Intelligent decision-making system for sewage whole-process technological parameters

Abstract

The invention discloses an intelligent decision-making system for full-flow process parameters of sewage, and belongs to the technical field of sewage treatment. Comprising a monitoring module connected with a sewage plant and used for monitoring inflow historical data of the sewage plant; the water inlet prediction module is connected with the monitoring module and predicts to obtain water inlet prediction data; the lifting operation optimization module is connected with the water inlet prediction module and used for calculating the lifting amount of the efficiency of the pump set in the next time period; the aeration intelligent decision-making module is connected with the water inlet prediction module and is used for calculating an air demand set value; the intelligent carbon source adding module is connected with the water inlet prediction module and calculates the optimal adding amount of the carbon source; and the intelligent phosphorus removal agent adding module is connected with the water inlet prediction module and is used for calculating the optimal adding amount of the phosphorus removal agent. The technical scheme has the beneficial effects that the future incoming water condition of the sewage plant can be accurately predicted, a data basis is provided for subsequent sewage treatment, pump set efficiency optimization, aeration optimization and carbon source and phosphorus removal agent dosage optimization are realized, and the efficiency and accuracy of the whole water treatment process are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of sewage treatment, and in particular to a process parameter intelligent decision-making system. Background Art

[0002] With the increasingly stringent environmental protection requirements and the development of the sewage treatment industry, municipal sewage plants with activated sludge as the core treatment process play a key role in urban sewage treatment. As a major carbon emitter in China, the sewage treatment industry includes direct and indirect carbon emissions, and its carbon emissions account for 1% to 2% of the total emissions of the whole society. In the process of indirect greenhouse gas emissions, it mainly comes from indirect greenhouse gas emissions generated by electricity consumption, agent consumption, and other auxiliary equipment. Among them, 70% to 90% of indirect greenhouse gas emissions come from electricity consumption. In the operation of the main equipment, the energy-consuming sewage treatment method is adopted. The aeration, sewage lifting and sludge treatment links consume the most electricity, accounting for about 80% of the total energy consumption of the sewage treatment plant.

[0003] There are many shortcomings in the traditional full-process sewage treatment operation and management methods. In terms of influent water quality and quantity prediction, there is a lack of effective data processing and prediction, making it difficult to accurately predict future water inflow conditions, resulting in a lack of scientific basis for subsequent process parameter adjustments; in terms of lifting operation, the operation strategy of the lifting pump group cannot be optimized in real time according to changes in influent water volume, which easily leads to energy waste of the pump group and low efficiency in the coordinated use of multiple pumps; in the aeration link, the air demand cannot be accurately adjusted dynamically according to the influent water quality, biochemical system process parameters and effluent water quality, resulting in excessive aeration energy consumption or insufficient aeration affecting the treatment effect; for the addition of carbon sources and phosphorus removers, unreasonable dosages often occur, which increases treatment costs and may cause effluent water quality to substandard. Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent decision-making system for process parameters of the entire sewage process to solve the above technical problems;

[0005] An intelligent decision-making system for sewage full-process process parameters, including:

[0006] A monitoring module, connected to a sewage treatment plant, for monitoring the historical data of water inflow and the quality of effluent from the sewage treatment plant;

[0007] A water inflow prediction module, connected to the monitoring module, for receiving and obtaining water inflow prediction data based on the water inflow historical data;

[0008] A lifting operation optimization module, connected to the water inflow prediction module, for receiving and calculating the increase in pump group efficiency in the next time period based on the water inflow prediction data;

[0009] an aeration intelligent decision-making module, connected to the water inlet prediction module and the monitoring module, for establishing an aeration optimization model according to the water inlet prediction data, and calculating a required air volume set value according to the aeration optimization model;

[0010] A carbon source intelligent dosing module, connected to the water inlet prediction module and the monitoring module, for establishing a carbon source dosing optimization model according to the water inlet prediction data and the water inlet historical data, and calculating the optimal carbon source dosing amount according to the carbon source dosing optimization model;

[0011] The intelligent dephosphorization agent dosing module is connected to the water inlet prediction module and the monitoring module, and is used to establish a dephosphorization agent dosing optimization model based on the water inlet prediction data and the water inlet historical data, and calculate the optimal dephosphorization agent dosage based on the dephosphorization agent dosing optimization model.

[0012] Preferably, the water inflow prediction module includes:

[0013] A first data collection unit, connected to the monitoring module, for collecting water inflow flow and water quality data in the water inflow history data to obtain a first data set;

[0014] A data preprocessing unit, connected to the data collection unit, for performing an outlier elimination process and a missing value compensation process on the water inflow history data to obtain preprocessed data;

[0015] The first model building unit is connected to the data preprocessing unit and is used to build a water inlet data prediction model according to the preprocessed data, and obtain the water inlet prediction data according to the water inlet data prediction model.

[0016] Preferably, the outlier elimination process includes:

[0017] Randomly divide the first data set based on the abnormal data identification model, calculate the abnormal value scores of the first data set, and remove the abnormal value scores exceeding a set value;

[0018] The calculation formula of the outlier score is:

[0019]

[0020] Among them, S(h ij ,u) represents the outlier score;

[0021] h ij represents a data point in the first data set, i and j are index identifiers of the data point in the first data set;

[0022] u represents the number of samples in the first data set;

[0023] E(hij ) means h ij an average path length in the anomaly data identification model;

[0024] C(u) represents the correction function;

[0025] The missing value compensation process includes:

[0026] Linear interpolation is performed on the data matrix after the outlier elimination process, and a filling matrix including target compensation variables is obtained through matrix transformation. The filling matrix is ​​then predicted through a regression model, and the compensation value is obtained by taking the average of the prediction results.

[0027] Preferably, the lifting operation optimization module includes:

[0028] A second data collection unit, connected to the water inflow prediction module, is used to collect the water inflow prediction data and obtain a second data set by combining the capacity of the booster pump pool and the liquid level information of the biochemical pool;

[0029] A second model building unit is connected to the second data collection unit, and builds a pump group optimization model based on the second data set and taking the minimum energy consumption of the pump group as the objective function;

[0030] A lifting optimization unit is connected to the second model building unit and the second data collection unit, and is used to calculate the lifting amount of the pump group in the next time period according to the second data set and the pump group optimization model, and generate an optimal operation strategy for the pump group.

[0031] Preferably, the calculation formula of the pump group efficiency is:

[0032]

[0033] Wherein, η represents the efficiency of the pump group;

[0034] P et Indicates the effective power of the pump set;

[0035] P at Indicates the input power of the pump set;

[0036] ρ represents the density of the liquid;

[0037] g represents the acceleration due to gravity;

[0038] Q t represents the pump flow rate of the pump group;

[0039] H t Indicates the pump head of the pump unit;

[0040] n represents the total number of pump groups;

[0041] P ai Indicates the input power of each pump in the pump group.

[0042] Preferably, the aeration intelligent decision-making module includes:

[0043] A third data collection unit, connected to the influent prediction module and the monitoring module, for collecting and integrating process indicators, influent water quality data and effluent water quality in the influent historical data to obtain a third data set;

[0044] a third model building unit, connected to the third data collection unit, and establishing the aeration optimization model according to the third data set;

[0045] An aeration optimization unit is connected to the third model building unit, calculates the air demand setting value based on the aeration optimization model, and controls the flow of the blower according to the air demand setting value.

[0046] Preferably, the aeration optimization model in the third model building unit uses a neural network model to obtain the maximum value of the nonlinear curve, and adopts a particle swarm optimization algorithm with inertia weight to obtain the optimal setting value of the operating parameter by minimizing the evaluation index, wherein the particle velocity and position iteration formulas are respectively,

[0047] v i,d (t+1)=wv i,d (t+1)+c 1 r 1 [p i,d -x i,d (t)]+c 2 r 2 [p g,d -x i,d (t)];

[0048] x i,d (t+1)=x i,d (t)+v i,d (t+1);

[0049] 1≤i≤N,1≤d≤D;

[0050] Among them, v i,d (t+1) represents the velocity of the i-th particle in the d-th dimension at time t+1;

[0051] w represents the inertia weight;

[0052] c 1 、c 2 represents the learning factor;

[0053] r 1 、r 2Represents a random number between 0 and 1;

[0054] N represents the number of particles;

[0055] p i,d represents the historical optimal position of the i-th particle in the d-th dimension;

[0056] x i,d (t) represents the position of the i-th particle in the d-th dimension at time t;

[0057] p g,d represents the global optimal position of all particles in the dth dimension;

[0058] D represents the number of solutions;

[0059] The expression of the inertia weight for solving the minimum fitness function is,

[0060]

[0061] in, The inertia weight representing the minimum fitness function;

[0062] w min Indicates the preset minimum inertia coefficient;

[0063] w max Indicates the preset maximum inertia coefficient;

[0064] represents the fitness value of the i-th particle in the d-th dimension;

[0065] Represents the minimum fitness value of all particles in the dth dimension at a certain iteration;

[0066] Represents the average fitness value of all particles in the dth dimension at a certain iteration.

[0067] Preferably, the carbon source intelligent dosing module comprises:

[0068] a fourth data collection unit, connected to the water inlet prediction module and the monitoring module, for collecting water quality in the water inlet prediction data, water quality data in the water inlet historical data, and process indicators to obtain a fourth data set;

[0069] The fourth model building unit is connected to the fourth data collection unit and is used to establish the carbon source addition optimization model according to the fourth data set, and calculate the optimal carbon source addition amount according to the carbon source addition optimization model.

[0070] Preferably, the dephosphorization agent intelligent dosing module comprises:

[0071] a fifth data collection unit connected to the water inflow prediction module and the monitoring module, collecting the water inflow prediction data and the dosage of the reagent and the process index in the water inflow history data to obtain a fifth data set;

[0072] The fifth model building unit is connected to the fifth data collection unit, establishes the dephosphorization agent dosage optimization model according to the fifth data set, and calculates the optimal dosage of the dephosphorization agent according to the dephosphorization agent dosage optimization model.

[0073] Preferably, it also includes:

[0074] A reflow intelligent decision-making module, connected to the monitoring module, is used to calculate the internal reflow amount according to the anoxic tank dissolved oxygen monitoring value and sludge concentration monitoring value of the process indicators in the influent historical data, the effluent total nitrogen monitoring value and the influent flow rate;

[0075] The sludge discharge intelligent decision-making module is connected to the monitoring module and is used to calculate the sludge discharge flow rate according to the biochemical pool sludge concentration and sludge age of the process indicators in the water inlet historical data.

[0076] The beneficial effects of the present invention are: it can accurately predict the future water inflow of the sewage treatment plant, provide a data basis for subsequent sewage treatment, achieve pump group efficiency optimization, aeration optimization, carbon source and phosphorus removal agent dosage optimization, and improve the efficiency and accuracy of the overall water treatment process. BRIEF DESCRIPTION OF THE DRAWINGS

[0077] Figure 1 It is a schematic diagram of the intelligent decision-making system and index system of the whole process process parameters of sewage in the present invention;

[0078] Figure 2 is a schematic diagram of a water inflow prediction module of the present invention;

[0079] Figure 3 is a schematic diagram of the lifting operation optimization module of the present invention;

[0080] Figure 4 is a schematic diagram of the aeration intelligent decision-making module of the present invention;

[0081] Figure 5 is a schematic diagram of the carbon source intelligent dosing module of the present invention;

[0082] Figure 6 It is a schematic diagram of the phosphorus removal agent intelligent dosing module of the present invention;

[0083] Figure 7 It is a process flow chart of the present invention.

[0084] In the attached drawings: 1. Monitoring module; 2. Water inlet prediction module; 21. First data collection unit; 22. Data preprocessing unit; 23. First model building unit; 3. Lifting operation optimization module; 31. Second data collection unit; 32. Second model building unit; 33. Lifting optimization unit; 4. Aeration intelligent decision module; 41. Third data collection unit; 42. Third model building unit; 43. Aeration optimization unit; 5. Carbon source intelligent dosing module; 51. Fourth data collection unit; 52. Fourth model building unit; 6. Phosphorus removal agent intelligent dosing module; 61. Fifth data collection unit; 62. Fifth model building unit; 7. Reflux intelligent decision module; 8. Sludge discharge intelligent decision module. DETAILED DESCRIPTION

[0085] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0086] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0087] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0088] An intelligent decision-making system for process parameters of the entire sewage process, such as Figure 1 As shown, including,

[0089] Monitoring module 1, connected to the sewage treatment plant, is used to monitor the historical data of the sewage treatment plant's influent and the quality of the effluent;

[0090] The water inflow prediction module 2 is connected to the monitoring module 1 and is used to receive and obtain water inflow prediction data based on the water inflow historical data;

[0091] The lifting operation optimization module 3 is connected to the water inflow prediction module 2, and is used to receive and calculate the improvement of the pump group efficiency in the next time period based on the water inflow prediction data;

[0092] The aeration intelligent decision-making module 4 is connected to the water inlet prediction module 2 and the monitoring module 1, and is used to establish an aeration optimization model according to the water inlet prediction data, and calculate the air demand set value according to the aeration optimization model;

[0093] The carbon source intelligent dosing module 5 is connected to the water inlet prediction module 2 and the monitoring module 1, and is used to establish a carbon source dosing optimization model according to the water inlet prediction data and the water inlet historical data, and calculate the optimal carbon source dosage according to the carbon source dosing optimization model;

[0094] The dephosphorization agent intelligent dosing module 6 is connected to the water inlet prediction module 2 and the monitoring module 1, and is used to establish a dephosphorization agent dosing optimization model based on the water inlet prediction data and the water inlet historical data, and calculate the optimal dephosphorization agent dosage based on the dephosphorization agent dosing optimization model.

[0095] Specifically, the present invention provides an intelligent decision-making system for process parameters of the entire sewage process. Through the monitoring module 1, the future water inflow of the sewage treatment plant can be accurately predicted, providing a data basis for subsequent sewage treatment. The pump group efficiency is optimized by improving the operation optimization module 3, aeration optimization is achieved through the aeration intelligent decision-making module 4, and the carbon source and dephosphorization agent dosages are optimized through the carbon source intelligent addition module 5 and the dephosphorization agent intelligent addition module 6, thereby improving the efficiency and accuracy of the overall water treatment process.

[0096] In a preferred embodiment, referring to Figure 2 , water inflow prediction module 2 includes,

[0097] A first data collection unit 21, connected to the monitoring module 1, is used to collect water inflow flow and water quality data in the water inflow history data to obtain a first data set;

[0098] The data preprocessing unit 22 is connected to the data collection unit and is used to perform abnormal value elimination and missing value compensation processing on the historical water inflow data to obtain preprocessed data;

[0099] The first model building unit 23 is connected to the data preprocessing unit 22 and is used to build a water inflow data prediction model according to the preprocessed data, and obtain water inflow prediction data according to the water inflow data prediction model.

[0100] Specifically, based on monitoring module 1, the historical influent data of the sewage treatment plant including flow, COD (chemical oxygen demand), ammonia nitrogen, TN (total nitrogen), TP (total phosphorus), SS (suspended solids), etc. are collected to establish an influent water quantity and water quality database; after preprocessing the data, a machine learning algorithm is used to establish a water quantity and water quality prediction model to predict the influent water quantity and water quality of the sewage treatment plant in the next 24 hours.

[0101] The influent water quality and quantity prediction establishes a database based on the influent historical data, and after data preprocessing by removing outliers and compensating for missing values, it is divided into a training set and a test set, with a ratio of 7:3. The machine learning algorithm with LSTM (Long Short-Term Memory Network) as the core is used to establish the influent water quantity and water quality prediction model. The influent historical data includes influent flow, water quality data, dosage of chemicals, process indicators and effluent water quality;

[0102] By predicting the incoming water volume and water quality in the next 24 hours, we can know the changing trend of water inflow in advance. For example, before the peak of water volume comes, we can adjust the operation strategy of the booster pump in advance to ensure sufficient processing capacity to cope with it; according to different water qualities, we can prepare appropriate processing parameters in advance to ensure stable treatment effect, avoid substandard treatment due to sudden changes in water inflow, and improve overall treatment efficiency.

[0103] Accurate water inflow prediction helps sewage treatment plants to rationally arrange energy and chemical use. When high-concentration organic wastewater is predicted to flow in, the aeration system is adjusted in advance to increase the aeration volume to meet the needs of microbial degradation and avoid energy waste; at the same time, according to the phosphorus content prediction, the amount of phosphorus removal agent is accurately controlled to reduce excessive consumption of chemicals and reduce operating costs.

[0104] The water inflow forecast data provides a basis for the coordinated operation of each treatment unit in the sewage plant. The operation optimization of the lifting pump can be carried out based on the water volume forecast; the intelligent decision-making of aeration, carbon source and phosphorus removal agent addition can also be combined with the water quality forecast data to achieve effective connection and coordination of each unit, breaking the traditional situation of independent operation of each section and improving the overall efficiency of the system.

[0105] Effectively respond to fluctuations in influent volume and quality, predict influent changes in advance, and enable sewage treatment plants to take timely response measures, such as adjusting process parameters, initiating emergency plans, etc., to reduce the impact of sudden changes in water quality and volume on the treatment system, ensure stable operation of sewage treatment plants, and reduce operating risks.

[0106] The prediction model built with LSTM as the core algorithm, combined with a large amount of historical data and scientific data preprocessing methods, has a high prediction accuracy. At the same time, the model can be continuously optimized according to new data, providing scientific decision-making support for the operation and management of sewage treatment plants, helping sewage treatment plants to continuously improve treatment processes and enhance management levels.

[0107] In a preferred embodiment, the outlier removal process includes:

[0108] Randomly divide the first data set based on the abnormal data identification model, calculate the abnormal value score of the first data set, and remove the abnormal value scores exceeding the set value;

[0109] The outlier score is calculated as:

[0110]

[0111] Among them, S(h ij ,u) represents the outlier score;

[0112] h ij represents a data point in the first data set, i and j are index identifiers of the data point in the first data set;

[0113] u represents the number of samples in the first data set;

[0114] E(h ij ) means h ij Average path length in anomaly data identification model;

[0115] C(u) represents the correction function;

[0116] Missing value compensation processing includes:

[0117] Linear interpolation is performed on the data matrix after outlier removal, and the filling matrix including the target compensation variable is obtained through matrix transformation. The filling matrix is ​​then predicted through the regression model, and the compensation value is obtained by taking the average of the prediction results.

[0118] Specifically, outlier removal is based on the abnormal data identification model of the isolation forest, which can quickly and accurately identify abnormal data. First, the known and continuous time data set is randomly divided, and then the difference between abnormal data and normal data is utilized. After the data is calculated by the IF (lonely forest) model, different high and low density areas are formed. The data outlier score is calculated to reflect the density area where the data is located, and the data with high scores are eliminated. The model output x ij The path length is h ij , x ij For the elements in the matrix X, through h ij jCalculate x ij The outlier score.

[0119] Missing value compensation performs linear interpolation on X after removing abnormal data, and then obtains the filling matrix containing the target compensation variables through matrix transformation; then the filling matrix is ​​predicted by the RF (random forest) regression model, and the output is output using the integration idea, and the mean of the results is taken as the compensation value, which improves the accuracy of the output results and completes the compensation of mixed type missing data.

[0120] In a preferred embodiment, referring to Figure 3 , the operation optimization module 3 includes:

[0121] The second data collection unit 31 is connected to the water inflow prediction module 2 and is used to collect water inflow prediction data and obtain a second data set by combining the capacity of the booster pump pool and the liquid level information of the biochemical pool;

[0122] The second model building unit 32 is connected to the second data collection unit 31, and builds a pump group optimization model based on the second data set and taking the minimum energy consumption of the pump group as the objective function;

[0123] The lifting optimization unit 33 is connected to the second model building unit 32 and the second data collection unit 31, and is used to calculate the lifting amount of the pump group in the next time period according to the second data set and the pump group optimization model, and generate an optimal operation strategy for the pump group.

[0124] Specifically, the lifting operation optimization is based on the prediction of the water inlet volume. Combined with the capacity of the lifting pump pool and the liquid level of the biochemical pool, the lifting volume of the lifting pump in the next period is calculated. Taking the minimum energy consumption of the pump group as the objective function, an intelligent lifting optimization model is established to obtain the optimal operation strategy of the lifting pump group, realize the coordinated use of multiple pumps, make the water pump operate in an efficient working area, and reduce energy waste.

[0125] The lifting operation optimization strategy is based on the lifting volume in the next period obtained by predicting the water inlet volume, combined with the water pump performance parameters, and the change law of pump group efficiency under different flow rates, different pump group combinations and different lifting pump pool liquid levels during historical operation, so as to establish a pump group optimization model and obtain the lifting operation optimization strategy.

[0126] An optimization model is established with the minimum energy consumption of the pump group as the objective function to accurately calculate the lifting volume of the pump group in the next time period, so that the water pump operates in the efficient working area. This avoids unnecessary energy consumption, reduces carbon emissions generated by the operation of the lifting pump in the sewage plant, meets the goal of energy conservation and carbon reduction, and reduces operating costs.

[0127] Combined with the water inflow forecast data, pump pool capacity and biochemical pool level information, the operation strategy is formulated to ensure that the pump group lifting capacity matches the actual treatment needs of the sewage plant. Avoid inefficient operation states such as idling and overloading of the pump group, extend the service life of the equipment, ensure that the sewage plant can continuously and stably treat sewage, and improve the overall treatment capacity.

[0128] Closely connected with the water inflow prediction module 2, data sharing and collaboration are achieved. The pump group operation is adjusted according to the water inflow prediction to better connect the lifting link with the subsequent sewage treatment link, avoid uneven load of subsequent treatment units due to unreasonable lifting volume, and improve the overall coordination and operation efficiency of the sewage treatment system.

[0129] Adapt to changes in water quality and quantity, improve stability, adjust operation strategies in real time based on the latest data, and quickly adapt to dynamic changes in water quality and quantity of sewage treatment plants. In the face of sudden increases in water volume or fluctuations in water quality, timely optimize pump operation, maintain stable operation of sewage treatment plants, and reduce the impact of changes in water quality and quantity on treatment effects.

[0130] By building models and generating optimal operation strategies, we provide scientific and quantitative decision-making basis for sewage treatment plant managers. We change the previous extensive management mode that relies on manual experience, realize the refined management of pump operation, and improve the overall operation and management level of sewage treatment plants.

[0131] In a preferred embodiment, the calculation formula for the pump group efficiency is:

[0132]

[0133] Among them, η represents the efficiency of the pump group;

[0134] P et Indicates the effective power of the pump set;

[0135] P at Indicates the input power of the pump set;

[0136] ρ represents the density of the liquid;

[0137] g represents the acceleration due to gravity;

[0138] Q t Indicates the pump flow rate of the pump group;

[0139] H t Indicates the pump head of the pump unit;

[0140] n represents the total number of pump sets;

[0141] P ai Indicates the input power of each pump in the pump group.

[0142] In a preferred embodiment, referring to Figure 4 , aeration intelligent decision module 4 includes,

[0143] The third data collection unit 41 is connected to the influent prediction module 2 and the monitoring module 1, and is used to collect and integrate the process indicators, influent water quality data and effluent water quality in the influent historical data to obtain a third data set;

[0144] A third model building unit 42, connected to the third data collection unit 41, establishes an aeration optimization model according to the third data set;

[0145] The aeration optimization unit 43 is connected to the third model building unit 42, calculates the air demand setting value based on the aeration optimization model, and controls the flow of the blower according to the air demand setting value.

[0146] Specifically, the aeration intelligent decision-making establishes an aeration optimization model based on the inlet water quality data, biochemical system process parameters and outlet water quality data, finds the optimal solution with energy consumption as the objective function, calculates the air demand set value, and thus realizes the optimization decision of the flow control of the blower.

[0147] In a preferred embodiment, the aeration optimization model in the third model building unit 42 uses a neural network model to obtain the maximum value of the nonlinear curve, and uses a particle swarm optimization algorithm with inertia weight to obtain the optimal setting value of the operating parameter by minimizing the evaluation index, wherein the particle velocity and position iteration formulas are respectively,

[0148] vi,d (t+1)=wv i,d (t+1)+c 1 r 1 [p i,d -x i,d (t)]+c 2 r 2 [p g,d -x i,d (t)];

[0149] x i,d (t+1)=x i,d (t)+v i,d (t+1);

[0150] 1≤i≤N,1≤d≤D;

[0151] Among them, v i,d (t+1) represents the velocity of the i-th particle in the d-th dimension at time t+1;

[0152] w represents the inertia weight;

[0153] c 1 、c 2 represents the learning factor;

[0154] r 1 、r 2 Represents a random number between 0 and 1;

[0155] N represents the number of particles;

[0156] p i,d represents the historical optimal position of the i-th particle in the d-th dimension;

[0157] x i,d (t) represents the position of the i-th particle in the d-th dimension at time t;

[0158] p g,d represents the global optimal position of all particles in the dth dimension;

[0159] D represents the number of solutions;

[0160] The expression for the inertia weight to solve the minimum fitness function is,

[0161]

[0162] in, Inertia weight representing the minimum fitness function;

[0163] w min Indicates the preset minimum inertia coefficient;

[0164] wmax Indicates the preset maximum inertia coefficient;

[0165] represents the fitness value of the i-th particle in the d-th dimension;

[0166] Represents the minimum fitness value of all particles in the dth dimension at a certain iteration;

[0167] Represents the average fitness value of all particles in the dth dimension at a certain iteration.

[0168] Specifically, the aeration optimization model uses the current inlet water, outlet water, and dissolved oxygen as neural network inputs, and the outlet water quality index with energy consumption penalty after 2 hours as the prediction target to train the neural network model. The neural network model is used to obtain the maximum value of the nonlinear curve, and the particle swarm optimization algorithm with inertia weight is used to obtain the optimal setting value of the operating parameters by minimizing the evaluation index.

[0169] In a preferred embodiment, referring to Figure 5 , the carbon source intelligent dosing module 5 includes,

[0170] The fourth data collection unit 51 is connected to the water inflow prediction module 2 and the monitoring module 1, and is used to collect water quality in the water inflow prediction data, water quality data in the water inflow historical data, and process indicators to obtain a fourth data set;

[0171] The fourth model building unit 52 is connected to the fourth data collection unit 51 and is used to establish a carbon source addition optimization model according to the fourth data set, and calculate the optimal carbon source addition amount according to the carbon source addition optimization model.

[0172] Specifically, the intelligent decision on carbon source addition analyzes and predicts the influent C / N based on the influent water quality monitoring data and the incoming water quality prediction data, and uses machine learning to establish a carbon source addition optimization model based on the secondary anoxic tank nitrate nitrogen monitoring data and the effluent total nitrogen monitoring data to give the optimal carbon source dosage decision.

[0173] Collect multi-source water quality data and process indicators to establish a carbon source addition optimization model, which can accurately calculate the optimal carbon source dosage. According to different influent C / N ratios and nitrate nitrogen, total nitrogen and other indicators, provide the carbon source amount that just meets the denitrification demand, enhance the denitrification reaction efficiency, improve the wastewater denitrification effect, and ensure that the total nitrogen in the effluent meets the discharge standards.

[0174] Avoid excessive or insufficient carbon source caused by traditional manual dosing based on experience. Excessive dosing causes waste of resources and increased costs, while insufficient dosing affects the denitrification effect. Accurate dosing can control the amount of carbon source within a reasonable range, reduce unnecessary carbon source consumption, reduce the cost of purchasing chemicals for sewage treatment plants, and improve economic benefits.

[0175] Dynamically adjust the carbon source dosage based on real-time and predicted water quality data to closely match the carbon source supply with the sewage denitrification demand. Stabilize the denitrification process, reduce abnormal microbial metabolism caused by carbon source fluctuations, ensure stable operation of the sewage treatment system, ensure that the effluent water quality is stable and meets the standards, and reduce the risk of water quality fluctuations.

[0176] Automated data collection and model calculation processes replace frequent manual testing and adjustments, reducing the workload of operators. With the help of intelligent means, it can quickly respond to changes in water quality, adjust carbon source addition strategies in a timely manner, improve the efficiency of sewage plant operation and management, and adapt to the development trend of intelligent modern sewage treatment.

[0177] The input parameters of the prediction model are influent COD, influent NH 4+ -N, influent TN, carbon source dosage and DO (dissolved oxygen), and the output is effluent COD, NH 4+ -N and TN. The prediction model is embedded into the particle swarm optimization (PSO) algorithm. PSO searches through an iteratively updated particle swarm. In each iteration, the particle updates itself by moving towards its previous best position (pbest) and the global best position (gbest) in the swarm. The particle velocity and position iteration formulas are:

[0178] v i =ωv i +c 3 r 1 (pbest i -p i )+c 4 r 2 (gbest-p i );

[0179] p i =p i +v i ;

[0180] Among them, v i represents the velocity of the particle;

[0181] ω represents the inertia weight;

[0182] c 3 、c 4 represents the acceleration factor;

[0183] r 1 、r 2 Represents a random number between 0 and 1;

[0184] pbest i represents the historical optimal position of the i-th particle;

[0185] pi Represents the current position of the i-th particle;

[0186] gbest represents the global optimal position of all particles.

[0187] The particle indicates the dosage. The effluent TN is the main control parameter for carbon source addition. Therefore, during the optimization process, the effluent TN should meet the standard.

[0188] In a preferred embodiment, referring to Figure 6 , the phosphorus removal agent intelligent dosing module 6 includes:

[0189] The fifth data collection unit 61 is connected to the water inflow prediction module 2 and the monitoring module 1, and collects the water inflow prediction data and the dosage of the reagents and the process indicators in the water inflow history data to obtain a fifth data set;

[0190] The fifth model building unit 62 is connected to the fifth data collection unit 61, establishes a dephosphorization agent dosage optimization model according to the fifth data set, and calculates the optimal dosage of the dephosphorization agent according to the dephosphorization agent dosage optimization model.

[0191] Specifically, the intelligent decision on the addition of phosphorus removers analyzes and predicts the influent C / P (carbon-phosphorus ratio) based on the influent water quality monitoring data and the incoming water quality prediction data, and uses machine learning to establish a phosphorus remover addition optimization model based on the secondary aerobic tank total phosphorus monitoring data and the effluent total phosphorus monitoring data, and gives the optimal dosage decision for the phosphorus remover.

[0192] Intelligent decision-making for phosphorus removal agent addition is based on the influent water quality monitoring data and influent water quality prediction data to analyze and predict the influent C / P. Based on the historical dosing data, the secondary aerobic pool total phosphorus monitoring data and the effluent total phosphorus monitoring data, the LSTM algorithm is used to establish a phosphorus removal agent addition optimization model. The Keras framework is used to build a LSTM layer with 50 neurons and Tanh activation function. The loss function selects Mae (mean absolute error), the optimizer selects Adam (adaptive moment estimation), and the explained variance is used to judge the accuracy of the model test set. Thus, the optimal dosage decision of the phosphorus removal agent is given.

[0193] In a preferred embodiment, it also includes:

[0194] The reflow intelligent decision module 7 is connected to the monitoring module 1 and is used to calculate the internal reflow amount according to the anoxic tank dissolved oxygen monitoring value and sludge concentration monitoring value of the process indicators in the influent historical data, the effluent total nitrogen monitoring value and the influent flow rate;

[0195] The sludge discharge intelligent decision module 8 is connected to the monitoring module 1 and is used to calculate the sludge discharge flow rate according to the biochemical pool sludge concentration and sludge age of the process indicators in the historical water inflow data.

[0196] Specifically, the intelligent decision-making of reflow determines the internal reflow volume based on the monitoring values ​​of dissolved oxygen and sludge concentration in the anoxic tank, the monitoring values ​​of total nitrogen in the effluent, and the inlet flow rate. The internal reflow volume needs to meet the following conditions:

[0197] (1) 0.2mg / L≤ DO of anoxic pool≤0.5mg / L.

[0198] (2) The MLSS (mixed liquor suspended solids concentration) of the aerobic pool is between 3000mg / L and 5000mg / L.

[0199] (3) The internal reflux ratio ranges from 50% to 200%.

[0200] The intelligent sludge discharge decision determines the sludge dry matter discharge amount according to the sludge concentration and sludge age in the biochemical pool, and then determines the sludge discharge flow rate according to the sludge concentration.

[0201] The intelligent decision on sludge discharge is based on the MLSS of the biochemical pool being between 3000mg / L and 5000mg / L and the sludge age being between 25 and 35 days, to determine the sludge discharge flow rate.

[0202] Reference Figure 7 The present invention is applicable to the refined operation and management of municipal sewage treatment plants with activated sludge process (especially multi-stage AO anaerobic-anoxic-aerobic process, sewage treatment process of multi-stage AO process includes two-stage and above AO process, such as five-stage Batumfu process, etc.) as the core treatment process, and the process flow of the sewage treatment plant includes a grid tank, a lifting pump tank, a primary sedimentation tank, a primary anaerobic tank, a primary anoxic tank, a primary aerobic tank, a secondary anoxic tank, a secondary aerobic tank, a secondary sedimentation tank, a high-efficiency sedimentation tank, an aerobic filter, a disinfection tank and a water outlet well for water outlet;

[0203] The grid tank, the lifting pump tank, the primary sedimentation tank, the primary anaerobic tank, the primary anoxic tank, the primary aerobic tank, the secondary anoxic tank, the secondary aerobic tank, the secondary sedimentation tank, the high-efficiency sedimentation tank, the aerobic filter tank, the disinfection tank and the outlet well for water discharge are connected in sequence according to the sewage treatment process;

[0204] The secondary sedimentation tank is connected to the primary anaerobic tank for sludge return, the secondary aerobic tank is connected to the primary anoxic tank for nitrification liquid return, and the secondary sedimentation tank is connected to the high-efficiency sedimentation tank, aerobic filter tank, and disinfection tank for sludge discharge. Dosing includes carbon source addition and phosphorus removal agent addition. The carbon source addition point is located at the water inlet of the secondary anoxic tank, and the phosphorus removal agent addition point is located at the water outlet of the secondary aerobic tank.

[0205] The primary aerobic pool and the secondary aerobic pool can be divided into the front section, the middle section and the final section.

[0206] Reference Figure 1The monitoring content of monitoring module 1 includes inlet flow and water quality (COD, ammonia nitrogen, TN, TP, SS) monitoring, power monitoring (total power of the whole plant, power of lift pump group, power of aeration fan), agent dosage monitoring (carbon source and phosphorus removal agent flow), process index monitoring (DO, MLSS, ammonia nitrogen, nitrate nitrogen, total phosphorus, sludge return flow, gas flow, etc.) and effluent water quality (COD, ammonia nitrogen, TN, TP, SS).

[0207] The inlet flow and water quality monitoring is set up in the lift pump room to monitor the inlet water volume and water quality; the total power consumption of the whole plant is monitored in the sewage treatment plant's main distribution cabinet to monitor the total power consumption of the whole plant; the power monitoring of the lift pump group is set up in the lift pump group distribution cabinet to monitor the total power consumption of the lift pump group; the power monitoring of the aeration fan is set up in the aeration fan distribution cabinet to monitor the total power consumption of the aeration fan; the carbon source flow monitoring is set at the outlet of the carbon source dosing pump; the phosphorus removal agent flow monitoring is set at the outlet of the phosphorus removal agent dosing pump, and the process index monitoring: DO, MLSS, ammonia nitrogen and nitrate nitrogen are set at the middle section of the primary aerobic tank and the outlet of the secondary aerobic tank, the total phosphorus is set at the outlet of the secondary aerobic tank, the sludge return volume is set on the internal return and external return pipelines, and the gas flow monitoring is set at the aeration main pipe and the main pipe.

[0208] The intelligent decision-making system for the whole-process process parameters of sewage provided in this application introduces the prediction of influent water quality and quantity, and establishes an optimization algorithm by associating aeration, carbon source and phosphorus removal agent addition, backflow and sludge discharge with influent and effluent water and system operation conditions. It can provide a system optimization plan for the process parameters of the sewage treatment plant, optimize the treatment capacity and energy consumption of the sewage treatment plant, adjust the process parameters in real time according to water quality, improve sewage treatment efficiency and stability, and achieve energy saving and consumption reduction.

[0209] The present invention introduces the prediction of influent water quality and quantity to achieve the organic combination of lift pump group optimization, aeration optimization, carbon source and phosphorus removal agent dosage optimization, backflow and sludge discharge optimization, thus solving the problem that the various processes in traditional sewage treatment plants are independent of each other and lack of systematicness and synergy.

[0210] In the sewage treatment process, the present invention realizes the automatic adjustment of each treatment link through the coordinated control of the parameter decision motor and the flow control valve. The system can optimize the process and resource utilization based on data collection and intelligent analysis.

[0211] The parameter-determined motor drives the intelligent swirl impeller to form a vortex effect, enhance the sand settling effect, improve the sand settling efficiency, and reduce the residence time of sand in the water.

[0212] Through intelligent control, the system can automatically adjust the opening of the flow control valve to ensure that the water intake matches the processing capacity of the treatment unit, avoiding a decrease in treatment effect or equipment damage due to excessive flow.

[0213] The present invention is equipped with an intelligent monitoring module, which can monitor the operating status of each link and water quality changes in real time, and automatically make adaptive adjustments according to actual conditions.

[0214] The system can automatically analyze water quality data, adjust the operating status of the intelligent pretreatment unit and multi-stage process purification box, and improve the flexibility and accuracy of water treatment. The intelligent monitoring module enables the system to have self-diagnosis and automatic correction functions to ensure the stability and reliability of the treatment process.

[0215] The system collects key data such as water quality, flow rate and equipment status in real time, establishes an intelligent database, and achieves continuous improvement and optimization of the process through big data analysis.

[0216] All key data and control parameters are automatically stored, and process optimization reports are generated through intelligent algorithms to provide operators with scientific process parameter decision support and improve the efficiency and accuracy of the overall water treatment process.

[0217] The above description is only a preferred embodiment of the present invention, and does not limit the implementation mode and protection scope of the present invention. For those skilled in the art, it should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the description and illustrations of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent decision-making system for sewage full-process process parameters, characterized in that: include, A monitoring module, connected to a sewage treatment plant, for monitoring the historical data of water inflow and the quality of effluent from the sewage treatment plant; A water inflow prediction module, connected to the monitoring module, for receiving and obtaining water inflow prediction data based on the water inflow historical data; A lifting operation optimization module, connected to the water inflow prediction module, for receiving and calculating the increase in pump group efficiency in the next time period based on the water inflow prediction data; an aeration intelligent decision-making module, connected to the water inlet prediction module and the monitoring module, for establishing an aeration optimization model according to the water inlet prediction data, and calculating a required air volume set value according to the aeration optimization model; A carbon source intelligent dosing module, connected to the water inlet prediction module and the monitoring module, for establishing a carbon source dosing optimization model according to the water inlet prediction data and the water inlet historical data, and calculating the optimal carbon source dosage according to the carbon source dosing optimization model; The intelligent dephosphorization agent dosing module is connected to the water inlet prediction module and the monitoring module, and is used to establish a dephosphorization agent dosing optimization model based on the water inlet prediction data and the water inlet historical data, and calculate the optimal dephosphorization agent dosage based on the dephosphorization agent dosing optimization model.

2. The intelligent decision-making system for sewage full-process process parameters according to claim 1 is characterized in that: The water inflow prediction module includes: A first data collection unit, connected to the monitoring module, for collecting water inflow flow and water quality data in the water inflow history data to obtain a first data set; A data preprocessing unit, connected to the data collection unit, for performing an outlier elimination process and a missing value compensation process on the water inflow history data to obtain preprocessed data; The first model building unit is connected to the data preprocessing unit and is used to build a water inlet data prediction model according to the preprocessed data, and obtain the water inlet prediction data according to the water inlet data prediction model.

3. The intelligent decision-making system for sewage full-process process parameters according to claim 2 is characterized in that: The outlier elimination process includes: Randomly divide the first data set based on the abnormal data identification model, calculate the abnormal value scores of the first data set, and remove the abnormal value scores exceeding a set value; The calculation formula of the outlier score is: Among them, S(h ij ,u) represents the outlier score; h ij represents a data point in the first data set, i and j are index identifiers of the data point in the first data set; u represents the number of samples in the first data set; E(h ij ) means h ij an average path length in the anomaly data identification model; C(u) represents the correction function; The missing value compensation process includes: Linear interpolation is performed on the data matrix after the outlier elimination process, and a filling matrix including target compensation variables is obtained through matrix transformation. The filling matrix is ​​then predicted through a regression model, and the compensation value is obtained by taking the average of the prediction results.

4. The intelligent decision-making system for sewage full-process process parameters according to claim 1 is characterized in that: The lifting operation optimization module includes: A second data collection unit, connected to the water inflow prediction module, is used to collect the water inflow prediction data and obtain a second data set by combining the capacity of the booster pump pool and the liquid level information of the biochemical pool; A second model building unit is connected to the second data collection unit, and builds a pump group optimization model based on the second data set and taking the minimum energy consumption of the pump group as the objective function; A lifting optimization unit is connected to the second model building unit and the second data collection unit, and is used to calculate the lifting amount of the pump group in the next time period according to the second data set and the pump group optimization model, and generate an optimal operation strategy for the pump group.

5. The intelligent decision-making system for sewage full-process process parameters according to claim 1 is characterized in that: The calculation formula of the pump group efficiency is: Wherein, η represents the efficiency of the pump group; P et Indicates the effective power of the pump set; P at Indicates the input power of the pump set; ρ represents the density of the liquid; g represents the acceleration due to gravity; Q t represents the pump flow rate of the pump group; H t Indicates the pump head of the pump unit; n represents the total number of pump groups; P ai Indicates the input power of each pump in the pump group.

6. The intelligent decision-making system for sewage full-process process parameters according to claim 1 is characterized in that: The aeration intelligent decision-making module includes: A third data collection unit, connected to the influent prediction module and the monitoring module, for collecting and integrating process indicators, influent water quality data and effluent water quality in the influent historical data to obtain a third data set; a third model building unit, connected to the third data collection unit, and establishing the aeration optimization model according to the third data set; An aeration optimization unit is connected to the third model building unit, calculates the air demand setting value based on the aeration optimization model, and controls the flow of the blower according to the air demand setting value.

7. The intelligent decision-making system for sewage full-process process parameters according to claim 6 is characterized in that: The aeration optimization model in the third model building unit uses a neural network model to obtain the maximum value of the nonlinear curve, and uses a particle swarm optimization algorithm with inertia weight to obtain the optimal setting value of the operating parameter by minimizing the evaluation index, wherein the particle velocity and position iteration formulas are respectively, v i,d (t+1)=wv i,d (t+1)+c1r1[p i,d -x i,d (t)]+c2r2[p g,d -x i,d (t)]; x i,d (t+1)=x i,d (t)+v i,d (t+1); 1≤i≤N,1≤d≤D; Among them, v i,d (t+1) represents the velocity of the i-th particle in the d-th dimension at time t+1; w represents the inertia weight; c1 and c2 represent learning factors; r1 and r2 represent random numbers between 0 and 1; N represents the number of particles; p i,d represents the historical optimal position of the i-th particle in the d-th dimension; x i,d (t) represents the position of the i-th particle in the d-th dimension at time t; p g,d represents the global optimal position of all particles in the dth dimension; D represents the number of solutions; The expression of the inertia weight for solving the minimum fitness function is, in, The inertia weight representing the minimum fitness function; w min Indicates the preset minimum inertia coefficient; w max Indicates the preset maximum inertia coefficient; represents the fitness value of the i-th particle in the d-th dimension; Represents the minimum fitness value of all particles in the dth dimension at a certain iteration; Represents the average fitness value of all particles in the dth dimension at a certain iteration.

8. The intelligent decision-making system for sewage full-process process parameters according to claim 1 is characterized in that: The carbon source intelligent dosing module comprises: a fourth data collection unit, connected to the water inlet prediction module and the monitoring module, for collecting water quality in the water inlet prediction data, water quality data in the water inlet historical data, and process indicators to obtain a fourth data set; The fourth model building unit is connected to the fourth data collection unit and is used to establish the carbon source addition optimization model according to the fourth data set, and calculate the optimal carbon source addition amount according to the carbon source addition optimization model.

9. The intelligent decision-making system for sewage full-process process parameters according to claim 1 is characterized in that: The dephosphorization agent intelligent dosing module comprises: a fifth data collection unit connected to the water inflow prediction module and the monitoring module, collecting the water inflow prediction data and the dosage of the reagent and the process index in the water inflow history data to obtain a fifth data set; The fifth model building unit is connected to the fifth data collection unit, establishes the dephosphorization agent dosage optimization model according to the fifth data set, and calculates the optimal dosage of the dephosphorization agent according to the dephosphorization agent dosage optimization model.

10. The intelligent decision-making system for sewage full-process process parameters according to claim 1 is characterized in that: Also includes, A reflux intelligent decision-making module, connected to the monitoring module, is used to calculate the internal reflux amount according to the anoxic tank dissolved oxygen monitoring value and sludge concentration monitoring value of the process indicators in the influent historical data, the effluent total nitrogen monitoring value and the influent flow rate; The sludge discharge intelligent decision-making module is connected to the monitoring module and is used to calculate the sludge discharge flow rate according to the biochemical pool sludge concentration and sludge age of the process indicators in the water inlet historical data.

Images