluated, and the optimal design scheme of the septic tank to be evaluated is selected based on the evaluation results.

[0045] In this embodiment, the method for obtaining the global performance prediction set is as follows: The correlation matrix and the set of node performance feature matrices are input into a pre-constructed neural network model. Graph convolution feature transfer and aggregation operations are performed on the set of node performance feature matrices through the correlation matrix to output the deep feature matrices of each node. Based on the deep feature matrices of each node, feature mapping is performed node by node through a built-in fully connected decoding layer to obtain the predicted values ​​of multiple key performance parameters of each node under multiple operating conditions for each effective scheme combination. The predicted values ​​of multiple key performance parameters of all nodes under the same operating condition are weighted and accumulated to obtain the local performance index value of each effective scheme combination under a single operating condition. All simulated operating conditions are traversed sequentially, and the local performance index value corresponding to each operating condition is calculated one by one. The local performance index values ​​corresponding to all operating conditions are integrated and arranged in the order of the operating conditions to obtain the global performance prediction set of the corresponding effective scheme combination.

[0046] In this embodiment, a detailed explanation is given using a target area containing N=3 nodes (nodes A, B, and C), each node corresponding to 11 preset operating conditions (such as steady-state conditions with 50% / 100% / 150% load, variable conditions with 5 fluctuation points, and long-term operating conditions for 12 / 24 / 36 weeks) and 6 key performance parameters (such as COD removal rate, BOD removal rate, SS removal rate, hydraulic retention time uniformity, flow field dead zone ratio, and sludge deposition rate). The neural network model uses a graph convolutional neural network as the core prediction model. The input includes an association matrix and a set of node performance feature matrices, corresponding to the association relationships between nodes and the performance features of the nodes themselves, respectively. The association matrix serves as the adjacency matrix input of the model, with a dimension of N*N (3*3 in this example). The matrix elements are the edge weights between nodes, quantifying the network topology association and sewage hydraulic coupling strength between nodes. This matrix is ​​an undirected pairwise pair. The matrix is ​​called a node self-loop weight (1.0 in this embodiment) on the diagonal to ensure that each node's features contain its own information during the transmission process. Then, the influence of node degree differences is eliminated by degree matrix normalization, resulting in a normalized association matrix A, ensuring a balanced distribution of node weights during feature transmission. The node performance feature matrix set serves as the node feature input to the model. Each element in the set is an N*D matrix (D is the node feature dimension; in this embodiment, D=11*6=66, corresponding to 11 working conditions × 6 key performance parameters). Each matrix corresponds to an effective scheme combination. The row dimension represents the septic tank nodes within the target area, and the column dimension represents the multi-working-condition performance feature vector of that node under the corresponding scheme. The graph convolutional neural network can perform weighted aggregation of node features based on the adjacency matrix, automatically capturing the coupling relationship between nodes, and achieving deep feature transmission and fusion. The specific process is as follows: The first layer of the graph convolutional neural network (GCN layer) performs first-order adjacency aggregation on the node performance feature matrix, outputting a first-order deep feature matrix, expressed as:

[0047] in, It is a first-order deep feature matrix. It is the ReLU activation function. The input is the node performance feature matrix. The first layer is a learnable weight matrix. This layer maps the initial 66-dimensional features of each node to 32-dimensional hidden features, and outputs a first-order hidden feature matrix with a dimension of N×32. Each row of the matrix corresponds to the first-order hidden feature vector of a node. The second graph convolutional layer performs second-order adjacency aggregation based on the first-order latent feature matrix output from the first layer, capturing the indirect coupling relationships between nodes (such as the indirect influence of upstream nodes on downstream nodes), and outputs the deep feature matrix of each node, expressed as:

[0048] in, The second layer is a learnable weight matrix, and the output is a node deep feature matrix with a dimension of N×16. Each row of this matrix is ​​a 16-dimensional deep feature vector of a single node, which integrates the node's own performance information, the direct neighbor coupling effect and the indirect neighbor transmission effect, and realizes deep modeling of the relationship between nodes.

[0049] Based on the output node deep feature matrix, feature mapping is performed node by node through the built-in fully connected decoding layer to map the deep features in the latent space back to the original performance parameter space. The specific implementation is as follows: The decoding layer takes each row of the node's deep feature matrix (i.e., the 16-dimensional deep feature vector of a single node) as input and uses a two-layer fully connected network for mapping: the first layer is a linear layer that maps the 16-dimensional deep features to a 66-dimensional vector (consistent with the original node feature dimensions); the second layer is a linear correction layer that outputs normalized performance parameter prediction values ​​through a sigmoid activation function, ensuring that the predicted values ​​match the range of the original simulation data; finally, it outputs the predicted value sequence of each node under each effective scheme combination, with a dimension of 11×6, corresponding one-to-one with the preset operating conditions and key performance parameters, including 11 operating conditions (such as steady-state 50% / 100% / 150% load, 5 fluctuation points under variable operating conditions, and long-term 12 / 24 / 36 weeks), and the corresponding predicted values ​​of 6 key performance parameters (such as COD removal rate, BOD removal rate, SS removal rate, hydraulic retention time uniformity, flow field dead zone ratio, and sludge deposition rate). For each effective scheme combination, the predicted values ​​of key performance parameters of all nodes under the same operating condition are weighted and accumulated to obtain the local performance index value of the wastewater system in that area under that operating condition. Specifically, the predicted values ​​of different types of performance parameters can be preprocessed first: positive indicators (such as COD removal rate and BOD removal rate) are positively normalized; negative indicators (such as the proportion of dead zones in the flow field and sludge deposition rate) are reversely normalized to eliminate the polarity difference of the indicators. Then, the predicted values ​​of each key performance parameter are weighted according to the importance of the project to obtain the local performance index value of each effective scheme combination under a single operating condition. Finally, the local performance index values ​​corresponding to all operating conditions are integrated and arranged in the order of operating condition time (steady-state operating condition → variable operating condition → long-term operating condition) to obtain the global performance prediction set of the corresponding effective scheme combination. This set fully reflects the global performance change trend of the effective scheme combination under different operating stages and different load conditions.

[0050] It should be noted that the graph convolutional neural network model used for global performance prediction in this scheme includes components such as an input layer, graph convolutional layer, temporal convolutional layer, activation function, normalization layer, and fully connected decoding output layer. Specifically: the input layer receives two types of structured data: a set of node performance feature matrices and an association matrix; the graph convolutional layer performs weighted propagation and spatial fusion of the node performance feature matrices based on the association matrix, capturing the spatial association characteristics between septic tank nodes and modeling the process of upstream node performance changes being transmitted to downstream nodes through pipe network topology and hydraulic coupling; the temporal convolutional layer processes the changes in node performance feature vectors over continuous operating time, extracting the performance evolution patterns of the septic tank system under steady-state, variable operating conditions, and long-term operating conditions, capturing the performance under different loads and operating stages. The changing trend; activation functions and normalization layers are used to introduce nonlinearity by applying activation functions such as ReLU between each convolutional layer, enhancing the model's ability to express complex coupling relationships. At the same time, the normalization layer stabilizes the training process and improves the robustness of feature representation. The output layer (fully connected decoding layer) maps and generates predicted values ​​of key performance parameters of each node under different working conditions based on the spatial coupling information between nodes captured by the graph convolutional layer and the working condition evolution information extracted by the temporal convolutional layer. Then, it integrates and calculates the local performance index value, and finally forms a global performance prediction set of effective scheme combination. The core of this scheme is to adapt the graph convolutional neural network to the node coupling characteristics and multi-working condition performance data of the septic tank area system to achieve global performance prediction. Its basic architecture is based on existing technology, and the specific construction process will not be described in detail here.

[0051] The method for evaluating the performance of all effective scheme combinations based on the global performance prediction set and selecting the optimal design scheme for the septic tank to be evaluated based on the evaluation results is as follows: The local performance index values ​​corresponding to each working condition in the global performance prediction set of the effective scheme combinations are weighted and accumulated to obtain the comprehensive energy efficiency score of each effective scheme combination. Based on the comprehensive energy efficiency score, the effective scheme combinations are sorted in descending order to obtain the effective scheme combination ranking table. The design scheme of the node corresponding to the septic tank to be evaluated in the effective scheme combination with the highest ranking in the effective scheme combination ranking table is selected as the optimal design scheme.

[0052] In this embodiment, weight values ​​are first preset for each working condition corresponding to each effective scheme combination. The weight values ​​can be set according to the experience and professional knowledge of those in the field. Then, the local efficiency index values ​​corresponding to each working condition are weighted and accumulated to obtain the comprehensive energy efficiency score of each effective scheme combination. The larger the comprehensive energy efficiency score, the better the global efficiency of the effective scheme combination. Therefore, the effective scheme combinations are sorted in descending order to obtain the effective scheme combination ranking table. During the ranking process, if multiple effective scheme combinations have the same comprehensive energy efficiency score, a secondary ranking can be performed using the following additional rules: prioritize combinations with lower risk of hydraulic coupling conflict between nodes (such as higher matching degree between upstream outflow and downstream inflow load), or prioritize combinations with lower construction costs (such as smaller tank volume and lower pipe specifications) to ensure that the ranking results are more in line with the actual needs of the project. Select the effective scheme combination ranked first in the effective scheme combination ranking table, and extract the design scheme of the corresponding node of the septic tank to be evaluated within the combination as the optimal design scheme of the septic tank.

[0053] This approach constructs a node association matrix that integrates the network topology and sewage hydraulic coupling, generates a set of node performance feature matrices through pruning and dimensionality reduction, and uses a graph convolutional neural network adapted to the node coupling characteristics to achieve global performance prediction under multiple operating conditions. Finally, a comprehensive energy efficiency score is generated through weighted accumulation for ranking and selection, choosing the optimal design scheme for the septic tank to be evaluated. This effectively solves the technical defects of traditional septic tank design, where the efficiency of a single tank is optimal but the overall efficiency of the entire area is poor, and there is hydraulic incompatibility between nodes. It improves the accuracy of global performance prediction by capturing the physical topology and hydraulic coupling relationship between nodes, and significantly reduces computational complexity through invalid combination pruning. At the same time, the comprehensive evaluation under multiple operating conditions takes into account the short-term stability, dynamic adaptability, and long-term reliability of the scheme, ensuring that the final optimal design scheme achieves global collaborative optimization and meets the hydraulic compatibility and cost control requirements of actual engineering projects, greatly improving the comprehensiveness and accuracy of septic tank design scheme evaluation.

[0054] In one embodiment, a septic tank simulation design performance evaluation system based on big data is disclosed, such as... Figure 2 As shown, the system includes: The digital simulation modeling module is used to establish a corresponding digital simulation model for each septic tank to be evaluated within the target area, and to set corresponding design parameters for each septic tank to be evaluated in the digital simulation model based on the predetermined design scheme. The simulation operation module is used to simulate different steady-state operating conditions, variable operating conditions and long-term operating conditions. Based on the design parameters, it performs CFD and biochemical reaction coupled simulation on the digital simulation model, and acquires the key performance parameters generated by each septic tank to be evaluated in real time during the simulation operation. The qualified design scheme evaluation module is used to obtain key performance benchmark information, combine key performance parameters to generate performance evaluation coefficients for each septic tank to be evaluated under each corresponding design scheme, evaluate the qualification of each design scheme in turn based on the performance evaluation coefficients, and build a qualified design scheme pool for each septic tank to be evaluated based on all qualified design schemes. The correlation matrix construction module is used to treat each septic tank to be evaluated in the target area as a node, calculate the pipe network topology correlation parameters and sewage hydraulic coupling parameters between any two nodes, and obtain the edge weights through fusion analysis, and construct the correlation matrix based on the edge weights; The feature matrix construction module is used to traverse the pool of qualified solutions for each node, filter out effective solution combinations, generate node performance feature matrices under different effective solution combinations, and construct a set of node performance feature matrices based on the node performance feature matrices. The optimal solution selection module is used to input the correlation matrix and the set of node performance feature matrices into a pre-built neural network model. The neural network model outputs a global performance prediction set of different effective solution combinations. Based on the global performance prediction set, the performance of all effective solution combinations is evaluated, and the optimal design scheme for each septic tank to be evaluated is selected based on the evaluation results.

[0055] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

[0056] All formulas in this manual are dimensionless and calculated numerically. 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 and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0057] Although embodiments of the invention have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for evaluating the design efficiency of septic tanks based on big data simulation, characterized in that, The methods include: A corresponding digital simulation model is established for each septic tank to be evaluated within the target area, and corresponding design parameters are set for each septic tank to be evaluated in the digital simulation model based on the predetermined design scheme. Simulate different steady-state, variable and long-term operating conditions, perform CFD and biochemical reaction coupled simulation on the digital simulation model based on design parameters, and obtain key performance parameters of each septic tank to be evaluated in real time during the simulation operation. Obtain key performance benchmark information, combine key performance parameters to generate performance evaluation coefficients for each septic tank to be evaluated under each corresponding design scheme, evaluate the qualification of each design scheme based on the performance evaluation coefficients, and construct a qualified design scheme pool for each septic tank to be evaluated based on all qualified design schemes. Each septic tank to be evaluated within the target area is taken as a node. The pipe network topology correlation parameters and sewage hydraulic coupling parameters between any two nodes are calculated and fused to obtain the edge weights. The correlation matrix is ​​constructed based on the edge weights. Traverse the pool of qualified solutions for each node, select effective solution combinations, and generate node performance feature matrices under different effective solution combinations. Construct a set of node performance feature matrices based on the node performance feature matrices. The correlation matrix and the set of node performance feature matrices are input into a pre-built neural network model, and the neural network model outputs a set of global performance predictions for different effective scheme combinations. The effectiveness of all effective scheme combinations is evaluated based on the global effectiveness prediction set, and the optimal design scheme for each septic tank to be evaluated is selected based on the evaluation results.

2. The method for evaluating the design efficiency of septic tanks based on big data according to claim 1, characterized in that, The method for evaluating the suitability of a design scheme based on performance evaluation coefficients is as follows: Under multiple steady-state conditions, the steady-state performance deviation coefficient of the septic tank to be evaluated is calculated according to the different simulated influent loads; under variable influent load conditions, the variable-condition adaptability deviation coefficient of the septic tank to be evaluated is calculated. Under long-term operating conditions, calculate the long-term degradation deviation coefficient of the septic tank to be evaluated; Set weighting coefficients for steady-state performance deviation coefficient, variable operating condition adaptability deviation coefficient, and long-term decay deviation coefficient. Multiply the steady-state performance deviation coefficient, variable operating condition adaptability deviation coefficient, and long-term decay deviation coefficient by their respective weighting coefficients and then sum them to obtain the comprehensive deviation coefficient. A preset comprehensive deviation coefficient threshold and a baseline confidence level are set. The comprehensive deviation coefficient is divided by the product of the comprehensive deviation coefficient threshold and the baseline confidence level to obtain the performance evaluation coefficient. If the performance evaluation coefficient is less than 1, the design scheme is deemed qualified; otherwise, the design scheme is deemed unqualified.

3. The method for evaluating the design efficiency of septic tanks based on big data according to claim 2, characterized in that, The calculation methods for the steady-state performance deviation coefficient and the variable operating condition adaptability deviation coefficient are as follows: Based on the different influent loads under the simulated m steady-state conditions, the detection cycle for each steady-state condition is set. Based on the changes in the key performance parameters of the septic tank to be evaluated over time under each steady-state condition within the detection cycle, the time variation function of each key performance parameter is proposed. Based on key performance benchmark information, a standard time variation function is constructed for the time variation function of each key performance parameter. Within the detection period, the absolute value of the first integral between the time variation function of the key performance parameter and the corresponding standard time variation function is calculated. The weighted summation of the absolute values ​​of the first integral of all key performance parameters yields the performance deviation of each steady-state condition. The weighted summation of the performance deviations of all steady-state conditions yields the steady-state performance deviation coefficient. Set up a variable operating condition with the influent load from minimum to maximum value, and formulate the load variation function of each key performance parameter based on the changes of each key performance parameter of the septic tank to be evaluated with the influent load. Based on key performance benchmark information, a standard load change function is constructed for the load change function of each key performance parameter. Under varying operating conditions, the second integral absolute value of the load change function of the key performance parameter and the corresponding standard load change function is calculated. The second integral absolute values ​​of all key performance parameters are weighted and summed to obtain the variation deviation coefficient of the varying operating conditions.

4. The method for evaluating the design efficiency of septic tanks based on big data according to claim 2, characterized in that, The method for calculating the long-term attenuation deviation coefficient is as follows: Under long-term operating conditions, n consecutive normal operating cycles are set, and the key performance parameters of the septic tank to be evaluated are obtained in each operating cycle. The average performance parameter of each key performance parameter in each operating cycle is calculated. The average performance parameter benchmark value corresponding to each average performance parameter is set in each operating cycle. Calculate the absolute value of the difference between the average performance parameter and the corresponding average performance parameter benchmark value in each operating cycle. Then, weight and sum the absolute values ​​of the differences of the key performance parameters in all operating cycles to obtain the attenuation coefficient of each key performance parameter. Finally, weight and sum the attenuation coefficients of all key performance parameters to obtain the long-term attenuation deviation coefficient.

5. The method for evaluating the design efficiency of septic tanks based on big data according to claim 2, characterized in that, The method for obtaining the baseline confidence level is as follows: Extract the number of valid samples and the total number of samples from the key performance benchmark information, and divide the number of valid samples by the total number of samples to obtain the data sufficiency sub-coefficient. The acquisition time of each valid sample is obtained, the average acquisition time of the valid samples is calculated, the current time of the simulation is recorded as the current evaluation time, the interval between the average acquisition time and the current evaluation time is obtained, a time decay coefficient is introduced, and the data timeliness coefficient is calculated through a preset exponential decay model. The operating condition parameters corresponding to each effective sample are extracted from the key performance benchmark information. Each operating condition parameter is pre-divided into different levels to form a gridded target operating condition combination. The number of effective samples in each target operating condition combination is counted. Based on the number of effective samples corresponding to each target operating condition combination, the corresponding sample proportion probability is calculated. The Shannon information entropy formula is introduced to calculate the actual operating condition distribution entropy value. The theoretical maximum information entropy corresponding to all target operating condition combinations is calculated. The ratio of the actual operating condition distribution entropy value to the theoretical maximum information entropy is used as the coverage integrity sub-coefficient. The baseline confidence level is obtained by multiplying the data sufficiency sub-coefficient, the data timeliness sub-coefficient, and the coverage completeness sub-coefficient together and then taking the cube root.

6. The method for evaluating the design efficiency of septic tanks based on big data according to claim 1, characterized in that, The method for constructing the correlation matrix is ​​as follows: Let any two nodes be labeled as node i and node j. Obtain the shortest merge path length between node i and node j in the pipeline topology. Based on the merge relationship between node i and node j in the pipeline topology, set the merge level weight between node i and node j. Obtain the pipe diameter of the manifold to which node i and node j are connected, and take the ratio of the minimum pipe diameter to the maximum pipe diameter as the pipe diameter correction factor. Multiply the reciprocal of the shortest merge path length, the merge level weight, and the pipe diameter correction factor to obtain the pipe network topology association parameters; Obtain the key performance parameters of node i and node j under different working conditions, construct the key performance parameter sequence, calculate the Pearson correlation coefficient between the same key performance parameter sequence of node i and node j, calculate the mean Pearson correlation coefficient of node i and node j under all working conditions, and obtain the mean correlation coefficient of the same key performance parameter sequence of node i and node j. The wastewater hydraulic coupling parameters are obtained by weighted summing of the mean correlation coefficients of all key performance parameter sequences between node i and node j. The edge weights between node i and node j are obtained by weighting and adding the pipeline topology association parameters and the sewage hydraulic coupling parameters. Construct an initial node matrix, and then map the edge weights between each node to the corresponding positions in the initial node matrix to obtain the correlation matrix.

7. The method for evaluating the design efficiency of septic tanks based on big data according to claim 1, characterized in that, The method for constructing the set of node performance feature matrices is as follows: Obtain the key performance parameters of each qualified solution at each node under different working conditions, and concatenate them according to the order of working conditions and the order of key performance parameters to obtain a one-dimensional feature vector of each qualified solution at each node. Traverse the pool of qualified solutions for each node, and select one design solution from each node's pool to form a combination of regional design solutions. Eliminate invalid regional design solution combinations using preset pruning rules to obtain a set of valid solution combinations. For any valid solution combination in the set of valid solution combinations, sort the one-dimensional feature vectors corresponding to the design solutions selected by each node in the corresponding valid solution combination according to the node order to generate the node performance feature matrix of the corresponding valid solution combination. Based on the node performance feature moments of all valid solution combinations, construct a set of node performance feature moments.

8. The method for evaluating the design efficiency of septic tanks based on big data according to claim 7, characterized in that, The method by which the neural network model outputs a set of global performance predictions for different combinations of effective solutions is as follows: The correlation matrix and the set of node performance feature matrices are input into the pre-built neural network model. The correlation matrix is ​​used to perform graph convolution feature transfer and aggregation operations on the set of node performance feature matrices, and the deep feature matrices of each node are output. Based on the deep feature matrix of each node, feature mapping and parsing are performed node by node through the built-in fully connected decoding layer to obtain the predicted values ​​of multiple key performance parameters of each node under multiple working conditions for each effective scheme combination. The predicted values ​​of multiple key performance parameters of all nodes under the same working condition are weighted and accumulated to obtain the local performance index value of each effective scheme combination under a single working condition. By iterating through all simulated operating conditions, calculating the local performance index value corresponding to each operating condition, and integrating and arranging the local performance index values ​​corresponding to all operating conditions in chronological order, a global performance prediction set of the corresponding effective scheme combination is obtained.

9. The method for evaluating the design efficiency of septic tanks based on big data according to claim 8, characterized in that, The method for selecting the optimal design scheme for the septic tank to be evaluated is as follows: The local efficiency index values ​​corresponding to each working condition within the global efficiency prediction set of the effective scheme combination are weighted and accumulated to obtain the comprehensive energy efficiency score of each effective scheme combination. Based on the comprehensive energy efficiency score, the effective scheme combinations are sorted in descending order to obtain the effective scheme combination ranking table. The design scheme corresponding to the node of the septic tank to be evaluated within the top-ranked effective scheme combination in the effective scheme combination ranking table is selected as the optimal design scheme.

10. A septic tank simulation design performance evaluation system based on big data, implementing the septic tank simulation design performance evaluation method based on big data as described in any one of claims 1-9, characterized in that, The system includes: The digital simulation modeling module is used to establish a corresponding digital simulation model for each septic tank to be evaluated within the target area, and to set corresponding design parameters for each septic tank to be evaluated in the digital simulation model based on the predetermined design scheme. The simulation operation module is used to simulate different steady-state operating conditions, variable operating conditions and long-term operating conditions. Based on the design parameters, it performs CFD and biochemical reaction coupled simulation on the digital simulation model, and acquires the key performance parameters generated by each septic tank to be evaluated in real time during the simulation operation. The qualified design scheme evaluation module is used to obtain key performance benchmark information, combine key performance parameters to generate performance evaluation coefficients for each septic tank to be evaluated under each corresponding design scheme, evaluate the qualification of each design scheme in turn based on the performance evaluation coefficients, and build a qualified design scheme pool for each septic tank to be evaluated based on all qualified design schemes. The correlation matrix construction module is used to treat each septic tank to be evaluated in the target area as a node, calculate the pipe network topology correlation parameters and sewage hydraulic coupling parameters between any two nodes, and obtain the edge weights through fusion analysis, and construct the correlation matrix based on the edge weights; The feature matrix construction module is used to traverse the pool of qualified solutions for each node, filter out effective solution combinations, generate node performance feature matrices under different effective solution combinations, and construct a set of node performance feature matrices based on the node performance feature matrices. The optimal solution selection module is used to input the correlation matrix and the set of node performance feature matrices into a pre-built neural network model. The neural network model outputs a global performance prediction set of different effective solution combinations. Based on the global performance prediction set, the performance of all effective solution combinations is evaluated, and the optimal design scheme for each septic tank to be evaluated is selected based on the evaluation results.

Images