Remote control system of wastewater treatment equipment based on internet of things

Abstract

The present application belongs to the technical field of wastewater treatment, and provides a remote control system for a wastewater treatment device based on the Internet of Things, which comprises: in view of the problems that traditional fixed DO concentration control cannot adapt to fluctuations in influent load, effluent is prone to be substandard or aeration energy consumption is high, and monitoring data is noisy and time series is dislocated, the scheme is as follows: a multidimensional real-time sensing network is constructed, flow, COD, BOD, ammonia nitrogen, water temperature, distributed DO and sludge activity sensors are deployed, discrete wavelet transform is used for noise reduction and data time series alignment; a three-layer load-DO-energy consumption dynamic correlation model is designed, the basic layer predicts the load trend through LSTM, the middle layer quantifies the DO demand by using a microbial metabolism model, and the optimization layer outputs the optimal DO set value by using a PPO algorithm; a dynamic decision system is developed, a DO interval is adaptively generated according to the load state, and the number and parameters of the fans are adjusted in linkage, the present application realizes dynamic adaptation of the load, guarantees that the effluent ammonia nitrogen is up to standard, reduces aeration energy consumption, and improves the remote control precision and equipment operation efficiency.

Description

Technical Field

[0001] This invention belongs to the field of wastewater treatment technology, specifically a remote control system for wastewater treatment equipment based on the Internet of Things. Background Technology

[0002] Existing wastewater treatment and control technologies have significant bottlenecks:

[0003] Firstly, the control mode is rigid. Most sewage treatment plants use a fixed DO concentration setpoint of 2.0 to 3.0 mg / L and rely on manual or PID adjustment of the aeration rate.

[0004] Because microbial metabolism has a strong nonlinear characteristic, when the influent load increases suddenly, insufficient DO can easily lead to excessive ammonia nitrogen in the effluent, while when the load drops suddenly, excessive aeration can cause energy waste—the energy consumption of the aeration system accounts for more than 50% of the total energy consumption of the plant, and the traditional control mode has higher energy consumption than the intelligent system.

[0005] Secondly, the monitoring dimensions are lacking. Traditional SCADA systems only collect basic parameters such as flow rate and COD, and lack monitoring of key indicators of sludge activity, which leads to inaccurate calculation of the microbial metabolic model (OUR) and makes it impossible to accurately guide aeration control.

[0006] Third, data quality defects. The raw sensor data suffers from noise interference and time-series misalignment issues. For example, there is a lag between the influent load data and the DO response. The existing system lacks an effective wavelet noise reduction and time-series alignment mechanism, which leads to a decrease in the accuracy of the prediction model.

[0007] At the control system level, traditional PLC and SCADA systems have inherent limitations: high expansion costs, difficulty in fault tracing, and inability to achieve global optimization.

[0008] Furthermore, existing technologies are not sufficiently adaptable to complex operating conditions. When parameters such as influent COD and flow rate fluctuate drastically, adjustment methods relying on human experience are inadequate and cannot meet the stringent requirements of the Class I discharge standard.

[0009] Therefore, developing a control system that can dynamically adapt to load fluctuations, integrate multi-dimensional sensing, and realize remote intelligent decision-making has become the key to breaking through existing technological bottlenecks and meeting the dual needs of environmental protection and energy conservation.

[0010] To this end, the present invention provides a remote control system for wastewater treatment equipment based on the Internet of Things. Summary of the Invention

[0011] In order to overcome the shortcomings of the prior art, at least one technical problem raised in the background art is solved.

[0012] The technical solution adopted by this invention to solve its technical problem is:

[0013] On one hand, the present invention provides a remote control system for wastewater treatment equipment based on the Internet of Things, including:

[0014] Multi-dimensional perception and data preprocessing module: Constructs a multi-dimensional real-time perception network, uses the multi-dimensional real-time perception network to collect multi-dimensional data, and performs noise reduction and time-series alignment processing on the collected multi-dimensional data;

[0015] The three-layer architecture load-DO-energy dynamic correlation model construction module is designed to establish a dynamic correlation model of load-DO concentration-energy consumption by designing a basic layer, an intermediate layer, and an optimization layer. The basic layer outputs the load trend over a period of time, the intermediate layer quantifies the DO concentration, and the optimization layer outputs the optimal DO concentration setting value that achieves effluent standards and minimizes aeration energy consumption.

[0016] Dynamic DO setting and aeration equipment linkage control module: Develop a dynamic setpoint decision system to achieve adaptive control from fixed DO concentration to DO concentration range. Based on the real-time load and the output of the load-DO concentration-energy consumption dynamic correlation model, dynamically generate the DO concentration setpoint range, and adjust the aeration equipment in linkage based on the dynamically generated DO concentration setpoint range.

[0017] As one embodiment, the specific process of constructing a multi-dimensional real-time sensing network is as follows:

[0018] Deploy flow sensors, COD sensors, BOD sensors, ammonia nitrogen sensors, and water temperature sensors at the water inlet end;

[0019] A distributed fiber optic DO sensor is installed every 5 meters in the aeration tank, and a new sludge activity sensor is added. The sludge activity sensor consists of an ATP-based probe for detecting microbial activity and an oxidation-reduction potential probe.

[0020] As one embodiment, the specific process of collecting multidimensional data using a multidimensional real-time sensing network is as follows:

[0021] The flow sensor is used to collect the influent flow rate, the COD sensor is used to collect the chemical oxygen demand (COD), the BOD sensor is used to collect the biochemical oxygen demand (BOD), the ammonia nitrogen sensor is used to collect the ammonia nitrogen concentration, the water temperature sensor is used to collect the influent temperature, the distributed fiber optic DO sensor collects the dissolved oxygen (DO) concentration in different areas of the aeration tank, and the sludge activity sensor is used to collect the sludge activity.

[0022] As one embodiment, the specific process of performing noise reduction and time-series alignment on the collected multidimensional data is as follows:

[0023] The time series data of influent flow rate and DO concentration in the collected data are denoised using Discrete Wavelet Transform (DWT). The steps are as follows:

[0024] Let the original signals of influent flow rate and DO concentration be x(n) (n=0,1,...,N-1), which are decomposed into approximate coefficients by passing through a low-pass filter h and a high-pass filter g. and detail coefficient :

[0025] Approximation coefficient: Detail factor: Where j is the decomposition level, k is the coefficient index, and h and g are the coefficients of the db4 wavelet filter;

[0026] For detail coefficients Apply Donoho threshold , Where σ is the noise standard deviation and N is the signal length. For symbolic functions, Thresholding shrinkage is applied to the absolute values ​​of the detail coefficients;

[0027] Combined with the processed approximation coefficients and detail coefficient The signal is recovered by inverse wavelet transform: ,in, , These are the coefficients of the inverse db4 wavelet filter. Let n be the signal reconstructed by inverse wavelet transform, and n be the sequence index of the signal. This represents the index offset of the filter coefficients;

[0028] Align the DO time series and calculate the water residence time. : Where L is the length of the aeration tank from the inlet to the DO monitoring point, v is the average flow velocity of the water, which is derived from the flow rate Q and the cross-sectional area S of the tank, S=w·h, where w is the width of the tank and h is the effective water depth;

[0029] pass Shift the timestamps of influent flow rate and DO concentration backward.

[0030] As one embodiment, the specific process by which the base layer outputs the load trend over a future period is as follows:

[0031] The base layer uses raw data collected by a multi-dimensional real-time sensing network. The raw data includes influent flow rate, COD, BOD, ammonia nitrogen concentration, and influent temperature after the timestamp is shifted backward. These data serve as influent load indicators. The raw data is processed by a Long Short-Term Memory (LSTM) network to predict the load trend over a future period and output the influent flow rate, COD, BOD, ammonia nitrogen concentration, and influent temperature for the future period.

[0032] As one embodiment, the specific process of quantifying DO concentration in the intermediate layer is as follows:

[0033] The intermediate layer constructs a microbial metabolic model, calculates the current actual oxygen consumption rate (OUR) of microorganisms based on sludge activity and ORP data, and quantifies DO demand;

[0034] Among them, ORP data is obtained by direct measurement through an ORP probe;

[0035] The formula for the microbial metabolic model is: ,in, This represents the actual oxygen consumption rate of microorganisms. The total oxygen transfer coefficient of the aeration system. This represents the saturation concentration of dissolved oxygen in the water under the current operating conditions. This represents the actual dissolved oxygen concentration in the aeration tank. The rate of change of dissolved oxygen concentration over time;

[0036] Higher sludge activity leads to higher oxygen utilization efficiency. The total oxygen transfer coefficient of the aeration system is corrected by adjusting the ATP concentration. : ,in, This is the corrected total oxygen transfer coefficient for the aeration system. Based on the actual activity of the current sludge, As a baseline sludge activity, To correct the index;

[0037] ORP mutations reflect an imbalance between DO supply and demand. An ORP compensation term is introduced to compensate for ORP; when ORP > 200mV, OUR increases by a factor of 1.2. .

[0038] As one embodiment, the specific process of the optimization layer outputting the optimal DO concentration setting value that achieves the standard for effluent and minimizes aeration energy consumption is as follows:

[0039] The optimization layer introduces a reinforcement learning algorithm with the dual objectives of achieving effluent standards and minimizing aeration energy consumption. It trains a DO setpoint decision model, and the action of the reinforcement learning algorithm is to output the DO concentration setpoint.

[0040] Design a reward function with the goals of achieving effluent quality standards and minimizing aeration energy consumption:

[0041] Piecewise function Based on the ammonia nitrogen concentration in the effluent Whether the standard is met or not, the threshold for meeting the standard is 5mg / L, and the reward is calculated in two cases:

[0042] When the standard is met The lower the concentration of hydrogen ammonia, the closer the reward is to 1;

[0043] When not meeting the standard The higher the ammonia nitrogen concentration, the greater the absolute value of the negative reward;

[0044] Aeration energy consumption bonus ( ):

[0045] With the lowest energy consumption in history Based on this benchmark, the lower the energy consumption, the higher the reward: ;

[0046] Weighted fusion dual objectives: , , These are weighting coefficients, and , E represents the effluent ammonia nitrogen concentration, and E represents the real-time aeration energy consumption.

[0047] The decision-making strategy for the DO setting value of the PPO update reinforcement learning algorithm is optimized through a proximal strategy. The core formula is the clipped objective function: ,in, For the policy network in state Down Output Action The probability, For the policy network in state Down Output Action The old probability, For the dominant function, The range of clip.

[0048] As one embodiment, the specific process of adapting from a fixed DO concentration to an adaptive DO concentration range is as follows:

[0049] To accurately determine the influent load status, a comprehensive load index L and its fluctuation are defined. ;

[0050] Comprehensive load index L: ;

[0051] volatility : ;

[0052] in, Chemical oxygen demand (COD) of the influent. This refers to the inlet water flow rate. The weights for influent chemical oxygen demand and influent flow rate are respectively, and , The current overall load, This represents the average total load over the past hour.

[0053] when At that time, the load fluctuation was in a stable state, maintaining a fixed DO concentration range;

[0054] when At that time, the load fluctuation was in a state of sudden increase, and the DO concentration range needed to be expanded;

[0055] when At that time, the load fluctuation was in a state of sudden drop, and the DO concentration range needed to be compressed.

[0056] As one embodiment, the specific process of dynamically generating the DO concentration setpoint range is as follows:

[0057] The optimal DO concentration value output based on the load-DO concentration-energy consumption dynamic correlation model Based on the load status, DO concentration ranges are dynamically generated. ;

[0058] When the load fluctuation is in a stable state, the interval width is evenly distributed, balancing accuracy and stability:

[0059] , The baseline value for steady-state control;

[0060] When the load fluctuates sharply, priority should be given to ensuring the lower limit of DO concentration, and the upper limit should be increased to meet the high load: , This serves as the baseline value for regulating a sudden increase in temperature.

[0061] When the load fluctuation is a sudden drop, the upper limit of compression reduces energy consumption, while maintaining the lower limit stability: , This is the baseline value for adjusting the sudden drop in temperature.

[0062] As one embodiment, the specific process of the linkage adjustment aeration device is as follows:

[0063] Based on the dynamically generated DO concentration setpoint range, the fan operating parameters are dynamically adjusted. When the load fluctuation is a sudden increase, the fan is allowed to operate at its maximum over-power limit. When the load fluctuation is a sudden decrease, the no-load energy consumption is reduced by frequency reduction. , For operating frequency, For the rated frequency, This refers to the operating power of the fan after frequency reduction.

[0064] The number of operating fans is dynamically adjusted according to the load: , The number of operating wind turbines. This represents the total number of operating wind turbines. This indicates rounding up.

[0065] On the other hand, the present invention provides a remote control method for wastewater treatment equipment based on the Internet of Things, including:

[0066] Step 1: Construct a multi-dimensional real-time sensing network, use the multi-dimensional real-time sensing network to collect multi-dimensional data, and perform noise reduction and time-series alignment processing on the collected multi-dimensional data;

[0067] Step 2: Design the basic layer, intermediate layer, and optimization layer to establish a dynamic correlation model of load-DO concentration-energy consumption. The basic layer outputs the load trend for a future period of time, the intermediate layer quantifies the DO concentration, and the optimization layer outputs the optimal DO concentration setting value that achieves the effluent standard and minimizes aeration energy consumption.

[0068] Step 3: Develop a dynamic setpoint decision system to achieve adaptive DO concentration from fixed to range-bound. Based on the real-time load and the output of the load-DO concentration-energy consumption dynamic correlation model, dynamically generate the DO concentration setpoint range, and adjust the aeration equipment accordingly.

[0069] The beneficial effects of this invention are as follows:

[0070] 1. Dynamically adapting to load fluctuations to ensure stable effluent quality compliance: This invention constructs a multi-dimensional real-time sensing network, deploys distributed fiber optic DO sensors and sludge activation sensors, and combines an LSTM network to predict influent load trends, accurately capturing sudden increases and decreases in parameters such as COD and flow rate; then, it quantifies DO demand through a microbial metabolic model and outputs the optimal DO setpoint using the PPO algorithm, avoiding the problem of excessive ammonia nitrogen in effluent caused by insufficient load adaptation in traditional fixed DO control, ensuring that effluent ammonia nitrogen stably meets the compliance threshold, and significantly improving the water quality compliance rate.

[0071] 2. Significantly reduces aeration energy consumption, achieving energy conservation and cost reduction: The optimization layer is designed with a reward function based on the dual objectives of achieving effluent standards and minimizing energy consumption, dynamically generating DO concentration ranges. During sudden load drops, fan frequency reduction reduces idle energy consumption; during sudden load increases, the upper limit of fan overload power is rationally controlled, and the number of operating fans is dynamically adjusted according to the load to avoid over-aeration. Compared to traditional fixed DO control modes, this effectively reduces aeration system energy consumption and significantly improves energy utilization efficiency.

[0072] 3. Improve data quality and model reliability to solidify the foundation of regulation: Discrete wavelet transform (DWT) is used to reduce noise in influent flow and DO concentration data, filtering out noise such as electromagnetic interference and sensor drift. At the same time, the data time sequence is aligned by calculating the water flow residence time, solving the time lag problem between influent load and DO response. This provides high-quality data input for LSTM load prediction models and microbial metabolism models, significantly improving the accuracy of model calculations and avoiding regulation inaccuracies caused by data defects.

[0073] 4. Simplified operation and maintenance process, improved remote control efficiency: Relying on the Internet of Things to build a remote control system, it eliminates the need for manual on-site inspection and adjustment. It can remotely monitor multi-dimensional parameters in real time and automatically adjust aeration equipment based on the output of the dynamic decision system. It is especially suitable for decentralized sewage treatment facilities, reducing manual operation and maintenance costs, shortening fault response time, avoiding the lag of traditional manual adjustment, and improving equipment operation stability and operation and maintenance efficiency. The system is highly adaptable and scalable to different scenarios: The multi-dimensional real-time sensing network supports flexible access to various sensors. The dynamic setpoint decision system can adaptively adjust the DO range and blower control strategy according to the load characteristics of different wastewater types. It can adapt to different treatment scales and different water quality fluctuation scenarios without large-scale hardware modification, meet the control requirements after the upgrade of environmental protection standards, and has strong promotion and application value. Attached Figure Description

[0074] The invention will now be further described with reference to the accompanying drawings.

[0075] Figure 1 This is a system module diagram of the remote control system for wastewater treatment equipment based on the Internet of Things of this invention;

[0076] Figure 2 This is a flowchart of the steps of the remote control method for wastewater treatment equipment based on the Internet of Things of the present invention. Detailed Implementation

[0077] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments.

[0078] Example 1

[0079] like Figure 1 As shown in the embodiment of the present invention, the remote control system for wastewater treatment equipment based on the Internet of Things includes:

[0080] Multi-dimensional perception and data preprocessing module: Constructs a multi-dimensional real-time perception network, uses the multi-dimensional real-time perception network to collect multi-dimensional data, and performs noise reduction and time-series alignment processing on the collected multi-dimensional data;

[0081] In the multi-dimensional perception and data preprocessing module, the first specific step in constructing the multi-dimensional real-time perception network is as follows:

[0082] Deploy flow sensors, COD sensors, BOD sensors, ammonia nitrogen sensors, and water temperature sensors at the water inlet end;

[0083] A distributed fiber optic DO sensor is deployed every 5 meters in the aeration tank to collect the dissolved oxygen (DO) concentration in different areas of the aeration tank, reflecting the spatial distribution of oxygen content in the water and avoiding local errors caused by single-point monitoring.

[0084] A new sludge activity sensor has been added, which consists of an ATP-based probe for detecting microbial activity and an oxidation-reduction potential probe.

[0085] ATP detection is used to collect data on microbial activity. ATP is the energy molecule of microbial metabolism and directly reflects the activity of the microbial community.

[0086] Redox potential probes are used to collect redox potentials in aeration tanks, indirectly reflecting the redox state of the system and used to correlate DO supply and demand balance.

[0087] In the multi-dimensional perception and data preprocessing module, the second specific step, which involves using a multi-dimensional real-time perception network to collect multi-dimensional data, is as follows:

[0088] Flow sensors are used to collect influent flow rate, reflecting the scale of wastewater influent; COD sensors are used to collect chemical oxygen demand (COD), reflecting the degree of organic pollution in wastewater; BOD sensors are used to collect biochemical oxygen demand (BOD), reflecting the oxygen demand of microorganisms decomposing organic matter; ammonia nitrogen sensors are used to collect ammonia nitrogen concentration, which is a core indicator of nitrogen pollution; water temperature sensors are used to collect influent temperature, affecting microbial activity and dissolved oxygen solubility; distributed fiber optic DO sensors collect dissolved oxygen (DO) concentration in different areas of the aeration tank; and sludge activity sensors are used to collect sludge activity.

[0089] In the multi-dimensional perception and data preprocessing module, the third specific step, which involves noise reduction and temporal alignment of the collected multi-dimensional data, is as follows:

[0090] The time series data of influent flow rate and DO concentration in the collected data were denoised using Discrete Wavelet Transform (DWT). The steps are as follows:

[0091] Let the original signals of influent flow rate and DO concentration be x(n) (n=0,1,...,N-1), which are decomposed into approximate coefficients by passing through a low-pass filter h and a high-pass filter g. and detail coefficient :

[0092] Approximation coefficient: Detail factor: Where j is the decomposition level, k is the coefficient index, and h and g are the coefficients of the db4 wavelet filter, which satisfy orthogonality;

[0093] For detail coefficients Apply Donoho threshold , Where σ is the noise standard deviation, estimated through the stationary segment of the signal, and N is the signal length. For symbolic functions, Thresholding shrinkage is applied to the absolute values ​​of the detail coefficients;

[0094] Combined with the processed approximation coefficients and detail coefficient The signal is recovered by inverse wavelet transform: ,in, , These are the coefficients of the inverse db4 wavelet filter, ensuring the accuracy of signal reconstruction. Let be the signal reconstructed by inverse wavelet transform, and n be the sequence index of the signal, representing the value of the reconstructed signal at time point n. This represents the index offset of the filter coefficients;

[0095] This enhances the signal-to-noise ratio of signal data and effectively filters out noise such as electromagnetic interference and sensor drift.

[0096] To achieve time matching between influent flow fluctuations and the DO response in the reactor, the DO time series needs to be aligned and the water residence time calculated. : Where L is the length of the aeration tank from the inlet to the DO monitoring point, v is the average flow velocity of the water, which is derived from the flow rate Q and the cross-sectional area S of the tank, S=w·h, where w is the width of the tank and h is the effective water depth;

[0097] pass Shift the timestamps of influent flow rate and DO concentration backward to ensure that changes in influent load and DO are causally correlated on the time axis.

[0098] The three-layer architecture load-DO-energy dynamic correlation model construction module is designed to establish a dynamic correlation model of load-DO concentration-energy consumption by designing a basic layer, an intermediate layer, and an optimization layer. The basic layer outputs the load trend over a period of time, the intermediate layer quantifies the DO concentration, and the optimization layer outputs the optimal DO concentration setting value that achieves effluent standards and minimizes aeration energy consumption.

[0099] In the three-tier architecture load-DO-energy dynamic correlation model construction module, the specific process of the base layer outputting the load trend over a future period is as follows:

[0100] The base layer uses raw data collected by a multi-dimensional real-time sensing network. The raw data includes influent flow rate, COD, BOD, ammonia nitrogen concentration, and influent temperature after the timestamp is shifted backward. These data serve as influent load indicators. The raw data is processed by a Long Short-Term Memory (LSTM) network to predict the load trend in the future and output the influent flow rate, COD, BOD, ammonia nitrogen concentration, and influent temperature for the future period, thus addressing the lag of influent shocks.

[0101] In the three-tier architecture load-DO-energy dynamic correlation model construction module, the second specific step, the intermediate layer quantification of DO concentration, is as follows:

[0102] The intermediate layer constructs a microbial metabolic model, calculates the current actual oxygen consumption rate (OUR) of microorganisms based on sludge activity and ORP data, and quantifies DO demand;

[0103] Among them, ORP data is obtained by direct measurement through an ORP probe;

[0104] The formula for the microbial metabolic model is: ,in, This represents the actual oxygen consumption rate of microorganisms. The total oxygen transfer coefficient of the aeration system. This represents the saturation concentration of dissolved oxygen in the water under the current operating conditions. The actual dissolved oxygen concentration in the aeration tank is measured in real time by a DO sensor. The rate of change of dissolved oxygen concentration over time;

[0105] Higher sludge activity leads to higher oxygen utilization efficiency. The total oxygen transfer coefficient of the aeration system is corrected by adjusting the ATP concentration. : ,in, This is the corrected total oxygen transfer coefficient for the aeration system. Based on the actual activity of the current sludge, The baseline sludge activity was obtained through experiments. To correct the index, it was calibrated experimentally;

[0106] ORP mutations reflect an imbalance between DO supply and demand. An ORP compensation term is introduced to compensate for ORP; when ORP > 200mV, OUR increases by a factor of 1.2. ;

[0107] In the three-layer architecture load-DO-energy consumption dynamic correlation model construction module, the specific process of the third optimization layer outputting the optimal DO concentration setpoint that achieves effluent compliance and minimizes aeration energy consumption is as follows:

[0108] The optimization layer introduces a reinforcement learning algorithm with the dual objectives of achieving effluent standards and minimizing aeration energy consumption. It trains a DO setpoint decision model, and the action of the reinforcement learning algorithm is to output the DO concentration setpoint.

[0109] Design a reward function with the goals of achieving effluent quality standards and minimizing aeration energy consumption:

[0110] Piecewise function Based on the ammonia nitrogen concentration in the effluent Whether the standard is met or not, the threshold for meeting the standard is 5mg / L, and the reward is calculated in two cases:

[0111] When the standard is met The lower the concentration of hydrogen ammonia, the closer the reward is to 1;

[0112] When not meeting the standard The higher the ammonia nitrogen concentration, the greater the absolute value of the negative reward;

[0113] Aeration energy consumption bonus ( ):

[0114] With the lowest energy consumption in history Based on this benchmark, the lower the energy consumption, the higher the reward: ;

[0115] Weighted fusion dual objectives: , , These are weighting coefficients, and , E represents the effluent ammonia nitrogen concentration, and E represents the real-time aeration energy consumption.

[0116] The decision policy for setting the objective (DO) of the reinforcement learning algorithm is updated through proximal policy optimization (PPO). The core formula is the clipped objective function: ,in, For the policy network in state Down Output Action The probability, For the policy network in state Down Output Action The old probability, For the dominant function, For the clip range;

[0117] The base layer, intermediate layer, and optimization layer are trained using historical operational data. Different load scenarios are simulated through a digital twin system to accelerate model convergence. The core reaction kinetic formula is the ammonia nitrogen oxidation rate. ,in, For the maximum specific ammonia nitrogen oxidation rate, This refers to the concentration of heterotrophic bacteria. The influent ammonia nitrogen concentration, The ammonia nitrogen half-saturation constant, The half-saturation constant of DO concentration. This refers to the dissolved oxygen concentration in the water.

[0118] By using a digital twin system to generate virtual scenes, the DO concentration requirements and effluent indicators under different loads are simulated, providing additional training data for the load-DO concentration-energy consumption dynamic correlation model and improving the model convergence efficiency.

[0119] Dynamic DO setting and aeration equipment linkage control module: Develop a dynamic setpoint decision system to achieve adaptive control from fixed DO concentration to DO concentration range. Based on the real-time load and the output of the load-DO concentration-energy consumption dynamic correlation model, dynamically generate the DO concentration setpoint range, and adjust the aeration equipment in linkage based on the dynamically generated DO concentration setpoint range.

[0120] In the dynamic DO setting and aeration equipment linkage control module, the first specific step is to develop a dynamic setpoint decision system to achieve the process of adapting from a fixed DO concentration to an adaptive DO concentration range.

[0121] To accurately determine the influent load status, a comprehensive load index L and its fluctuation are defined. ;

[0122] Comprehensive load index L: ;

[0123] volatility : ;

[0124] in, Chemical oxygen demand (COD) of the influent. This refers to the inlet water flow rate. The weights for influent chemical oxygen demand and influent flow rate are respectively, and , The current overall load, This represents the average total load over the past hour.

[0125] when At that time, the load fluctuation was in a stable state, maintaining a fixed DO concentration range;

[0126] when At that time, the load fluctuation was in a state of sudden increase, and the DO concentration range needed to be expanded;

[0127] when At that time, the load fluctuation was in a state of sudden drop, and the DO concentration range needed to be compressed;

[0128] In the dynamic DO setting and aeration equipment linkage control module, the second specific process for dynamically generating the DO concentration setpoint range based on the real-time load and the output of the load-DO concentration-energy consumption dynamic correlation model is as follows:

[0129] The optimal DO concentration value output based on the load-DO concentration-energy consumption dynamic correlation model Based on the load status, DO concentration ranges are dynamically generated. ;

[0130] When the load fluctuation is in a stable state, the interval width is evenly distributed, balancing accuracy and stability:

[0131] ,in, To establish a reference value for steady-state control, in this invention, ;

[0132] When the load fluctuates sharply, priority should be given to ensuring the lower limit of DO concentration, and the upper limit should be increased to meet the high load: , As a reference value for regulating the sudden rise state, in this invention, ;

[0133] When the load fluctuation is a sudden drop, the upper limit of compression reduces energy consumption, while maintaining the lower limit stability: , As a reference value for regulating the sudden drop state, in this invention, ;

[0134] In the dynamic DO setting and aeration equipment linkage control module, the third specific process, based on the dynamically generated DO concentration setpoint range, for linkage adjustment of the aeration equipment is as follows:

[0135] Based on the dynamically generated DO concentration setpoint range, the fan operating parameters are dynamically adjusted. When the load fluctuation is a sudden increase, the fan is allowed to operate at its maximum over-power limit. When the load fluctuation is a sudden decrease, the no-load energy consumption is reduced by frequency reduction. , For operating frequency, For the rated frequency, This refers to the operating power of the fan after frequency reduction.

[0136] The number of operating fans is dynamically adjusted according to the load: , The number of operating wind turbines. This represents the total number of operating wind turbines. This is represented as rounding up;

[0137] The technical solution of this invention is as follows:

[0138] Construct a multi-dimensional real-time sensing network, use the multi-dimensional real-time sensing network to collect multi-dimensional data, and perform noise reduction and time-series alignment processing on the collected multi-dimensional data.

[0139] The design establishes a dynamic correlation model of load-DO concentration-energy consumption by designing a basic layer, intermediate layer, and optimization layer. The basic layer outputs the load trend over a period of time, the intermediate layer quantifies the DO concentration, and the optimization layer outputs the optimal DO concentration setting value that achieves effluent standards and minimizes aeration energy consumption.

[0140] Develop a dynamic setpoint decision system to achieve adaptive DO concentration from fixed to range-bound. Based on the real-time load and the output of the load-DO concentration-energy consumption dynamic correlation model, dynamically generate the DO concentration setpoint range, and adjust the aeration equipment accordingly.

[0141] Example 2

[0142] like Figure 2 As shown, based on Example 1, the present invention provides a remote control method for wastewater treatment equipment based on the Internet of Things, including:

[0143] Step 1: Construct a multi-dimensional real-time sensing network, use the multi-dimensional real-time sensing network to collect multi-dimensional data, and perform noise reduction and time-series alignment processing on the collected multi-dimensional data;

[0144] Step 2: Design the basic layer, intermediate layer, and optimization layer to establish a dynamic correlation model of load-DO concentration-energy consumption. The basic layer outputs the load trend for a future period of time, the intermediate layer quantifies the DO concentration, and the optimization layer outputs the optimal DO concentration setting value that achieves the effluent standard and minimizes aeration energy consumption.

[0145] Step 3: Develop a dynamic setpoint decision system to achieve adaptive DO concentration from fixed to range-bound. Based on the real-time load and the output of the load-DO concentration-energy consumption dynamic correlation model, dynamically generate the DO concentration setpoint range, and adjust the aeration equipment accordingly.

[0146] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely illustrative of the principles of the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended claims and their equivalents.

Claims

1. A remote control system for wastewater treatment equipment based on the Internet of Things, characterized in that: include: Multi-dimensional perception and data preprocessing module: Constructs a multi-dimensional real-time perception network, uses the multi-dimensional real-time perception network to collect multi-dimensional data, and performs noise reduction and time-series alignment processing on the collected multi-dimensional data. The specific process for denoising and temporal alignment of the collected multidimensional data is as follows: The time series data of influent flow rate and DO concentration in the collected data are denoised using Discrete Wavelet Transform (DWT). The steps are as follows: Let the original signals of influent flow rate and DO concentration be x(n) (n=0,1,...,N-1), which are decomposed into approximate coefficients by passing through a low-pass filter h and a high-pass filter g. and detail coefficient : Approximation coefficient: Detail factor: Where j is the decomposition level, k is the coefficient index, and h and g are the coefficients of the db4 wavelet filter; For detail coefficients Apply Donoho threshold , Where σ is the noise standard deviation and N is the signal length. For symbolic functions, Thresholding shrinkage is applied to the absolute values ​​of the detail coefficients; Combined with the processed approximation coefficients and detail coefficient The signal is recovered by inverse wavelet transform: ,in, , These are the coefficients of the inverse db4 wavelet filter. Let n be the signal reconstructed by inverse wavelet transform, and n be the sequence index of the signal. This represents the index offset of the filter coefficients; Align the DO time series and calculate the water residence time. : Where L is the length of the aeration tank from the inlet to the DO monitoring point, v is the average flow velocity of the water, which is derived from the flow rate Q and the cross-sectional area S of the tank, S=w·h, where w is the width of the tank and h is the effective water depth; pass Shift the timestamps of influent flow rate and DO concentration backward; The three-layer architecture load-DO-energy dynamic correlation model construction module: designing a basic layer, intermediate layer and optimization layer to establish a load-DO concentration-energy dynamic correlation model. The basic layer outputs the load trend for a future period of time, the intermediate layer quantifies the DO concentration, and the optimization layer outputs the optimal DO concentration setting value that achieves effluent standards and minimizes aeration energy consumption. The specific process of quantifying DO concentration in the intermediate layer is as follows: The intermediate layer constructs a microbial metabolic model, calculates the current actual oxygen consumption rate (OUR) of microorganisms based on sludge activity and ORP data, and quantifies DO demand; Among them, ORP data is obtained by direct measurement through an ORP probe; The formula for the microbial metabolic model is: ,in, This represents the actual oxygen consumption rate of microorganisms. The total oxygen transfer coefficient of the aeration system. This represents the saturation concentration of dissolved oxygen in the water under the current operating conditions. This represents the actual dissolved oxygen concentration in the aeration tank. The rate of change of dissolved oxygen concentration over time; Higher sludge activity leads to higher oxygen utilization efficiency. The total oxygen transfer coefficient of the aeration system is corrected by adjusting the ATP concentration. : ,in, This is the corrected total oxygen transfer coefficient for the aeration system. Based on the actual activity of the current sludge, As a baseline sludge activity, To correct the index; ORP mutations reflect an imbalance between DO supply and demand. An ORP compensation term is introduced to compensate for ORP; when ORP > 200mV, OUR increases by a factor of 1.

2. ; Dynamic DO setting and aeration equipment linkage control module: Develop a dynamic setpoint decision system to achieve adaptive control from fixed DO concentration to DO concentration range. Based on the real-time load and the output of the load-DO concentration-energy consumption dynamic correlation model, dynamically generate the DO concentration setpoint range, and adjust the aeration equipment in linkage based on the dynamically generated DO concentration setpoint range.

2. The remote control system for wastewater treatment equipment based on the Internet of Things as described in claim 1, characterized in that: The specific process of constructing the multi-dimensional real-time sensing network is as follows: Deploy flow sensors, COD sensors, BOD sensors, ammonia nitrogen sensors, and water temperature sensors at the water inlet end; A distributed fiber optic DO sensor is installed every 5 meters in the aeration tank, and a new sludge activity sensor is added. The sludge activity sensor consists of an ATP-based probe for detecting microbial activity and an oxidation-reduction potential probe.

3. The remote control system for wastewater treatment equipment based on the Internet of Things as described in claim 1, characterized in that: The specific process of collecting multidimensional data using a multidimensional real-time sensing network is as follows: The flow sensor is used to collect the influent flow rate, the COD sensor is used to collect the chemical oxygen demand (COD), the BOD sensor is used to collect the biochemical oxygen demand (BOD), the ammonia nitrogen sensor is used to collect the ammonia nitrogen concentration, the water temperature sensor is used to collect the influent temperature, the distributed fiber optic DO sensor collects the dissolved oxygen (DO) concentration in different areas of the aeration tank, and the sludge activity sensor is used to collect the sludge activity.

4. The remote control system for wastewater treatment equipment based on the Internet of Things according to claim 1, characterized in that: The specific process by which the base layer outputs the load trend for a future period is as follows: The base layer uses raw data collected by a multi-dimensional real-time sensing network. The raw data includes influent flow rate, COD, BOD, ammonia nitrogen concentration, and influent temperature after the timestamp is shifted backward. These data serve as influent load indicators. The raw data is processed by a Long Short-Term Memory (LSTM) network to predict the load trend over a future period and output the influent flow rate, COD, BOD, ammonia nitrogen concentration, and influent temperature for the future period.

5. The remote control system for wastewater treatment equipment based on the Internet of Things according to claim 1, characterized in that: The specific process by which the optimization layer outputs the optimal DO concentration setting value that meets the standards for effluent and minimizes aeration energy consumption is as follows: The optimization layer introduces a reinforcement learning algorithm with the dual objectives of achieving effluent standards and minimizing aeration energy consumption. It trains a DO setpoint decision model, and the action of the reinforcement learning algorithm is to output the DO concentration setpoint. Design a reward function with the goals of achieving effluent quality standards and minimizing aeration energy consumption: Piecewise function Based on the ammonia nitrogen concentration in the effluent Whether the standard is met or not, the threshold for meeting the standard is 5mg / L, and the reward is calculated in two cases: When the standard is met The lower the concentration of hydrogen ammonia, the closer the reward is to 1; When not meeting the standard The higher the ammonia nitrogen concentration, the greater the absolute value of the negative reward; Aeration energy consumption bonus ( ): With the lowest energy consumption in history Based on this benchmark, the lower the energy consumption, the higher the reward: ; Weighted fusion dual objectives: , , These are weighting coefficients, and , E represents the effluent ammonia nitrogen concentration, and E represents the real-time aeration energy consumption. The decision-making strategy for the DO setting value of the PPO update reinforcement learning algorithm is optimized through a proximal strategy. The core formula is the clipped objective function: ,in, For the policy network in state Down Output Action The probability, For the policy network in state Down Output Action The old probability, For the dominant function, The range is clip.

6. The remote control system for wastewater treatment equipment based on the Internet of Things according to claim 1, characterized in that: The specific process of adapting from a fixed DO concentration to an adaptive DO concentration range is as follows: To accurately determine the influent load status, a comprehensive load index L and its fluctuation are defined. ; Comprehensive load index L: ; volatility : ; in, Chemical oxygen demand (COD) of the influent. This refers to the inlet water flow rate. The weights for influent chemical oxygen demand and influent flow rate are respectively, and , The current overall load, This represents the average total load over the past hour. when At that time, the load fluctuation was in a stable state, maintaining a fixed DO concentration range; when At that time, the load fluctuation was in a state of sudden increase, and the DO concentration range needed to be expanded; when At that time, the load fluctuation was in a state of sudden drop, and the DO concentration range needed to be compressed.

7. The remote control system for wastewater treatment equipment based on the Internet of Things according to claim 1, characterized in that: The specific process for dynamically generating the DO concentration setpoint range is as follows: The optimal DO concentration value output based on the load-DO concentration-energy consumption dynamic correlation model Based on the load status, DO concentration ranges are dynamically generated. ; When the load fluctuation is in a stable state, the interval width is evenly distributed, balancing accuracy and stability: , The baseline value for steady-state control; When the load fluctuates sharply, priority should be given to ensuring the lower limit of DO concentration, and the upper limit should be increased to meet the high load: , This serves as the baseline value for regulating a sudden increase in temperature. When the load fluctuation is a sudden drop, the upper limit of compression reduces energy consumption, while maintaining the lower limit stability: , This is the baseline value for adjusting the sudden drop in temperature.

8. The remote control system for wastewater treatment equipment based on the Internet of Things according to claim 1, characterized in that: The specific process of adjusting the aeration equipment in a coordinated manner is as follows: Based on the dynamically generated DO concentration setpoint range, the fan operating parameters are dynamically adjusted. When the load fluctuation is a sudden increase, the fan is allowed to operate at its maximum over-power limit. When the load fluctuation is a sudden decrease, the no-load energy consumption is reduced by frequency reduction. , For operating frequency, For the rated frequency, This refers to the operating power of the fan after frequency reduction. The number of operating fans is dynamically adjusted according to the load: , The number of operating wind turbines. This represents the total number of operating wind turbines. This indicates rounding up.

Images