Sewage zero discharge control method and system for service facilities of high-cold and high-altitude expressway

Abstract

The invention discloses a sewage zero discharge control method and system for service facilities of high-cold and high-altitude highways. The method is applied to a high-cold high-altitude highway service facility sewage zero-discharge integrated treatment system with the altitude being not less than 2000 m and the extreme temperature in winter being not higher than-20 DEG C. The method comprises the following steps: collecting operation data and water quality data as input parameters; a BP neural network prediction method and a genetic algorithm optimization method are adopted to generate control setting parameters; monitoring the effluent quality in real time to form feedback parameters; based on the deviation between the feedback parameter and a preset effluent quality target value, a fuzzy control method is adopted to generate a control setting parameter; the control setting parameters are converted into equipment operation instructions, and the equipment operation instructions at least act on an aeration blower, an internal reflux pump, a flocculant metering pump, a stirrer and a filtering and backwashing device respectively. Under the conditions of low temperature, low oxygen, strong fluctuation and few / unattended operation, the system can still stably reach the standard for a long time, and zero emission is realized.

Description

Technical Field

[0001] This application relates to the field of wastewater treatment and intelligent control technology, and in particular to a method and system for zero-discharge control of wastewater from highway service facilities in high-altitude and cold regions. Background Technology

[0002] High-altitude and cold regions have harsh climates with large diurnal temperature variations and frequent extreme low temperatures in winter. Under these conditions, wastewater temperatures remain consistently low, significantly reducing biochemical reaction rates. When the water temperature drops to approximately 5°C, microbial activity is drastically inhibited, noticeably impacting treatment efficiency and stability. Simultaneously, high altitudes result in lower atmospheric pressure and thinner oxygen levels, leading to a decrease in dissolved oxygen saturation in the wastewater and increasing the difficulty of oxygen supply to the biological treatment units. The cold environment also poses a risk of equipment and pipeline freezing, further increasing the challenge of continuous and stable operation.

[0003] Wastewater from highway service facilities (such as service areas and management stations) exhibits significant fluctuations in both quantity and quality, as well as distinct compositional characteristics: First, large peak-to-valley differences and frequent abrupt changes occur; in monitoring of certain road sections and time periods, the daily peak flow and valley flow can differ by several times, and the daily variation coefficient of water quality concentration is also significant. Second, high nitrogen load and C / N imbalance are observed; concentrated discharge of toilet wastewater causes a significant increase in influent ammonia nitrogen at peak times, increasing the risk of free ammonia inhibiting microorganisms. Third, strong seasonal and intraday heterogeneity exists; summer sees more travelers, while winter is colder and has less foot traffic, with alternating midday peaks and nighttime troughs. The combination of these factors makes the biological treatment system more susceptible to effluent fluctuations and even short-term exceedances under low-temperature, low-oxygen, and highly volatile operating conditions.

[0004] Existing wastewater treatment in service areas mostly employs combined processes centered on anaerobic-anoxic-aerobic processes, biological treatment followed by sedimentation / filtration, or MBR. These processes achieve good effluent quality under normal temperature and pressure conditions with relatively stable influent flow, but they often face three common problems in high-altitude and cold environments: (1) Insufficient adaptability to low temperature and low oxygen: Dissolved oxygen supply is limited and reaction rate decreases, making it difficult to balance removal rate and energy consumption; (2) Lagging control strategy: Under strong fluctuating load, single-loop constant value PID or fixed threshold control is prone to causing large fluctuations in DO and frequent operation of valves / fans. Chemical dosing relies on manual experience and is also prone to lag and overdosing, resulting in instability of effluent and energy and chemical consumption. (3) High pressure on operation and maintenance: Service areas are often located in remote sections of the road and have limited operation and maintenance resources. Equipment is at risk of freezing and pressure difference increase under low temperature conditions, making it difficult to operate stably and economically for a long time under conditions of few or no people on duty.

[0005] In summary, in the zero-emission scenario of highway service facilities in high-altitude and cold regions (in this article, it refers to the on-site reuse of treated effluent without discharge into the external environment), there is an urgent need for an intelligent operation and control method that can perform multi-parameter coordinated regulation under strong fluctuation conditions, take into account energy consumption / chemical consumption and filtration pressure difference constraints, and is suitable for unattended operation, so as to maintain stable compliance and economical operation under low temperature, low oxygen and sudden load conditions. Summary of the Invention

[0006] The purpose of this invention is to provide a method for zero-discharge control of wastewater from highway service facilities in high-altitude and cold regions. This method addresses the challenges of low temperature, low oxygen, and high volatility by integrating artificial intelligence technologies such as BP neural network prediction, fuzzy control, and genetic algorithm optimization to achieve automatic control and predictive optimization of the wastewater treatment process. Through intelligent analysis of real-time monitoring data, this invention can dynamically adjust key parameters of biochemical reactions and advanced treatment, enabling the system to maintain efficient and stable operation under harsh environments and drastic load changes, and ensuring that all effluent meets standards for reuse to achieve the goal of "zero discharge."

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for zero-discharge control of wastewater from highway service facilities in high-altitude and cold regions, applicable to an integrated wastewater treatment system for highway service facilities in high-altitude and cold regions with an altitude of not less than 2000 meters and extreme winter temperatures not exceeding -20℃, comprising: Collect operational data and water quality data as input parameters; Based on the input parameters, a BP neural network prediction method and a genetic algorithm optimization method are used to generate a baseline setpoint for the control parameters; Real-time monitoring of effluent water quality generates feedback parameters; Based on the deviation between the feedback parameters and the preset effluent water quality target value, a fuzzy control method is used to generate incremental corrections for the control setting parameters; The control setting parameters, synthesized from the baseline setpoint and the incremental correction of the control setting parameters, are converted into equipment operation commands, which are respectively applied to at least the aeration blower, internal return pump, flocculant metering pump, mixer, and filtration and backwashing device. The deviation between the feedback parameters and the preset effluent water quality target value or preset operating constraints is calculated, and a PID algorithm is used to correct the closed loop, applying amplitude constraints and rate of change limits to the synthesized control setting parameters.

[0008] Preferably, the input parameters include at least: five-day biochemical oxygen demand (BOD) of the influent, influent ammonia nitrogen concentration, water temperature, pH, mixed liquor suspended solids concentration, oxidation-reduction potential, effluent turbidity, flow rate, filtration pressure difference, and pool level; the feedback parameters include at least: five-day biochemical oxygen demand (BOD) of the effluent, effluent ammonia nitrogen concentration, effluent turbidity, real-time dissolved oxygen value, and filtration pressure difference; the control setting parameters include at least: target dissolved oxygen value, internal reflux ratio, flocculant dosage, stirring intensity, filtration speed, and backwashing cycle.

[0009] Preferably, the input parameters of the BP neural network prediction method are the five-day biochemical oxygen demand (BOD) of the influent, the influent ammonia nitrogen concentration, the water temperature, the pH, the mixed liquor suspended solids concentration, the oxidation-reduction potential, the effluent turbidity, the flow rate, the filtration pressure difference, and the pool level. The output control parameters are the target dissolved oxygen value and the internal reflux ratio. The BP neural network has a three-layer structure, including an input layer, a hidden layer, and an output layer. The activation function of the hidden layer is the sigmoid function, the activation function of the output layer is a linear function, the training method is the backpropagation algorithm, and the loss function is the mean squared error (MSE). Its calculation process includes: Input layer to hidden layer: ; in, The input parameters are: five-day biochemical oxygen demand (BOD) of the influent, influent ammonia nitrogen concentration, water temperature, pH, mixed liquor suspended solids concentration, oxidation-reduction potential, effluent turbidity, flow rate, filtration pressure difference, and pool level, corresponding to 10 neurons in the input layer. For hidden layer output, The weights from the input layer to the hidden layer. For bias, For the Sigmoid function, : Number of neurons in the hidden layer, a positive integer; Index of hidden layer neurons, from 1 to ; Hidden layer to output layer: ; in, Output parameters include target dissolved oxygen value and internal reflux ratio. The weights from the hidden layer to the output layer. For bias, It is a linear function; Loss function: ; in, Mean square error, No. The target output of the sample is 1. For the first The predicted output for each sample. The number of training samples, Training sample index, from 1 to .

[0010] Preferably, the input to the fuzzy control method is the deviation between the feedback parameters and the preset target values. Specifically, the input includes deviations in the effluent's five-day biochemical oxygen demand (BOD), effluent ammonia nitrogen concentration, effluent turbidity, and real-time dissolved oxygen values. The output control setting parameter is the internal reflux ratio correction value. The input and output fuzzy linguistic variables are divided into three levels: "low," "medium," and "high," and the membership function uses a triangular function with the following form: The fuzzy control rules employ the Mamdani inference mechanism, and the defuzzification method is the centroid method. ; in, The input variables are normalized versions of the deviations in effluent biochemical oxygen demand (BOD) over five days, ammonia nitrogen concentration, turbidity, and dissolved oxygen real-time values, with values ​​ranging from [−1, 1]. :variable x Regarding fuzzy sets A The membership function values, a, b, c: the left endpoint, peak, and right endpoint of the triangle membership function, satisfying a≤b≤c, and the parameter values ​​corresponding to the three levels of "low", "medium", and "high" are "low" (a, b, c) = (−1, −1, 0), "medium" (a, b, c) = (−1, 0, 1), and "high" (a, b, c) = (0, 1, 1); The dimensionless output value obtained from defuzzification is used to generate the internal reflux ratio correction value. Output the discretized values ​​within the universe of discourse. Output fuzzy set In taking values Membership degree of location Output the number of values ​​for the discretized universe of discourse.

[0011] In the above technical solution, the fuzzy control method refers to the real-time dissolved oxygen deviation as the difference between the target dissolved oxygen value obtained by synthesis from the BP output and the real-time dissolved oxygen value. The five-day biochemical oxygen demand (BOD) deviation, effluent ammonia nitrogen concentration deviation, and effluent turbidity deviation are the differences between the five-day BOD, effluent ammonia nitrogen concentration, and effluent turbidity and the preset effluent water quality target values ​​(normalized and mapped to [−1,1] as fuzzy inputs), where positive values ​​indicate exceeding the upper limit and negative values ​​indicate being better than the upper limit; the preset effluent water quality target values ​​are: daily BOD ≤ 10 mg / L, ammonia nitrogen concentration ≤ 5 mg / L, and turbidity ≤ 5 NTU.

[0012] Preferably, the genetic algorithm optimization method uses the following input parameters as the basic data for optimization calculation: five-day biochemical oxygen demand of the influent, influent ammonia nitrogen concentration, water temperature, pH, mixed liquor suspended solids concentration, oxidation-reduction potential, effluent turbidity, flow rate, filtration pressure difference, and pool level. It uses the effluent turbidity and filtration pressure difference from the feedback parameters as filtration performance constraints to optimize the flocculant dosage, stirring intensity, filtration speed, and backwashing cycle in the control parameters. The genetic algorithm uses binary encoding to represent chromosomes, single-point crossover, and gene position flipping as the mutation method. The fitness function is: ; in, The five-day biochemical oxygen demand removal rate. For ammonia nitrogen removal rate, Turbidity removal rate Energy consumption per unit This refers to the amount of medicine consumed per unit. For the filtration pressure difference constraint penalty term, α is the weighting coefficient of the five-day biochemical oxygen demand removal rate, β is the weighting coefficient of the ammonia nitrogen removal rate, γ is the weighting coefficient of the turbidity removal rate, λ is the weighting coefficient of the unit energy consumption, μ is the weighting coefficient of the unit reagent consumption, and ν is the weighting coefficient of the filtration pressure difference constraint penalty term.

[0013] A preferred approach involves calculating the deviation between the feedback parameters and the preset effluent water quality target value, adjusting the control setting parameters, and using a PID algorithm for corrective closed-loop control, including: Its control output calculation formula is: ; And on Apply amplitude constraints and rate of change limits; The preset effluent water quality targets are: five-day biochemical oxygen demand ≤10mg / L, ammonia nitrogen concentration ≤5mg / L, and turbidity ≤5NTU; Preset operating constraints: Dissolved oxygen ≥ 2 mg / L, filtration pressure difference not exceeding the warning threshold; in, Control output / adjustment amount :error, Proportional gain, Integral gain Differential gain Error integral term, : Error differential term, Current time : Integral variable.

[0014] In the above technical solution, based on input parameters, the BP neural network prediction method and the genetic algorithm optimization method respectively generate the baseline setpoints for the control settings; based on the deviation between the feedback parameters and the preset effluent water quality target value, a fuzzy control method is used to generate the incremental corrections for the control settings. Under amplitude constraints and rate of change limitations, the system combines the baseline setpoints and the incremental corrections to form the final control settings, and issues them for execution accordingly. The synthesis rules for different objects are as follows: Dissolved oxygen setting: The device side is tracked by PID; Internal reflux ratio: ; Flocculant dosage / stirring intensity / backwashing cycle: Baseline values ​​are taken from the genetic algorithm optimization method; Filtration speed: Backwashing should be performed in advance when ΔP reaches the warning value.

[0015] in, The target dissolved oxygen value is given by the BP neural network. The final dissolved oxygen setpoint (after amplitude / slope limiting processing) is sent to the PID loop. Simultaneously apply amplitude and rate of change limits. Final internal reflux ratio setting, The internal reflux ratio given by the BP neural network is a baseline. Fuzzy output increment, :Will Gain coefficient converted to actual correction amount Final filtration speed setting, The benchmark for filtering speed optimized by the genetic algorithm optimization method. : The PID fine-tuning variable is the filter pressure difference ΔP as the controlled variable. Pressure difference deviation.

[0016] Among them, the flocculant dosage, stirring intensity and backwashing cycle are slow / discrete variables, and their effects on effluent turbidity and filterability have a minute-level lag. Therefore, the baseline values ​​of the genetic algorithm (GA) rolling optimization are taken and amplitude / rate of change limits are implemented. The filtration speed is a fast variable and the primary controllable quantity of the filtration pressure difference. Therefore, PID fine-tuning with ΔP as the controlled variable is superimposed on the GA baseline. When ΔP reaches the warning threshold, backwashing is triggered in advance to ensure both effluent stability and equipment safety.

[0017] More preferably, the method by which the control setting parameters are converted into device operating instructions and applied to the execution device is as follows: The target dissolved oxygen value is achieved by adjusting the speed and air supply of the aeration blower to maintain the dissolved oxygen concentration in the reaction tank within a preset range. The internal recirculation ratio is controlled by adjusting the flow rate of the internal recirculation pump to control the proportion of aerobic effluent recirculated to the anoxic zone. The dosage of flocculant is adjusted by regulating the dosing rate of the flocculant metering pump to promote the formation of settleable flocs from suspended solids and colloidal particles. The stirring intensity is applied to the mixer by adjusting the speed of the mixer to ensure that the reagents and wastewater in the reaction tank or mixing tank are evenly mixed and kept in suspension; The filtration speed and backwashing cycle are adjusted by regulating the operating mode of the filtration and backwashing devices to control the hydraulic conditions of the filtration process and to initiate backwashing when the pressure difference reaches the threshold or the operating time reaches the set value.

[0018] Preferably, the fuzzy control method further includes online learning and parameter update steps, specifically including: Real-time data on five-day biochemical oxygen demand, ammonia nitrogen concentration, turbidity, and dissolved oxygen in the effluent are continuously collected and stored in a historical database. Based on the preset effluent water quality target values, calculate the time series of deviations in five-day biochemical oxygen demand, ammonia nitrogen concentration, turbidity, and real-time dissolved oxygen values ​​in the effluent; Without changing the form of the triangular membership function and the three-level linguistic variable structure of "low", "medium" and "high", the left endpoint, peak and right endpoint of the triangular membership function corresponding to each input variable are slightly increased or decreased according to the recent deviation distribution; the adjustment range does not exceed the preset ratio, the order of the endpoints is kept such that the left endpoint is not greater than the peak and the peak is not greater than the right endpoint, and the endpoint values ​​are kept within the domain of discourse. The contribution of each rule in the four input three-level rule bases is statistically analyzed. The weight of rules with lower contribution is reduced, and the weight of rules with higher contribution is increased. The weight of the rules is limited to between zero and one. Set a fixed learning rate to limit the magnitude of a single parameter update, and set minimum, maximum, and endpoint order constraints for the membership function endpoints; The updated membership function and rule weights are used for short-term operation verification; when the effluent quality meets the preset effluent quality target value, the updated result is solidified into a new fuzzy control parameter set; when it does not meet the target value, it reverts to the previous solidified parameter set and reduces the subsequent update ratio. When data loss, sensor malfunction, or communication interruption occurs, online learning and parameter updates are paused, the most recent fixed parameter set is retained, and an alarm is triggered.

[0019] A preferred option also includes a secure handover mechanism, specifically including: The safety switching mechanism will be triggered under any of the following circumstances: sensor failure or communication timeout; water tank level exceeding the limit; filtration pressure difference exceeding the warning threshold; effluent five-day biochemical oxygen demand, effluent ammonia nitrogen concentration, effluent turbidity failing to meet the preset effluent water quality target values, or dissolved oxygen real-time value failing to meet the preset operating constraints; ambient temperature or equipment compartment temperature lower than the antifreeze protection temperature. Pause the calculation output of the equipment operation command, switch to safe setting value operation, set the dissolved oxygen target value, internal reflux ratio, flocculant dosage, stirring intensity, filtration speed and backwash cycle to the corresponding safe setting values, and continue to execute amplitude constraints and rate of change limits; After the fault is cleared and the feedback parameters have continuously reached the preset effluent water quality target value for a preset period of time, automatic control, online learning, and parameter updates are restored; alarms are generated throughout the process and the triggering reasons and key parameters are recorded.

[0020] In the above technical solution, sensors refer to online instruments and process transmitters that provide signals for the input / feedback parameters, specifically including: online dissolved oxygen meter, effluent turbidity meter, online ammonia nitrogen analyzer, online BOD5 analyzer, pH / ORP / water temperature probe, MLSS (mixed liquor suspended solids) meter, differential pressure / pressure transmitter of the filtration (or membrane) unit, influent and internal return flow meter, and level gauges of each tank, etc. When any of the above sensors experiences communication timeout, invalid / limit-over, jamming / sudden change, or self-test / maintenance alarm, online learning and parameter updates are suspended, the most recent fixed parameter set is maintained, and an alarm is triggered; when the anomaly involves key control / protection quantities such as dissolved oxygen, filtration differential pressure, effluent turbidity, level, or influent / internal return flow, or causes the feedback parameters to fail to reach the preset effluent water quality target value, a safety switching mechanism is triggered.

[0021] A zero-discharge control system for wastewater from highway service facilities in high-altitude and cold regions includes: The data acquisition module is used to collect operational data and water quality data as input parameters; and to perform real-time monitoring of the effluent water quality to generate feedback parameters; The control module is used to calculate and generate the baseline setpoint of the control setting parameters based on the input parameters using a BP neural network prediction method and a genetic algorithm optimization method; to calculate and generate the incremental correction of the control setting parameters based on the deviation between the feedback parameters and the preset effluent water quality target value using a fuzzy control method; to calculate the deviation between the feedback parameters and the preset effluent water quality target value or preset operating constraints, and to use a PID algorithm to correct the closed loop, thereby applying amplitude constraints and rate of change limits to the synthesized control setting parameters. The execution module is used to convert the control setting parameters synthesized from the baseline setpoint and the incremental correction of the control setting parameters into equipment operation commands, which are respectively applied to at least the aeration blower, the internal return pump, the flocculant metering pump, the mixer, and the filtration and backwashing device.

[0022] In the normal automatic control mode, this invention operates according to a synthetic strategy of "given value - correction - steady-state tracking - constraint - execution": First, a BP neural network provides a baseline given value for key setpoints representing rapid process dynamics (such as the target value of dissolved oxygen (DO) and the internal recirculation ratio); for deviations caused by the five-day biochemical oxygen demand (BOD), ammonia nitrogen, turbidity, and DO in the effluent, fuzzy control calculates incremental corrections and only adds them to setpoints that require water quality closed-loop correction (such as the internal recirculation ratio) to improve compliance robustness; for operational variables constrained by multiple objectives such as energy consumption, chemical consumption, and filtration pressure difference... Quantities such as flocculant dosage, stirring intensity, filtration speed, and backwash cycle are optimized by a genetic algorithm under the constraint of "effluent turbidity / filtration pressure difference" to achieve system-level optimum. For physical quantities requiring precise process tracking and vibration suppression (DO, filtration pressure difference), PID control on the equipment side uses measurements and sets errors for steady-state tracking and overshoot suppression: DO uses "BP input + PID tracking" to drive the blower, and the filtration speed uses "GA reference + PID fine-tuning with pressure difference as the controlled variable." When the pressure difference reaches the warning threshold, backwashing is triggered in advance to cover the original cycle plan. It should be noted that the internal recirculation ratio is "BP (neural network prediction method) input + fuzzy incremental correction," without introducing PID; while the flocculant dosage and stirring intensity are mainly given by GA (genetic algorithm optimization method). All synthesized setpoints undergo amplitude constraints and rate-of-change limits before being issued. They are then converted into specific instructions for blowers / internal reflux pumps / metering pumps / mixers / filters and backwashing devices through the "Set-up → Device" execution mapping, ensuring smooth output, equipment safety, and control consistency. Online learning only updates and verifies the fuzzy membership functions and rule weights in small steps, without changing the above synthesis framework. When a safety condition is triggered, the system pauses the calculation output of BP / fuzzy / GA / PID, and each control variable switches to the safe setpoint. The synthesis strategy is then smoothly restored after the effluent continuously meets the standard for a preset time.

[0023] The controlled object of this invention is an integrated wastewater zero-discharge treatment system for highway service facilities in high-altitude and cold regions. The system comprises at least: an inlet, a biochemical unit, a deep treatment unit, and a reuse / storage unit; preferably, it also includes a sludge treatment unit. Aeration blowers, internal return pumps, flocculant metering pumps, mixers, and filtration and backwashing devices are the execution equipment for the aforementioned process units and are all components of the integrated treatment system. Corresponding online sensors and status acquisition devices constitute a data acquisition module, which, together with the control module and execution module, forms a control system. The system boundary is defined by the raw water interface flange, the reclaimed water interface flange, and the sludge transport interface; the pipeline network and reuse terminal facilities outside the flanges are not part of this system. The innovation of this invention lies in the collaborative control logic and modular integration of the control method and the integrated treatment system; the aforementioned conventional process units and equipment are preferred implementations of existing technologies.

[0024] Compared with the prior art, the beneficial effects of the present invention are: 1. Compared to traditional "fixed threshold / simple PID" systems, this invention utilizes a multi-layered control chain: a BP neural network feedforward setpoint (dissolved oxygen target value, internal recirculation ratio), fuzzy control for online correction based on effluent deviation, and PID closed-loop steady-state vibration suppression. This enables early response to load changes and rapid elimination of deviations. Under conditions of high altitude, low temperature, low oxygen, and sudden load changes, the system can pre-increase / decrease the dissolved oxygen target and recirculation intensity based on load trends, reducing the "hysteresis response after ammonia nitrogen shock" and significantly improving dynamic steady-state compatibility (while being constrained by amplitude / rate of change limitations to avoid control overshoot).

[0025] 2. This invention simultaneously adjusts key variables such as dissolved oxygen target value, internal reflux ratio, flocculant dosage, stirring intensity, filtration speed, and backwash cycle. BP (Browser Optimization) is responsible for the "given value," while fuzzy logic is responsible for "incremental correction based on effluent deviation." The Genetic Algorithm (GA) comprehensively optimizes the dosing / stirring / filtration operation under constraints of effluent turbidity and filtration pressure difference, balancing removal rate, energy consumption, and chemical consumption. Ultimately, the set equipment is mapped to the coordinated control of the blower, internal reflux pump, metering pump, stirrer, and filtration and backwashing devices. Compared to traditional "isolated parameter adjustment," this invention achieves coordinated control and rolling optimization, ensuring each component remains stable at its optimal operating point at the system level.

[0026] 3. In the control structure, the BP neural network outputs the target dissolved oxygen value and the internal reflux ratio based on real-time / historical input conditions, using feedforward prediction. The genetic algorithm (GA) performs rolling optimization (including filtration pressure difference penalty) on flocculant / stirring / filtration speed / backwashing cycle. Fuzzy + PID performs online correction and steady-state vibration suppression for deviations in five-day biochemical oxygen demand, effluent ammonia nitrogen concentration, effluent turbidity, and real-time dissolved oxygen values. Therefore, the system can preemptively enhance treatment capacity before water quality deteriorates or load surges, and automatically reduce aeration / dosing intensity when the load decreases, achieving the dual goals of stable compliance and reduced energy and chemical consumption. Compared to lagging manual / single-loop PID, this invention can suppress deviations in advance.

[0027] 4. This invention introduces online learning to perform small-step updates, verification, and rollback of the membership function endpoints / rule weights of fuzzy control, enabling the system to adapt to seasonal and inflow water structure changes. Simultaneously, a safety switching mechanism is implemented: when the ambient / equipment chamber temperature falls below the anti-freeze temperature, the liquid level exceeds the limit, the filtration differential pressure exceeds the threshold, the sensor / communication malfunctions, or the effluent fails to meet standards, the advanced algorithm output is automatically paused and the system switches to a safe setpoint for operation, smoothly exiting after reaching stability. This combined design significantly reduces manual intervention and supports long-term stable operation in unattended scenarios at high altitudes and cold regions.

[0028] 5. This invention uses effluent quality targets as the core of a closed loop: five-day biochemical oxygen demand ≤10mg / L, ammonia nitrogen concentration ≤5mg / L, turbidity ≤5NTU, and preset operating constraints: dissolved oxygen ≥2mg / L. Through the linkage control of filtration speed / backwash cycle and filtration pressure difference, stable product water and controllable membrane (filtration) load are maintained. Based on meeting the above indicators, the effluent can be reused on-site for toilet flushing, vehicle washing, road greening, and road cleaning, achieving on-site recycling of service facilities and the goal of "zero discharge" of wastewater.

[0029] In summary, this invention addresses the zero-emission scenario of high-altitude and cold-climate service facilities at altitudes ≥2000m and extreme temperatures ≤−20℃. It constructs an integrated intelligent control closed loop encompassing "data acquisition—prediction—optimization—execution—feedback—protection." Through the coordinated use of feedforward (BP) and feedback (fuzzy logic) methods, superimposed with PID steady-state correction, it achieves multi-variable linkage and rolling optimization of key parameters such as DO, internal reflux, flocculant, stirring intensity, and filtration rate / backwashing. Under constraints of effluent targets (five-day biochemical oxygen demand / ammonia nitrogen / turbidity / dissolved oxygen real-time value deviation) and filtration pressure difference and effluent turbidity, it proactively addresses sudden load changes and reduces overshoot and hysteresis. Furthermore, it employs online learning and self-correction parameters, along with a safety switching mechanism to ensure backup operation in situations such as low-temperature freezing / sensor malfunction / substandard water quality. Therefore, even under conditions of low temperature and low oxygen, strong fluctuations, and minimal / unmanned operation, the system can still achieve long-term stable compliance, lower costs, and simpler maintenance, supporting on-site effluent reuse and realizing "zero discharge" of wastewater.

[0030] This invention achieves specific stable control effects under low-temperature, low-oxygen, and highly disturbed operating conditions, such as altitude ≥2000m and extreme winter temperatures ≤−20℃, through an interactive mechanism of "Genetic Algorithm (GA) - BP Neural Network - Fuzzy and PID hierarchical collaboration": The Genetic Algorithm (GA) performs slow-time-domain global optimization of flocculant dosage, stirring intensity, filtration speed, and backwashing cycle under constraints of effluent turbidity and filtration pressure difference, ensuring that the load at the separation end and energy / chemical consumption are at the system-level optimal; The BP Neural Network provides mid-time-domain feedforward settings for dissolved oxygen target values ​​and internal reflux ratio based on influent / effluent water quality and operating conditions, anticipating load mutations; Fuzzy control calculates incremental corrections in real time based on the deviations of five-day biochemical oxygen demand / ammonia nitrogen / turbidity / DO in the effluent, and superimposes them onto the setpoints requiring closed-loop correction; PID performs steady-state tracking and vibration suppression of fast-time-domain variables such as DO and filtration pressure difference on the equipment side, and uniformly applies amplitude / rate of change limits. The complementarity of the three in time-domain decoupling (slow / medium / fast) and object division of labor (biochemical side / separation side) enables the system to maintain the effluent target with small overshoot, fast convergence and low fluctuation even under day-night peak-valley switching and extreme low temperature, while keeping the filtration pressure difference below the warning threshold, taking into account both stable compliance and minimizing energy and chemical consumption. This synergistic stability effect is difficult to achieve by applying each single algorithm independently. Attached Figure Description

[0031] The present invention will be further described with reference to the accompanying drawings, but the embodiments in the drawings do not constitute any limitation on the present invention. For those skilled in the art, other drawings can be obtained based on the following drawings without creative effort.

[0032] Figure 1 This is a schematic diagram of the workflow of a zero-discharge control method for wastewater from highway service facilities in high-altitude and cold regions according to the present invention. Detailed Implementation

[0033] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0034] Please see Figure 1 This invention provides a method for zero-discharge control of wastewater from highway service facilities in high-altitude and cold regions. It is applied to an integrated wastewater treatment system for highway service facilities at altitudes of at least 2000 meters and extreme winter temperatures not exceeding -20°C, comprising: Collect operational data and water quality data as input parameters; Based on the input parameters, a BP neural network prediction method and a genetic algorithm optimization method are used to generate a baseline setpoint for the control parameters; Real-time monitoring of effluent water quality (online / periodic sampling, sampling period ≤T, such as 30–120 min) to generate feedback parameters; Based on the deviation between the feedback parameters and the preset effluent water quality target value, a fuzzy control method is used to generate incremental corrections for the control setting parameters; The control setting parameters, synthesized from the baseline setpoint and the incremental correction of the control setting parameters, are converted into equipment operation commands, which are respectively applied to at least the aeration blower, internal return pump, flocculant metering pump, mixer, and filtration and backwashing device. The deviation between the feedback parameters and the preset effluent water quality target value or preset operating constraints is calculated, and a PID algorithm is used to correct the closed loop, applying amplitude constraints and rate of change limits to the synthesized control setting parameters.

[0035] The input parameters include at least: five-day biochemical oxygen demand (BOD) of the influent, influent ammonia nitrogen concentration, water temperature, pH, mixed liquor suspended solids concentration, oxidation-reduction potential, effluent turbidity, flow rate, filtration pressure difference, and pool level; the feedback parameters include at least: five-day biochemical oxygen demand (BOD) of the effluent (which can be estimated through equivalent soft measurement), effluent ammonia nitrogen concentration, effluent turbidity, real-time dissolved oxygen value, and filtration pressure difference; the control setting parameters include at least: target dissolved oxygen value, internal reflux ratio, flocculant dosage, stirring intensity, filtration speed, and backwash cycle.

[0036] The input parameters of the BP neural network prediction method are the five-day biochemical oxygen demand (BOD) of the influent, the influent ammonia nitrogen concentration, water temperature, pH, mixed liquor suspended solids concentration, oxidation-reduction potential, effluent turbidity, flow rate, filtration pressure difference, and pool level. The output control parameters are the target dissolved oxygen value and the internal recirculation ratio. The BP neural network has a three-layer structure, including an input layer, a hidden layer, and an output layer. The activation function of the hidden layer is the sigmoid function, and the activation function of the output layer is a linear function. The training method is the backpropagation algorithm, and the loss function is the mean squared error (MSE). Its calculation process includes: Input layer to hidden layer: ; in, Input parameters: five-day biochemical oxygen demand (BOD) of influent, influent ammonia nitrogen concentration, water temperature, pH, mixed liquor suspended solids concentration, oxidation-reduction potential, effluent turbidity, flow rate, filtration pressure differential, and tank level. For hidden layer output, The weights from the input layer to the hidden layer. For bias, For the Sigmoid function, : Number of neurons in the hidden layer, a positive integer; Index of hidden layer neurons, from 1 to ; Hidden layer to output layer: ; in, Output parameters include target dissolved oxygen value and internal reflux ratio. The weights from the hidden layer to the output layer. For bias, It is a linear function; Loss function: ; in, Mean square error, No. The target output of the sample is 1. For the first The predicted output for each sample. The number of training samples, Training sample index, from 1 to .

[0037] In the above scheme, without changing the 10 input parameters, the input layer of the BP neural network is configured according to the 10 operating conditions. In order to improve the stability of training and online inference, time alignment (including effluent side analysis delay compensation), outlier / missing measurement processing and scale normalization (preferably scaled to [−1,1]) are performed on each input before the data enters the network. The preprocessing is only numerical transformation and does not add new input dimensions.

[0038] Preferred internal reflux ratio: ; Internal reflux ratio : Internal reflux pump flow rate (nitrified liquor / mixed liquor reflux flow rate from aerobic section to anoxic section). : System inlet water flow rate (flow rate of the main inlet pipe inside the raw water interface flange).

[0039] The input to the fuzzy control method is the deviation between the feedback parameters and the preset target values. Specifically, the input includes deviations in the effluent's five-day biochemical oxygen demand (BOD), effluent ammonia nitrogen concentration, effluent turbidity, and real-time dissolved oxygen values. The output control setting parameter is the internal reflux ratio correction value. The input and output fuzzy linguistic variables are divided into three levels: "low," "medium," and "high." The membership function uses a triangular function with the following form: The fuzzy control rules employ the Mamdani inference mechanism, and the defuzzification method is the centroid method. ; in, The input variables are normalized versions of the deviations in effluent biochemical oxygen demand (BOD) over five days, ammonia nitrogen concentration, turbidity, and dissolved oxygen real-time values, with values ​​ranging from [−1, 1]. :variable x Regarding fuzzy sets A The membership function values, a, b, c: the left endpoint, peak, and right endpoint of the triangle membership function, satisfying a≤b≤c, and the parameter values ​​corresponding to the three levels of "low", "medium", and "high" are "low" (a, b, c) = (−1, −1, 0), "medium" (a, b, c) = (−1, 0, 1), and "high" (a, b, c) = (0, 1, 1); The dimensionless output value obtained from defuzzification is used to generate the internal reflux ratio correction value. Output the discretized values ​​within the universe of discourse. Output fuzzy set In taking values Membership degree of location Output the number of values ​​for the discretized universe of discourse.

[0040] The genetic algorithm optimization method uses the following input parameters as the basic data for optimization calculation: five-day biochemical oxygen demand of influent, influent ammonia nitrogen concentration, water temperature, pH, mixed liquor suspended solids concentration, oxidation-reduction potential, effluent turbidity, flow rate, filtration pressure difference, and pool level. It uses effluent turbidity and filtration pressure difference as feedback parameters as filtration performance constraints to optimize the flocculant dosage, stirring intensity, filtration speed, and backwashing cycle in the control parameters. The genetic algorithm uses binary encoding to represent chromosomes, single-point crossover, and gene position flipping as the mutation method. The fitness function is: ; in, The five-day biochemical oxygen demand removal rate. For ammonia nitrogen removal rate, Turbidity removal rate Energy consumption per unit This refers to the amount of medicine consumed per unit. For the filtration pressure difference constraint penalty term, α is the weighting coefficient of the five-day biochemical oxygen demand removal rate, β is the weighting coefficient of the ammonia nitrogen removal rate, γ is the weighting coefficient of the turbidity removal rate, λ is the weighting coefficient of the unit energy consumption, μ is the weighting coefficient of the unit reagent consumption, and ν is the weighting coefficient of the filtration pressure difference constraint penalty term.

[0041] Calculate the deviation between the feedback parameters and the preset effluent water quality target value, adjust the control setting parameters, and use a PID algorithm for corrective closed-loop control, including: Its control output calculation formula is: ; And on Apply amplitude constraints and rate of change limits; The preset effluent water quality targets are: five-day biochemical oxygen demand ≤10mg / L, ammonia nitrogen concentration ≤5mg / L, and turbidity ≤5NTU; Preset operating constraints: Dissolved oxygen ≥ 2 mg / L, filtration pressure difference not exceeding the warning threshold (e.g., pressure difference 50 kPa); in, Control output / adjustment amount :error, Proportional gain, Integral gain Differential gain Error integral term, : Error differential term, Current time : Integral variable.

[0042] The control setting parameters are converted into device operation commands and then applied to the execution devices in the following manner: The target dissolved oxygen value is achieved by adjusting the speed and air supply of the aeration blower to maintain the dissolved oxygen concentration in the reaction tank within a preset range. The internal recirculation ratio is controlled by adjusting the flow rate of the internal recirculation pump to control the proportion of aerobic effluent recirculated to the anoxic zone. The dosage of flocculant is adjusted by regulating the dosing rate of the flocculant metering pump to promote the formation of settleable flocs from suspended solids and colloidal particles. The stirring intensity is applied to the mixer by adjusting the speed of the mixer to ensure that the reagents and wastewater in the reaction tank or mixing tank are evenly mixed and kept in suspension; The filtration speed and backwashing cycle are adjusted by regulating the operating mode of the filtration and backwashing devices to control the hydraulic conditions of the filtration process and to initiate backwashing when the pressure difference reaches the threshold or the operating time reaches the set value.

[0043] The fuzzy control method also includes online learning and parameter update steps, specifically including: Real-time data on five-day biochemical oxygen demand, ammonia nitrogen concentration, turbidity, and dissolved oxygen in the effluent are continuously collected and stored in a historical database. Based on the preset effluent water quality target values, calculate the time series of deviations in five-day biochemical oxygen demand, ammonia nitrogen concentration, turbidity, and real-time dissolved oxygen values ​​in the effluent; Without changing the form of the triangular membership function and the three-level linguistic variable structure of "low", "medium" and "high", the left endpoint, peak and right endpoint of the triangular membership function corresponding to each input variable are slightly increased or decreased according to the recent deviation distribution; the adjustment range does not exceed the preset ratio, the order of the endpoints is kept such that the left endpoint is not greater than the peak and the peak is not greater than the right endpoint, and the endpoint values ​​are kept within the domain of discourse. The contribution of each rule in the four input three-level rule bases is statistically analyzed. The weight of rules with lower contribution is reduced, and the weight of rules with higher contribution is increased. The weight of the rules is limited to between zero and one. Set a fixed learning rate to limit the magnitude of a single parameter update, and set minimum, maximum, and endpoint order constraints for the membership function endpoints; The updated membership function and rule weights are used for short-term operation verification; when the effluent quality meets the preset effluent quality target value, the updated result is solidified into a new fuzzy control parameter set; when it does not meet the target value, it reverts to the previous solidified parameter set and reduces the subsequent update ratio. When data loss, sensor malfunction, or communication interruption occurs, online learning and parameter updates are paused, the most recent fixed parameter set is retained, and an alarm is triggered.

[0044] It also includes a secure switching mechanism, specifically including: The safety switching mechanism will be triggered under any of the following circumstances: sensor failure or communication timeout; water tank level exceeding the limit; filtration pressure difference exceeding the warning threshold; effluent five-day biochemical oxygen demand, effluent ammonia nitrogen concentration, effluent turbidity failing to meet the preset effluent water quality target values, or dissolved oxygen real-time value failing to meet the preset operating constraints; ambient temperature or equipment compartment temperature lower than the antifreeze protection temperature. Pause the calculation output of the equipment operation command, switch to safe setting value operation, set the dissolved oxygen target value, internal reflux ratio, flocculant dosage, stirring intensity, filtration speed and backwash cycle to the corresponding safe setting values, and continue to execute amplitude constraints and rate of change limits; After the fault is cleared and the feedback parameters have continuously reached the preset effluent water quality target value for a preset period of time, automatic control, online learning, and parameter updates are restored; alarms are generated throughout the process and the triggering reasons and key parameters are recorded.

[0045] This invention also provides a zero-discharge control system for wastewater from highway service facilities in high-altitude and cold regions, comprising: The data acquisition module is used to collect operational data and water quality data as input parameters; and to perform real-time monitoring of the effluent water quality to generate feedback parameters; The control module is used to calculate and generate the baseline setpoint of the control setting parameters based on the input parameters using a BP neural network prediction method and a genetic algorithm optimization method; to calculate and generate the incremental correction of the control setting parameters based on the deviation between the feedback parameters and the preset effluent water quality target value using a fuzzy control method; to calculate the deviation between the feedback parameters and the preset effluent water quality target value or preset operating constraints, and to use a PID algorithm to correct the closed loop, thereby applying amplitude constraints and rate of change limits to the synthesized control setting parameters. The execution module is used to convert the control setting parameters synthesized from the baseline setpoint and the incremental correction of the control setting parameters into equipment operation commands, which are respectively applied to at least the aeration blower, the internal return pump, the flocculant metering pump, the mixer, and the filtration and backwashing device.

[0046] The controlled object of this invention is an integrated wastewater zero-discharge treatment system for highway service facilities in high-altitude and cold regions. The system comprises at least: an inlet, a biochemical unit, a deep treatment unit, and a reuse / storage unit; preferably, it also includes a sludge treatment unit. Aeration blowers, internal return pumps, flocculant metering pumps, mixers, and filtration and backwashing devices are the execution equipment for the aforementioned process units and are all components of the integrated treatment system. Corresponding online sensors and status acquisition devices constitute a data acquisition module, which, together with the control module and execution module, forms a control system. The system boundary is defined by the raw water interface flange, the reclaimed water interface flange, and the sludge transport interface; the pipeline network and reuse terminal facilities outside the flanges are not part of this system. The innovation of this invention lies in the coordinated control logic and modular integration of the control method and the integrated treatment system; the aforementioned conventional process units and equipment are preferred implementations of existing technologies. The filtration and backwashing devices include, but are not limited to, membrane filtration units.

[0047] The following are the specific measurement locations in the integrated zero-discharge wastewater treatment system for highway service facilities in high-altitude and cold regions, corresponding to the 10 input parameters and 5 feedback parameters (all located within the system boundary: between the raw water interface flange and the reclaimed water interface flange; differential pressure is taken from the inlet and outlet of the filtration and backwashing devices).

[0048] Five-day biochemical oxygen demand (BOD5) of influent: Sampling / analysis point for online biochemical oxygen demand at the influent end of the raw water inlet main pipe.

[0049] Influent ammonia nitrogen concentration: At the influent end, the online ammonia nitrogen sampling / analysis point of the raw water inlet main pipe.

[0050] Water temperature: Location of the mixed liquor temperature probe in the biochemical unit, aerobic MBR tank.

[0051] pH: Location of pH meter for mixed liquor in the aerobic MBR tank of the biochemical unit.

[0052] Mixed liquor suspended solids concentration (MLSS): Online sensor for MLSS (sludge concentration) in the aerobic MBR tank of the biochemical unit.

[0053] Oxidation-reduction potential: Biochemical unit, ORP electrode position in the anoxic pool mixture.

[0054] Effluent turbidity: Online turbidity measurement point on the main filtration outlet pipe of the advanced treatment unit.

[0055] Flow rate: At the inlet end, the location of the electromagnetic flow meter on the raw water inlet main pipe.

[0056] Filtration differential pressure: Location of differential pressure transmitters at the inlet and outlet of the deep processing unit, filtration and backwashing device (same set of differential pressure measuring points).

[0057] Water level in the pool: the water inlet end, and the position of the level gauge in the equalization and regulating pool.

[0058] Five-day biochemical oxygen demand (BOD) of effluent: Online BOD measuring point on the main effluent pipe of the deep treatment unit (inside the reclaimed water interface flange).

[0059] Ammonia nitrogen concentration in effluent: Online ammonia nitrogen measuring point on the main effluent pipe of the advanced treatment unit.

[0060] Dissolved oxygen (DO) real-time value: Biochemical unit, aerobic MBR tank mixed liquor dissolved oxygen probe location.

[0061] Example: This embodiment applies the aforementioned zero-discharge control method for wastewater from highway service facilities in high-altitude and cold regions to the wastewater treatment system of a tunnel service area on a section of highway. This service area is located in a high-altitude, cold region with an average altitude of approximately 3000 meters, where extreme winter temperatures can drop below -20°C, meeting the altitude and climatic conditions required by this invention. This embodiment deploys an integrated wastewater treatment system in the service area, along with a supporting control system to achieve fully automated closed-loop control of the wastewater treatment process. This integrated treatment system is housed in an insulated equipment compartment to withstand the frigid climate; the system integrates carbon oxidation, nitrification / denitrification, and deep solid-liquid separation functions, enabling stable operation in high-altitude and low-temperature environments, achieving effluent quality that meets reuse standards, and realizing "zero discharge" of wastewater. The specific structure and control process of this embodiment will be described below.

[0062] First, the overall system structure and deployment environment: Domestic sewage from the service area is collected and enters the integrated treatment system. This system includes sequentially connected pretreatment, anoxic tank, aerobic membrane bioreactor, coagulation, filtration, and backwashing functional units, equipped with necessary pipelines and valves. To adapt to the high-altitude and cold environment, the equipment compartment uses insulation materials and is equipped with anti-freeze heating devices to ensure the internal temperature does not drop below 0℃; the selection of key equipment such as aeration blowers takes into account the thin air at high altitudes, increasing power margins to ensure sufficient oxygen supply even under low oxygen pressure. The designed daily treatment capacity of this system is, for example, 50 m³ / h. 3 It is capable of treating high-concentration organic wastewater from service area toilets and catering (typical influent five-day biochemical oxygen demand is about 300 mg / L, ammonia nitrogen is about 60 mg / L). After pretreatment by a screen to remove large particulate impurities, the wastewater undergoes denitrification in an anoxic tank, nitrification and organic matter degradation in an aerobic MBR tank, and solid-liquid separation through submerged membrane modules. The implementing equipment includes: an aeration blower (supplying oxygen to the aerobic tank), an internal return pump (returning effluent from the aerobic tank to the anoxic tank), a flocculant metering pump (adding coagulant to the membrane tank or sedimentation unit), a mixer (for stirring in the anoxic tank and dosing unit), and a suction pump for the membrane modules (or equivalent filtration-driven / effluent method) and a backwashing device. These units constitute a complete wastewater treatment hardware system. Correspondingly, the control system consists of a data acquisition module, a control calculation module, and an execution module, which are centrally controlled by a PLC industrial control computer. The system comprises a data acquisition module responsible for collecting various operating and water quality parameters, a control calculation module for running prediction, optimization, and feedback control algorithms, and an execution module for sending control commands to each executing device. The entire system has a compact structure and is suitable for long-term stable deployment and operation in environments above 2000 meters in altitude and below -20°C.

[0063] Input and feedback parameters and their acquisition: In this embodiment, the data acquisition module is equipped with various online sensors and analyzers to monitor key parameters in the wastewater treatment process in real time and acquire input and feedback parameters. The input parameters include at least: influent water quality and operating status parameters, such as five-day biochemical oxygen demand (BOD), influent ammonia nitrogen concentration, water temperature, pH value, mixed liquor suspended solids concentration, oxidation-reduction potential (ORP), effluent turbidity, influent flow rate of the treatment system, membrane module filtration pressure differential, and biological tank level, etc. To obtain the above parameters, the system is equipped with: an online BOD5 water quality analyzer and an ammonia nitrogen analyzer (ion-selective electrode method or colorimetric method) on the influent pipeline for continuous monitoring of five-day biochemical oxygen demand and ammonia nitrogen concentration in the influent; a built-in temperature sensor and pH meter in the biological treatment tank for real-time monitoring of water temperature and pH; a suspended solids concentration sensor in the aerobic tank to measure the mixed liquor sludge concentration, and an ORP sensor in the anoxic tank to monitor changes in oxidation-reduction potential; a turbidity meter on the effluent pipeline to monitor the turbidity of the effluent; in addition, electromagnetic flow meters are installed on the influent and return pipes to monitor flow rate, pressure sensors are installed on both sides of the membrane module to measure the filtration pressure difference, and a level gauge is also installed in the biological treatment tank to monitor liquid level changes. These sensors input signals to the PLC, where the data acquisition module filters and standardizes the signals before providing them as input parameters to the control algorithm. On the other hand, the feedback parameters include at least: effluent water quality indicators, such as five-day biochemical oxygen demand, effluent ammonia nitrogen concentration, and effluent turbidity, as well as process operation indicators such as real-time dissolved oxygen values ​​and filtration pressure difference. The aforementioned feedback parameters are primarily acquired using the same set of sensors / analyzers. For example, the five-day biochemical oxygen demand (BOD) and ammonia nitrogen in the effluent are measured periodically by an online water quality analyzer; the effluent turbidity is read in real-time by a sensor at the effluent outlet; the dissolved oxygen sensor installed in the aerobic tank continuously monitors the dissolved oxygen concentration; and the membrane filtration differential pressure is provided in real-time by a pressure sensor. When there is a delay in the effluent water quality analyzer's measurement, real-time values ​​can be estimated using historical data and a soft sensor model to assist in control. After comparing all feedback parameters with their corresponding preset effluent target values, the deviation signal is input into the feedback control algorithm. Through the aforementioned sensor network, this embodiment achieves comprehensive perception of important state parameters and effluent water quality throughout the entire wastewater treatment process, providing sufficient data support for subsequent control decisions.

[0064] In this embodiment, the control module comprehensively utilizes a multi-level control strategy combining BP neural network prediction, fuzzy control, genetic algorithm optimization, and PID correction to ensure stable compliance with standards under various conditions. The control process includes the following levels: feedforward predictive control, optimized setpoint control, feedback fuzzy regulation, and PID closed-loop correction.

[0065] First, the system uses a backpropagation (BP) neural network to predict and calculate the treatment process, generating initial control parameters. The input to the BP neural network is the aforementioned input parameter vector, including ten parameters: five-day biochemical oxygen demand (BOD) of the influent, influent ammonia nitrogen concentration, water temperature, pH, mixed liquor suspended solids concentration, oxidation-reduction potential, effluent turbidity, flow rate, filtration pressure difference, and pool level. The network output consists of two key control parameters: target dissolved oxygen value and internal recirculation ratio. This embodiment uses a three-layer BP neural network (input layer - hidden layer - output layer). The activation function for the hidden layer neurons is the sigmoid function, and the output layer uses a linear function. This network is trained based on a large amount of historical operating data using the backpropagation algorithm, with mean squared error (MSE) as the loss function. During operation, the normalized real-time input parameters are transmitted from the input layer to the hidden layer. The hidden layer extracts the wastewater treatment operating conditions and performs a sigmoid nonlinear mapping, then transmits the results to the output layer for linear combination, ultimately outputting the predicted target dissolved oxygen value and internal recirculation ratio. These two outputs reflect the dissolved oxygen concentration settings required by the system to achieve the effluent targets under the current influent water quality and operating conditions, as well as the appropriate ratio of effluent from the aerobic unit to the anoxic unit. For example, when the influent organic load and ammonia nitrogen concentration are high, the BP network predicts that the target dissolved oxygen value needs to be increased and the internal recirculation ratio increased to enhance aerobic degradation and nitrification, accelerate anoxic denitrification, and thus prevent the effluent five-day biochemical oxygen demand or ammonia nitrogen from exceeding the standard. Through predictive control of the BP neural network, feedforward adjustment of key biochemical parameters is achieved, providing reasonable control settings at the beginning of pollutant entry into the treatment unit, and improving the system's response speed to water quality fluctuations.

[0066] Meanwhile, the control module also uses a genetic algorithm (GA) to optimize some control settings to balance treatment efficiency and energy and chemical consumption. The GA optimization calculation uses the aforementioned input parameters as basic data. Specifically, it inputs current operating data such as five-day biochemical oxygen demand (BOD), ammonia nitrogen concentration, water temperature, pH, mixed liquor suspended solids concentration, redox potential, effluent turbidity, flow rate, filtration pressure difference, and liquid level into the algorithm model. Effluent turbidity and filtration pressure difference are considered as constraints reflecting the performance of the filtration unit. The goal of the GA optimization is to perform global optimization on four other control settings: flocculant dosage, stirring intensity, filtration speed, and backwashing cycle. In this embodiment, the GA uses binary encoding to represent candidate solutions for the control parameters, discretizing the values ​​of each parameter to a certain precision and then concatenating them into chromosomes. The population size and generation number are set as needed, for example, a population of 50 individuals and a generation number of 100. In genetic operations, single-point crossover is used to recombine chromosomes, and gene position flipping is used to achieve mutation, maintaining population diversity. The fitness function is designed based on the overall effect of wastewater treatment: on the one hand, it encourages the improvement of pollutant removal rates, such as maximizing the removal rates of five-day biochemical oxygen demand (BOD5), ammonia nitrogen, and turbidity; on the other hand, it penalizes excessive energy consumption and reagent dosage (by adding negative benefits of unit energy consumption and unit reagent consumption terms to the fitness function). Simultaneously, for candidate solutions that exceed filtration performance constraints (such as excessive membrane pressure difference or excessive suspended solids in the effluent), a penalty factor is applied to the fitness function to reduce its fitness value. This fitness function can be characterized by a weighted sum, for example: Fitness F = α·five-day BOD5 removal rate + β·ammonia nitrogen removal rate + γ·turbidity removal rate - λ·unit energy consumption - μ·unit reagent consumption - ν·(penalty for exceeding filtration pressure difference limit), where α, β, γ, λ, μ, and ν are weight coefficients set based on operational experience. The genetic algorithm continuously improves the average fitness of the population through generational evolution, eventually converging to obtain an approximately optimal parameter combination. In this embodiment, the optimal control settings output by GA calculation are the values ​​of flocculant dosage, stirring intensity, filtration speed, and backwashing cycle. These optimized values ​​can reduce energy and chemical consumption while meeting the effluent quality indicators, and at the same time ensure that the membrane filtration unit does not operate under overload (the membrane pressure difference is kept within a safe range).

[0067] While providing initial settings through the aforementioned feedforward prediction and optimization control, the system also performs real-time fuzzy feedback control on deviations in effluent water quality to achieve closed-loop regulation. This embodiment pre-sets target values ​​for effluent water quality: 5-day biochemical oxygen demand (BOD) ≤ 10 mg / L, ammonia nitrogen ≤ 5 mg / L, turbidity ≤ 5 NTU, and membrane filtration differential pressure not exceeding a warning threshold (e.g., 50 kPa). The control module continuously collects the latest effluent water quality feedback parameters and compares them with the aforementioned target values ​​to obtain deviation signals. The inputs to the fuzzy control are these deviations, including: 5-day BOD deviation, ammonia nitrogen concentration deviation, turbidity deviation, and real-time dissolved oxygen deviation (the real-time dissolved oxygen deviation is the difference between the synthesized target dissolved oxygen value output by the BP and the real-time dissolved oxygen value). For each input deviation, a corresponding fuzzy linguistic variable is defined, divided into three levels: low, medium, and high. A symmetrical triangular membership function is used to describe the membership degree distribution of each level. For example, after the deviation value is normalized and mapped to the interval [-1, 1], the membership function triangle (a, b, c) = (−1, −1, 0) is defined for "low"; (−1, 0, 1) for "medium"; and (0, 1, 1) for "high". On the output side, the output control setting parameter of the fuzzy controller is the internal reflux ratio correction value, which is also represented by the fuzzy language "low / medium / high" and its membership degree is defined by the triangular membership function. Based on the experience of professional technicians and a large amount of experimental data, a fuzzy rule base containing four input variables and three combinations of linguistic values ​​for each input was pre-developed (the number of rules is 3^4 = 81, which can be appropriately simplified for some impossible combinations). Rule reasoning adopts a Mamdani-type reasoning mechanism, that is, the output fuzzy set is derived from the input fuzzy set through the "if...then..." rule, and the centroid method is used for defuzzification to obtain the internal reflux ratio correction value ΔR. The fuzzy controller runs every few seconds, dynamically adjusting the internal recirculation ratio control parameters based on the current deviation. For example, when high ammonia nitrogen and negative dissolved oxygen deviation (below the set value) are detected in the effluent, the fuzzy rules will output a signal to "increase the internal recirculation ratio" to increase the nitrification liquor recirculation flow and enhance the denitrification capacity and nitrification liquor recirculation in the anoxic zone, thereby further reducing ammonia nitrogen and total nitrogen content. Similarly, when the five-day biochemical oxygen demand (BOD) of the effluent is slightly higher than the target while ammonia nitrogen is normal, it may only be necessary to increase the aeration rate of the aerobic tank to promote the degradation of residual organic matter by heterotrophic bacteria. Through the coordinated adjustment of fuzzy logic, the control system can make intelligent decisions for complex and changing operating conditions, compensating for the shortcomings of single predictive or PID control in handling nonlinear and strongly coupled problems in a timely manner, ensuring that all effluent indicators consistently meet standards.

[0068] To further improve control accuracy and system steady-state consistency, this embodiment incorporates PID (Proportional-Integral-Derivative) closed-loop control in the feedback loop to correct control setpoints in real time. Specifically, the PID controller uses the effluent quality deviation as the control error signal, and obtains the correction amount for relevant control parameters through proportional, integral, and derivative operations. This correction is then added to the control setpoints determined by the aforementioned BP / fuzzy / GA methods to form the final execution command. The introduction of PID control is primarily aimed at unit processes with deterministic relationships, such as maintaining dissolved oxygen concentration and controlling membrane filtration pressure differential. For example, the deviation between the actual dissolved oxygen value in the aerobic tank and the effluent setpoint of 2 mg / L is directly applied to the aeration blower frequency via an adjustment signal output by the PID controller to fine-tune the air supply in real time and precisely control DO near the target value. Similarly, when the membrane filtration pressure differential approaches the threshold, the PID controller can appropriately reduce the filtration rate or trigger a backwash earlier based on the rate of pressure differential increase to prevent the pressure differential from continuously rising and causing membrane fouling. The PID controller has a high adjustment frequency and rapid response, continuously correcting effluent deviations caused by load fluctuations. It is important to note that, to avoid the system being impacted by rapid changes in control variables, this embodiment applies amplitude constraints and rate-of-change limits to the adjusted control settings in the PID output stage. That is, each increment of the PID output has a preset maximum amplitude, and the rate of parameter change per unit time is limited to a safe range. This mechanism ensures that even with large errors, the adjustment of control parameters is gradual and will not cause shock loads to the biological treatment unit. Through the synergistic effect of PID closed-loop correction and the aforementioned fuzzy control, the entire control system forms a complete closed loop: it has both feedforward prediction and optimization, as well as feedback fuzzy adjustment and fine correction, enabling it to maintain stable effluent water quality under various dynamic conditions, and ensuring closed-loop consistency and high implementability of the control logic.

[0069] The control setting parameters' effect on the execution equipment: After the above multi-level control calculations, the control setting parameters obtained in this embodiment include the dissolved oxygen target value, internal reflux ratio, flocculant dosage, stirring intensity, filtration speed, and backwashing cycle. The control system converts these parameters into operating instructions for specific equipment and sends them to the corresponding execution equipment through the PLC's analog / digital output interface or the inverter's communication interface, thereby achieving automatic control of each unit. The specific operation method is as follows: Dissolved oxygen target value: The control system transmits the calculated target dissolved oxygen concentration to the frequency converter of the aeration blower, adjusting the blower motor speed and the opening of the air supply valve to change the air supply and maintain the DO of the aerobic tank mixed liquor above 2 mg / L. For example, when the target dissolved oxygen increases, the PLC increases the blower frequency output, accelerating the air delivery; when the target decreases or the actual DO exceeds the requirement, the blower speed decreases. Through this control, the real-time dissolved oxygen value in the reaction tank always stays within the set range, meeting the oxygen requirements of nitrifying microorganisms while avoiding excessive oxygen supply and energy waste.

[0070] Internal recirculation ratio: The control system adjusts the speed or operating time of the internal recirculation pump according to the internal recirculation ratio setpoint, thereby changing the flow rate of the recirculated liquid between the anoxic and aerobic tanks. The internal recirculation ratio is defined as the ratio of the recirculation flow rate to the influent flow rate (for example, the optimized value may be in the range of 200%~400%). When the internal recirculation ratio is increased, the PLC increases the frequency of the internal recirculation pump or operates it for a longer period, allowing more nitrified liquid to recirculate from the aerobic section to the anoxic section; conversely, it reduces the recirculation flow rate. By adjusting the internal recirculation ratio, the proportion of nitrified liquid recirculation flow rate to the total treatment volume can be controlled, thereby affecting the supply of feedstock (nitrate) for denitrification in the anoxic zone. For example, in this embodiment, the internal recirculation ratio is controlled at approximately 300%, which has proven to be the optimal recirculation ratio under this condition, ensuring sufficient supply of nitrate required for denitrification in the anoxic section without introducing excessive dissolved oxygen.

[0071] Flocculant Dosage: Based on the flocculant dosage settings, the control system adjusts the speed of the flocculant metering pump to precisely control the dosage of flocculants (such as PAC polyaluminum chloride and PAM high-molecular organic flocculants). Typically, the flocculant is added before the membrane filtration unit from the biological effluent to promote the formation of larger flocs from colloids and fine suspended particles in the water, reducing effluent turbidity and membrane fouling tendency. When GA optimization or manual experience requires increasing the flocculant dosage, the PLC increases the metering pump frequency to add more reagent per hour; when the water quality is clear or the effluent turbidity is consistently low, the pump speed is appropriately reduced and PAM is added to achieve better flocculation. Automatic control of the flocculant dosage ensures that the solid-liquid separation unit always operates at high efficiency.

[0072] Mixing Intensity: The control system adjusts the mixer speed based on the set mixing intensity value. One mechanical mixer is installed in each of the anoxic tank and the chemical mixing tank to ensure thorough mixing of the liquids and prevent sedimentation dead zones. Mixing intensity is typically expressed as mixer speed or power. When enhanced mixing is needed, the PLC increases the mixer motor frequency and speed to increase mixing power; conversely, when the reaction is complete and excessive mixing may waste energy, the speed is reduced. Adjusting the mixing intensity ensures rapid and uniform mixing of flocculants and other added chemicals with the wastewater, while maintaining the sludge in the biological treatment tank in a suspended state, preventing a decrease in treatment efficiency due to localized sedimentation.

[0073] Filtration Rate and Backwash Cycle: The membrane filtration unit consists of a suction pump and an air backwashing device. The control system adjusts the operating frequency of the membrane suction pump or the opening of the outlet valve according to the filtration rate setpoint to control the instantaneous flux (filtration rate) of the membrane filtration. For example, when the filtration rate setpoint is increased, the suction pump accelerates water extraction, increasing the amount of water passing through the membrane per unit time; conversely, it slows down suction to reduce the filtration rate. Simultaneously, according to the backwash cycle setpoint, the PLC times and automatically triggers the membrane module's backwashing procedure after the set time interval. Backwashing can be performed by back-suction of permeate water or air wiping to clean the membrane surface. Between two timed backwashes, the control system continuously monitors the membrane's filtration pressure differential. When the pressure differential reaches a warning threshold (indicating membrane fouling accumulation), the system will trigger a backwash prematurely to restore membrane flux, even if the planned cycle time has not yet arrived. During backwashing, the corresponding solenoid valve switching procedure is executed: the suction pump is paused, the backwash water valve is opened, or the air compressor is started to inflate and flush the membrane, and normal filtration resumes after the set duration. By controlling the filtration speed and backwashing in a coordinated manner as described above, the membrane filtration unit can fully exert its interception effect while avoiding damage to the membrane module or reduction of water output due to excessive pressure differential.

[0074] By controlling the parameters set above and the action path of the physical devices, this embodiment achieves a seamless connection from the output of the control algorithm to the actual process adjustment. Each execution device works collaboratively under the control command, enabling the sewage treatment system to maintain stable and efficient operation in high-altitude and cold environments.

[0075] Online learning and parameter update mechanism: The fuzzy control module in this embodiment also has online learning capabilities, which can update fuzzy control parameters (such as membership functions and rule weights) in real time to adapt to slow changes in operating conditions or seasonal changes during long-term operation, thereby improving the adaptability of the control system. The specific implementation of online learning and parameter update is as follows: Data Accumulation and Deviation Calculation: The control system continuously collects and stores real-time data on key effluent water quality parameters, including five-day biochemical oxygen demand (BOD), ammonia nitrogen concentration, turbidity, and dissolved oxygen levels. This data is recorded in a historical database in chronological order. Based on a set time window (e.g., data from the most recent week or month), the system periodically extracts the difference between actual effluent values ​​and corresponding target values, forming a time series of deviations for each parameter. For example, it calculates the mean and variance of the five-day BOD difference over a certain period, and the trend of ammonia nitrogen deviation over time, providing a basis for subsequent fuzzy parameter adjustments.

[0076] Membership function fine-tuning: While maintaining the "triangular membership function" form of the fuzzy controller's input / output linguistic variables and the low, medium, and high three-level structure, the system makes small adjustments to the membership function parameters corresponding to each input variable based on recent deviation data. Specifically, if an input deviation has been consistently positively high recently, the triangle representing "high" in the membership function is shifted slightly to the left (towards lower deviation) to reduce its membership threshold, allowing the controller to more sensitively identify high deviations. Conversely, if a deviation has been consistently close to zero or negative, the corresponding "low" trigonometric function is shifted slightly to the right to avoid over-response. During the adjustment process, each adjustment does not exceed a preset proportion (e.g., each change does not exceed 5% of the original interval width), and the three endpoints of the trigonometric function are strictly maintained in the order of left, middle, and right, i.e., left endpoint ≤ peak value ≤ right endpoint. Simultaneously, the values ​​of each endpoint remain within the original universe of discourse [-1, 1], and do not exceed reasonable limits. Such gradual adjustments allow the fuzzy set definition of the fuzzy controller to be optimized step by step as the operating conditions change. For example, the judgment criteria for "high / medium / low" deviation can be gradually corrected in response to the trend of water deviation caused by seasonal changes.

[0077] Rule weight adjustment: The system statistically analyzes the usage of all rules in the fuzzy rule base, calculating the trigger frequency and contribution of each rule to the control output in recent operations. If some rules are rarely triggered or contribute little to output adjustment even when triggered, their importance is determined to be low, and their weight coefficient is reduced accordingly. Conversely, for rules that are frequently triggered and have a significant impact on the control output, their weights are increased. Rule weights are limited to a range of 0 to 1. For example, initially, all rule weights are 1, but after a period of online learning, the weights of different rules may differ within the range of 0.5 to 1. A fixed learning rate is also used to limit the amount of weight change in a single instance during adjustment to avoid drastic weight fluctuations due to extreme single cases. By adjusting rule weights, the fuzzy inference process can focus more on recently effective control patterns, improving control accuracy.

[0078] Parameter Update Verification and Consolidation: After obtaining new fuzzy membership functions and rule weights in each round of online learning, the system does not immediately and permanently replace the original parameters. Instead, it uses these as a temporary parameter set for short-term control operation tests (such as several control cycles). During this verification period, the system monitors whether the effluent quality still meets the preset target requirements (five-day biochemical oxygen demand, ammonia nitrogen, turbidity, etc. all meet the standards). If the results show that the effluent indicators are stably up to standard, it indicates that the parameter update is effective, and the system consolidates and stores the updated membership functions and weights as a new fuzzy control parameter set for continued use. If the update causes fluctuations and exceedances in the effluent indicators, the system immediately reverts to the previously consolidated parameter set, abandons the current adjustment, and reduces the adjustment magnitude or slows down the learning rate in subsequent learning to avoid adverse effects again. Through the online verification mechanism of "running and correcting simultaneously," the safety and effectiveness of parameter updates are ensured.

[0079] Learning halt under abnormal operating conditions: When abnormal situations such as data loss, critical sensor malfunction, or communication interruption occur, the system will pause the online learning and parameter update process. At this time, the control module will temporarily freeze the fuzzy controller parameters to the most recent reliable parameter set and will not automatically adjust them until the data returns to normal. During this period, if the deviation is large, operation will be maintained primarily by conventional control methods such as PID control, while simultaneously triggering alarms to remind maintenance personnel to check for sensor or communication failures. This safety design ensures that online learning only occurs when the data is reliable, avoiding erroneous parameter adjustments caused by incorrect data.

[0080] In summary, through the online learning mechanism, the fuzzy control parameters of this embodiment can continuously improve themselves during operation, enabling the control system to have a certain degree of self-adaptation and self-optimization capabilities. Even during long-term operation, it can still effectively cope with the slow changes in the quality of sewage in the service area and maintain the best control effect.

[0081] Triggering, Execution, and Exit of the Safety Switching Mechanism: Considering the special characteristics of high-altitude and frigid environments and highway service area operations, this embodiment establishes a safety switching mechanism to ensure system safety and the lower limit of effluent water quality in special or faulty situations. This safety switching mechanism includes three aspects: triggering conditions, execution strategy, and recovery conditions. Triggering Conditions: The safety switching mechanism will be immediately triggered to enter safe mode operation when any of the following conditions occur: 1) Critical monitoring equipment malfunctions or data communication times out (e.g., sensor failure or network interruption, preventing the control system from obtaining accurate real-time data); 2) The water tank level exceeds the limit, for example, the level in the regulating tank is too high and close to overflow, or too low and may cause the pump to run dry; 3) The membrane filtration pressure difference exceeds the warning threshold (indicating a serious risk of membrane clogging); 4) The effluent water quality does not meet the preset target value, i.e., any one of the following indicators—five-day biochemical oxygen demand, ammonia nitrogen, or turbidity—continuously exceeds the standard (failing to meet the above effluent compliance requirements), or the real-time dissolved oxygen value does not meet the preset operating constraints; 5) The ambient temperature or the internal temperature of the equipment compartment is lower than the anti-freeze protection temperature (e.g., the internal temperature is below 0℃ or the ambient temperature drops to below -30℃, posing a risk of freezing). The occurrence of any of the above conditions indicates that the system is in an abnormal or potentially dangerous state, requiring temporary exit from optimized control to prioritize safety. The control system identifies these situations by monitoring sensor status, self-diagnosis, and operating data, and issues a safety switching command once the conditions are met.

[0082] Execution Strategy (Safety Mode): Upon triggering the safety switch, the control module will pause the output of the advanced control algorithms involved in the claims, temporarily halting the real-time adjustment of the equipment by calculations such as BP neural networks, fuzzy control, and genetic algorithms, and switching to operation at the safety setpoints. Specifically, the system resets all key control settings to pre-set safety values, including dissolved oxygen target value, internal reflux ratio, flocculant dosage, stirring intensity, filtration speed, and backwashing cycle, all switched to fixed values ​​in safety mode. For example, in safety mode, the aeration rate can be set to 80% of the normal value to avoid both hypoxia and over-aeration; the internal reflux pump can maintain a moderate flow rate (e.g., 50% of the design maximum); the reagent dosage can be maintained at a minimum level to maintain basic treatment functions; the mixer can operate at a medium speed to prevent sedimentation in the tank; and the filtration device can reduce suction intensity to alleviate membrane load and increase backwashing frequency to prevent clogging. Even if the advanced algorithm detects a deviation, it will not adjust the output but will prioritize operation at the safety setpoints. Meanwhile, the control system continues to apply amplitude constraints and rate-of-change limits to the control of each actuator, ensuring a smooth transition of parameters caused by the switch and preventing new shocks to the system. Although processing efficiency may be slightly reduced in safe mode, it maximizes the protection against catastrophic accidents, maintains effluent quality within a basically acceptable range, and provides maintenance personnel with time to troubleshoot. Throughout the switchover process, the system will issue audible and visual alarms or remote alarms, indicating that it is currently operating in safe mode and the reason for the alarm.

[0083] Recovery Conditions: After operating in safe mode, maintenance personnel should promptly check for faults and take appropriate measures. Once the fault is resolved or the abnormal situation disappears, the system will not immediately switch back to automatic mode. Instead, it will first check whether the effluent feedback parameters have returned to standard and stabilized. Once the effluent's five-day biochemical oxygen demand, ammonia nitrogen, turbidity, and other indicators continuously reach the preset target values ​​and maintain this level for a predetermined period (e.g., continuous operation for 2 hours), the system is considered to have returned to normal. At this point, the safety switching mechanism will exit safe mode: the control module will gradually resume the normal computational output of BP neural network prediction, genetic algorithm optimization, and fuzzy control, smoothly transitioning the control settings from safe values ​​back to automatically calculated values, and regaining control of the optimized control of each actuator. Simultaneously, the previously suspended online learning and parameter update functions will be reactivated. However, in the initial recovery phase, a more conservative adjustment may still be used to observe the system response. Throughout the entire process of triggering, maintaining, and recovering from safe mode, the control system records relevant logs, including trigger time, trigger reason, key parameter values, adopted safe settings, and recovery time. These records can be used for future analysis and improvement of system operation strategies to prevent similar situations from recurring. Through the aforementioned safety switching mechanism, this embodiment ensures that the wastewater treatment system can still operate safely under extreme environmental conditions or unexpected failures, avoiding excessive effluent discharge or equipment damage, and achieving the robustness and safety requirements of the control logic.

[0084] Example of Key Operating Parameters: Based on actual operating data from a tunnel service area, the following example illustrates the effectiveness of the control method in this embodiment. During a typical winter operation, the influent water quality of the service area fluctuated significantly: during the morning peak hours, the five-day biochemical oxygen demand (BOD) was monitored to be approximately 300 mg / L, ammonia nitrogen approximately 60 mg / L, turbidity approximately 250 NTU, and the water temperature was only around 8°C. Faced with this challenge of high organic and nitrogen loads, low temperature, and low oxygen, the control system automatically adjusted various control parameters: BP neural network prediction increased the dissolved oxygen target to approximately 4.5 mg / L and the internal reflux ratio to 350% to enhance nitrification and denitrification capabilities; GA optimization increased the flocculant dosage by 20% and slightly increased the filtration speed to cope with the potential increase in turbidity; fuzzy control detected the increasing trend of the five-day BOD in the effluent and quickly issued a large-scale aeration command. Through a series of coordinated control measures, all units of the system operated smoothly, and the effluent water quality gradually met the standards. At the end of the day's treatment, the measured five-day biochemical oxygen demand (BOD) of the effluent dropped to approximately 8 mg / L, far below the set upper limit of 10 mg / L; the effluent ammonia nitrogen dropped below 2 mg / L, indicating good nitrification; the effluent turbidity remained around 0.5 NTU (thanks to sufficient flocculation and deep membrane interception); and the dissolved oxygen remained stable within the range of 4–4.5 mg / L, with an effluent dissolved oxygen content greater than 2 mg / L. The membrane filtration pressure differential fluctuated around 25 kPa, without reaching the 50 kPa warning line, and the membrane module was automatically backwashed every 8 hours. Therefore, under the control method of this embodiment, the wastewater treatment system in this high-altitude service area can still achieve efficient removal of various pollutants even under low temperature and high organic load conditions. The effluent quality is comprehensively superior to the designed reuse standards, truly achieving zero discharge of wastewater "within the premises." Throughout the process, data acquisition—prediction / optimization calculation—on-site execution—feedback adjustment constitutes a closed-loop consistent automatic control circuit. The coordinated action of each module ensures that the system always operates in a controlled and safe state. This embodiment fully verifies the feasibility and effectiveness of the method of the present invention in actual engineering.

[0085] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Contents not described in detail in this specification belong to prior art known to those skilled in the art.

Claims

1. A high-cold high-altitude highway service facility sewage zero discharge control method, characterized in that, An integrated wastewater treatment system for zero discharge, applicable to high-altitude and cold-climate highway service facilities at altitudes of at least 2000 meters and winter extreme temperatures not exceeding -20°C, includes: Collect operational data and water quality data as input parameters; Based on the input parameters, a BP neural network prediction method and a genetic algorithm optimization method are used to generate a baseline setpoint for the control parameters; Real-time monitoring of effluent water quality generates feedback parameters; Based on the deviation between the feedback parameters and the preset effluent water quality target value, a fuzzy control method is used to generate incremental corrections for the control setting parameters; The control setting parameters, synthesized from the baseline setpoint and the incremental correction of the control setting parameters, are converted into equipment operation commands, which are respectively applied to at least the aeration blower, internal return pump, flocculant metering pump, mixer, and filtration and backwashing device. The deviation between the feedback parameters and the preset effluent water quality target value or preset operating constraints is calculated, and a PID algorithm is used to correct the closed loop, applying amplitude constraints and rate of change limits to the synthesized control setting parameters.

2. The method of claim 1, wherein the method is used for the control of wastewater zero discharge of high-cold high-altitude highway service facilities. The input parameters include at least: five-day biochemical oxygen demand (BOD) of the influent, influent ammonia nitrogen concentration, water temperature, pH, mixed liquor suspended solids concentration, oxidation-reduction potential, effluent turbidity, flow rate, filtration differential pressure, and pool level; the feedback parameters include at least: five-day biochemical oxygen demand (BOD) of the effluent, effluent ammonia nitrogen concentration, effluent turbidity, real-time dissolved oxygen value, and filtration differential pressure; the control setting parameters include at least: target dissolved oxygen value, internal reflux ratio, flocculant dosage, stirring intensity, filtration speed, and backwashing cycle.

3. The method of claim 1, wherein the method is used for the zero discharge of sewage from a high-cold high-altitude highway service facility. The input parameters of the BP neural network prediction method are the five-day biochemical oxygen demand of the influent, the influent ammonia nitrogen concentration, water temperature, pH, mixed liquor suspended solids concentration, oxidation-reduction potential, effluent turbidity, flow rate, filtration pressure difference, and pool level. The output control parameters are the target dissolved oxygen value and the internal reflux ratio. The BP neural network has a three-layer structure, including an input layer, a hidden layer, and an output layer. The activation function of the hidden layer is the sigmoid function, the activation function of the output layer is a linear function, the training method is the backpropagation algorithm, and the loss function is the mean squared error (MSE). Its calculation process includes: Input layer to hidden layer: ; wherein, are input parameters: influent five-day biochemical oxygen demand, influent ammonia nitrogen concentration, water temperature, pH, mixed liquor suspended solids concentration, oxidation-reduction potential, effluent turbidity, flow rate, filtration pressure difference, and tank level, are hidden layer outputs, are weights from input layer to hidden layer, are biases, are Sigmoid functions, : number of hidden layer neurons, is a positive integer; : index of hidden layer neurons, from 1 to ; hidden layer to output layer: ; wherein, is an output parameter including a dissolved oxygen target value, an internal reflux ratio, is a weight from the hidden layer to the output layer, is a bias, is a linear function; Loss function: ; wherein, is the mean squared error, the first is the target output for the sample, the first is the predicted output for the sample, is the number of training samples, : training sample index, from 1 to .

4. The method of claim 1, wherein the method is used for the control of wastewater zero discharge of high-cold high-altitude highway service facilities. The input to the fuzzy control method is the deviation between the feedback parameters and the preset target values. Specifically, the input includes deviations in the effluent's five-day biochemical oxygen demand (BOD), effluent ammonia nitrogen concentration, effluent turbidity, and real-time dissolved oxygen values. The output control setting parameter is the internal reflux ratio correction value. The input and output fuzzy linguistic variables are divided into three levels: "low," "medium," and "high." The membership function uses a triangular function with the following form: The fuzzy control rules employ the Mamdani inference mechanism, and the defuzzification method is the centroid method. ； in, The input variables are normalized versions of the deviations in effluent biochemical oxygen demand (BOD) over five days, ammonia nitrogen concentration, turbidity, and dissolved oxygen real-time values, with values ​​ranging from [−1, 1]. :variable x Regarding fuzzy sets A The membership function values, a, b, c: the left endpoint, peak, and right endpoint of the triangle membership function, satisfying a≤b≤c, and the parameter values ​​corresponding to the three levels "low", "medium", and "high" are "low" (a, b, c) = (−1, −1, 0), "medium" (a, b, c) = (−1, 0, 1), and "high" (a, b, c) = (0, 1, 1); The dimensionless output value obtained from defuzzification is used to generate the internal reflux ratio correction value. Output the discretized values ​​within the universe of discourse. Output fuzzy set In taking values Membership degree of location Output the number of values ​​for the discretized universe of discourse.

5. The method for zero-discharge control of wastewater from high-altitude and cold-climate highway service facilities as described in claim 1, characterized in that: The genetic algorithm optimization method uses the following input parameters as the basic data for optimization calculation: five-day biochemical oxygen demand of influent, influent ammonia nitrogen concentration, water temperature, pH, mixed liquor suspended solids concentration, oxidation-reduction potential, effluent turbidity, flow rate, filtration pressure difference, and pool level. It uses effluent turbidity and filtration pressure difference as feedback parameters as filtration performance constraints to optimize the flocculant dosage, stirring intensity, filtration speed, and backwashing cycle in the control parameters. The genetic algorithm uses binary encoding to represent chromosomes, single-point crossover, and gene position flipping as the mutation method. The fitness function F is: ； in, The five-day biochemical oxygen demand removal rate. For ammonia nitrogen removal rate, Turbidity removal rate Energy consumption per unit This refers to the amount of medicine consumed per unit. For the filtration pressure difference constraint penalty term, α is the weighting coefficient of the five-day biochemical oxygen demand removal rate, β is the weighting coefficient of the ammonia nitrogen removal rate, γ is the weighting coefficient of the turbidity removal rate, λ is the weighting coefficient of the unit energy consumption, μ is the weighting coefficient of the unit reagent consumption, and ν is the weighting coefficient of the filtration pressure difference constraint penalty term.

6. The method for zero-discharge control of wastewater from high-altitude and cold-climate highway service facilities as described in claim 1, characterized in that: Calculate the deviation between the feedback parameters and the preset effluent water quality target value, adjust the control setting parameters, and use a PID algorithm for corrective closed-loop control, including: Its control output calculation formula is: ; And on Apply amplitude constraints and rate of change limits; The preset effluent water quality targets are: five-day biochemical oxygen demand ≤10mg / L, ammonia nitrogen concentration ≤5mg / L, and turbidity ≤5NTU; Preset operating constraints: Dissolved oxygen ≥ 2 mg / L, filtration pressure difference not exceeding the warning threshold; in, Control output / adjustment amount :error, Proportional gain, Integral gain Differential gain Error integral term, : Error differential term, Current time : Integral variable.

7. The method for zero-discharge control of wastewater from high-altitude and cold-climate highway service facilities as described in claim 1, characterized in that: The control setting parameters are converted into device operation commands and then applied to the execution devices in the following manner: The target dissolved oxygen value is achieved by adjusting the speed and air supply of the aeration blower to maintain the dissolved oxygen concentration in the reaction tank within a preset range. The internal recirculation ratio is controlled by adjusting the flow rate of the internal recirculation pump to control the proportion of aerobic effluent recirculated to the anoxic zone. The dosage of flocculant is adjusted by regulating the dosing rate of the flocculant metering pump to promote the formation of settleable flocs from suspended solids and colloidal particles. The stirring intensity is applied to the mixer by adjusting the speed of the mixer to ensure that the reagents and wastewater in the reaction tank or mixing tank are evenly mixed and kept in suspension; The filtration speed and backwashing cycle are adjusted by regulating the operating mode of the filtration and backwashing devices to control the hydraulic conditions of the filtration process and to initiate backwashing when the pressure difference reaches the threshold or the operating time reaches the set value.

8. The method for zero-discharge control of wastewater from service facilities on high-altitude and cold-climate highways as described in claim 4, characterized in that: The fuzzy control method also includes online learning and parameter update steps, specifically including: Real-time data on five-day biochemical oxygen demand, ammonia nitrogen concentration, turbidity, and dissolved oxygen in the effluent are continuously collected and stored in a historical database. Based on the preset effluent water quality target values, calculate the time series of deviations in five-day biochemical oxygen demand, ammonia nitrogen concentration, turbidity, and real-time dissolved oxygen values ​​in the effluent; Without changing the form of the triangular membership function and the three-level linguistic variable structure of "low", "medium" and "high", the left endpoint, peak and right endpoint of the triangular membership function corresponding to each input variable are slightly increased or decreased according to the recent deviation distribution; the adjustment range does not exceed the preset ratio, the order of the endpoints is kept such that the left endpoint is not greater than the peak and the peak is not greater than the right endpoint, and the endpoint values ​​are kept within the domain of discourse. The contribution of each rule in the four input three-level rule bases is statistically analyzed. The weight of rules with lower contribution is reduced, and the weight of rules with higher contribution is increased. The weight of the rules is limited to between zero and one. Set a fixed learning rate to limit the magnitude of a single parameter update, and set minimum, maximum, and endpoint order constraints for the membership function endpoints; The updated membership function and rule weights are used for short-term operation verification; when the effluent quality meets the preset effluent quality target value, the updated result is solidified into a new fuzzy control parameter set; when it does not meet the target value, it reverts to the previous solidified parameter set and reduces the subsequent update ratio. When data loss, sensor malfunction, or communication interruption occurs, online learning and parameter updates are paused, the most recent fixed parameter set is retained, and an alarm is triggered.

9. The method for zero-discharge control of wastewater from high-altitude and cold-climate highway service facilities as described in claim 1, characterized in that: It also includes a secure switching mechanism, specifically including: The safety switching mechanism will be triggered under any of the following circumstances: sensor failure or communication timeout; water tank level exceeding the limit; filtration pressure difference exceeding the warning threshold; effluent five-day biochemical oxygen demand, effluent ammonia nitrogen concentration, effluent turbidity failing to meet the preset effluent water quality target values, or dissolved oxygen real-time value failing to meet the preset operating constraints; ambient temperature or equipment compartment temperature lower than the antifreeze protection temperature. Pause the calculation output of the equipment operation command, switch to safe setting value operation, set the dissolved oxygen target value, internal reflux ratio, flocculant dosage, stirring intensity, filtration speed and backwash cycle to the corresponding safe setting values, and continue to execute amplitude constraints and rate of change limits; After the fault is cleared and the feedback parameters have continuously reached the preset effluent water quality target value for a preset period of time, automatic control, online learning, and parameter updates are restored; alarms are generated throughout the process and the triggering reasons and key parameters are recorded.

10. A zero-discharge control system for wastewater from highway service facilities in high-altitude and cold regions, characterized in that: include: The data acquisition module is used to collect operational data and water quality data as input parameters; The effluent water quality is monitored in real time to generate feedback parameters; The control module is used to calculate and generate the baseline setpoint of the control parameters based on the input parameters using a BP neural network prediction method and a genetic algorithm optimization method. Based on the deviation between the feedback parameters and the preset effluent water quality target value, the fuzzy control method is used to calculate and generate the incremental correction of the control setting parameters; The deviation between the feedback parameters and the preset effluent water quality target value or preset operating constraints is calculated, and the PID algorithm is used to correct the closed loop, applying amplitude constraints and rate of change limits to the synthesized control setting parameters. The execution module is used to convert the control setting parameters synthesized from the baseline setpoint and the incremental correction of the control setting parameters into equipment operation commands, which are respectively applied to at least the aeration blower, the internal return pump, the flocculant metering pump, the mixer, and the filtration and backwashing device.

Images