Alarm processing system for water outlet of industrial wastewater treatment station

Abstract

The invention discloses an alarm processing system for a water outlet of an industrial wastewater treatment station, and belongs to the technical field of industrial wastewater treatment, and the alarm processing system comprises a water quality on-line monitoring device integrating a plurality of sensor fusion settings and having a self-adaptive calibration algorithm; an embedded computer is arranged in the PLC control box and supports execution of a mathematical model and a machine learning algorithm so as to predict the water quality change trend; an edge computing technology is combined with an FPGA acceleration card; setting a block chain technology to construct a decentralized data recording mechanism; when the water quality exceeds the standard, the intelligent decision-making module automatically selects an emergency scheme based on a preset rule base and a dynamic adjustment strategy, and continuously improves a response strategy by using a Bayesian optimization algorithm to realize emergency treatment. The system can timely give an alarm to an operator on duty to find abnormal water discharge, can timely and automatically perform linkage treatment on the drainage problem, and does not allow excessive wastewater to continue to be discharged outwards.

Description

Technical Field

[0001] The invention belongs to the technical field of industrial wastewater treatment, in particular to an alarm processing system for a water outlet of an industrial wastewater treatment station. Background Art

[0002] Urban sewage treatment plants are generally categorized into three types: large, medium, and small, based on investment scale and wastewater treatment capacity. Large-scale plants treat over 200,000 tons per day, medium-scale plants treat 50,000 to 200,000 tons per day, and small-scale plants treat under 50,000 tons per day. In recent years, with the vigorous implementation of development policies and documents related to sewage treatment facility construction, these policies have significantly boosted the development of sewage treatment plants.

[0003] At present, small-scale wastewater treatment stations are mainly distributed in small towns or industrial parks. We are now mainly targeting small-scale wastewater treatment stations in industrial parks. Its characteristics are small treatment scale, large number, scattered distribution, simple treatment equipment and unstable water output. On-site staff are generally full-time non-professionals or part-time staff. They cannot promptly detect abnormalities in the daily treatment of wastewater treatment stations. In addition, the frequency of on-site inspections by wastewater treatment station staff is generally 1-2 hours. When the wastewater treatment system of the wastewater treatment station is abnormal, the online monitoring equipment of the water quality at the outlet will show that it does not meet the standards (the naked eye may not be able to tell whether the wastewater quality meets the discharge standards). The wastewater treatment equipment of the wastewater treatment station has been in operation. If the wastewater that exceeds the standard is not discovered in time, it will continue to be discharged to the outlet and then discharged into the municipal sewage network. The continuous discharge will affect the subsequent water bodies and pollute the surrounding environment.

[0004] Traditional systems typically rely on single sensors to measure water quality parameters. These sensors are easily affected by factors such as temperature, humidity, and vibration, resulting in significant measurement deviations. A single sensor cannot fully reflect changes in water quality in complex environments. The limitations of single sensors are particularly pronounced in highly volatile industrial environments. Inaccurate measurement results can cause the system to misjudge water quality conditions, leading to incorrect emergency measures or even overlooking excessive discharges, causing environmental pollution. Frequent manual calibration and adjustments increase operating costs and reduce system reliability. For example, at a chemical plant's wastewater treatment station, the pH sensor, affected by temperature, produced high measurement results. The system failed to promptly detect excessive acidic wastewater discharges, polluting surrounding rivers and damaging the ecological environment. The company faced significant fines and reputational damage. Summary of the Invention

[0005] The object of the present invention is to provide an alarm processing system for the water outlet of an industrial wastewater treatment station to solve the problems raised in the background technology.

[0006] To achieve the above-mentioned object, the present invention provides the following technical solution: an alarm processing system for the outlet of an industrial wastewater treatment station, comprising:

[0007] An online water quality monitoring device that integrates multiple sensor fusion settings and an adaptive calibration algorithm automatically corrects measurement deviations caused by environmental factors by comparing readings between different sensors in real time, ensuring data accuracy.

[0008] The PLC control box has a built-in embedded computer that supports the execution of mathematical models and machine learning algorithms to predict water quality trends;

[0009] Utilize edge computing technology combined with FPGA accelerator cards to quickly process large amounts of real-time data streams locally, reduce latency, and encrypt and transmit them to the cloud or remote servers through built-in security protocols to ensure information security.

[0010] Blockchain technology is used to build a decentralized data recording mechanism. All records are time-stamped and hashed to form an unalterable data link.

[0011] Among them, when the water quality exceeds the standard, the intelligent decision-making module automatically selects the emergency plan based on the preset rule library and dynamic adjustment strategy, and uses the Bayesian optimization algorithm to continuously improve the response strategy to achieve emergency handling.

[0012] In this solution, a remote monitoring platform based on the Internet of Things architecture is constructed in the system. The platform also integrates an image recognition module to analyze abnormal situations in the video stream and provide visual guidance to on-site staff through AR glasses;

[0013] The remote monitoring platform adopts a microservice architecture design. Each functional module is independently deployed and combined through the API gateway to support on-demand expansion of new features. The target detection in the image recognition module uses the YOLOv5 algorithm, whose formula is:

[0014] Among them, λ coord and λ noobj are the weights of coordinate loss and non-object confidence loss respectively; S is the number of grids; B is the number of bounding boxes per grid; and Indicator functions for the presence and absence of an object, respectively; x i ,y i are the center coordinates of the predicted bounding box; is the center coordinate of the ground-truth bounding box; C i and are the predicted and true confidence scores respectively.

[0016] This solution is preferred. The system has an intelligent analysis unit. The intelligent analysis unit uses a distributed deep learning framework to train a water quality prediction model. The model is deployed on edge cloud nodes and connected to PLC control boxes in various locations through a 5G low-latency network to ensure that all locations can obtain the latest model updates and service support. At the same time, federated learning technology is used to protect user privacy and avoid uploading sensitive data directly to the central server. For the model, the FedAvg algorithm is used:

[0017] Among them, w t represents the value of the global model parameters after the tth iteration, K is the set of clients participating in the learning, and n k is the number of samples for client k, are the local model parameters of client k after the tth iteration.

[0018] Preferably, the system also has an adaptive sound and light alarm module based on acoustic fingerprint recognition. The module establishes a database according to the sound characteristics of different areas. When an abnormality occurs, the module automatically matches the corresponding alarm mode according to the current environmental soundscape to ensure the effective transmission of the alarm information.

[0019] Preferably, in this solution, a mobile application is provided in the system for receiving alarm information and managing equipment, and a blockchain points incentive strategy is introduced. The application has a built-in intelligent assistant based on the knowledge graph.

[0020] In this solution, the PLC control box is preferably provided with an uninterruptible power supply and a solar auxiliary power supply module. The redundant power supply module of the system is designed with hot-swappable capability, allowing replacement of faulty components without interrupting system operation. The operating formula of the MPPT controller in the solar auxiliary power supply module is:

[0021] Among them, V mppt is the maximum power point voltage, V pv is the open circuit voltage of the photovoltaic array, P max is the power at the maximum power point, I pv is the maximum current of the photovoltaic array.

[0022] In the preferred embodiment of this solution, the wastewater treatment station has an intermediate water tank, and the submersible drainage pump in the intermediate water tank adopts magnetic bearing technology and liquid cooling, combined with dual protection measures of float level switch and ultrasonic level meter, and introduces a laser rangefinder as the third protection.

[0023] Preferably, the water quality online monitoring device has a self-cleaning function, uses high-pressure water flow or air flow to regularly clean the sensor surface, and the device shell is made of nano-coating material.

[0024] Preferably, the system is equipped with a soft starter and a variable frequency speed regulator, and the VFD is equipped with an energy feedback device. The VFD has a built-in preset fault prediction algorithm to identify possible faults in advance and take preventive measures.

[0025] Preferably, the system of this solution also includes: modular design and plug-and-play architecture, zero-trust architecture, and digital twin technology.

[0026] Compared with the prior art, the technical effects and advantages of the present invention are:

[0027] This alarm processing system for the outlet of an industrial wastewater treatment plant uses multi-sensor fusion and an adaptive calibration algorithm to address potential errors inherent in single sensors, significantly improving the accuracy and reliability of water quality parameter measurements. Automatic calibration reduces reliance on specialized personnel, lowering maintenance costs and operational complexity. TinyML efficiently runs complex models in resource-constrained environments, enabling real-time prediction of water quality trends and ensuring rapid response to anomalies. Edge computing enables local data processing, reducing transmission time and cloud processing latency, and improving system responsiveness.

[0028] This solution leverages FPGA accelerator cards to efficiently process large amounts of real-time data streams, ensuring initial analysis is completed in milliseconds, significantly improving the system's processing power and response speed. Data transmission is encrypted through built-in security protocols (such as TLS), ensuring information security and preventing data leakage or tampering. Blockchain technology establishes a decentralized data recording mechanism. All records are timestamped and hashed to form an unalterable data link, enhancing data security and transparency. All historical records cannot be tampered with, facilitating subsequent traceability and auditing, and improving system credibility and compliance.

[0029] This solution utilizes an uninterruptible power supply (UPS) and a solar-assisted power system, along with hot-swappable redundant power modules. This allows for the replacement of faulty components without system downtime, reducing downtime and maintenance costs. Furthermore, automatic fault detection and early warning capabilities prevent potential problems before they occur, further reducing the risk of unplanned downtime.

[0030] In summary, this solution can promptly alert on-duty personnel when abnormal water discharge is discovered, and can automatically and promptly handle drainage problems to prevent excessive wastewater from being discharged. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0032] Figure 1 This is an overall flow chart of an alarm processing system for an outlet of an industrial wastewater treatment station according to the present invention;

[0033] Figure 2 This is a branch and detailed flow chart of an alarm processing system for an outlet of an industrial wastewater treatment station according to the present invention. DETAILED DESCRIPTION

[0034] In the following description, numerous specific details are provided to provide a more thorough understanding of the present invention. However, it will be apparent to those skilled in the art that the present invention may be practiced without one or more of these details. In other instances, certain technical features well known in the art are not described to avoid confusion with the present invention.

[0035] Unless otherwise defined, the directions of up, down, left, right, front, back, inside and outside involved in this document are based on the directions of up, down, left, right, front, back, inside and outside shown in the figures of the present invention, and are explained here together.

[0036] This embodiment provides Figure 1 and Figure 2 The alarm processing system for the water outlet of an industrial wastewater treatment station shown includes:

[0037] Online water quality monitoring equipment that integrates multiple sensor fusion settings and an adaptive calibration algorithm automatically corrects measurement deviations caused by environmental factors by comparing the reading differences between different sensors in real time, ensuring data accuracy. Multiple sensors (such as pH, chemical oxygen demand (COD), biochemical oxygen demand (BOD), ammonia nitrogen, etc.) are installed in the online water quality monitoring equipment. The sensor readings are transmitted to the PLC control box via RS485 or Modbus protocol. The PLC's built-in algorithm compares the differences in readings from different sensors in real time and automatically adjusts the deviations. For example, when the pH sensor shows an abnormality, the system will correct it based on the data from other sensors (such as the conductivity sensor) to ensure the accuracy of the final output data. Through multi-sensor fusion and adaptive calibration algorithms, the error problem that may exist in a single sensor is solved, significantly improving the accuracy and reliability of water quality parameter measurements. Automatic calibration reduces dependence on professionals, lowers maintenance costs and operational difficulty.

[0038] The PLC control box has a built-in high-performance embedded computer with a dedicated operating system and real-time computing environment. This supports the execution of mathematical models and machine learning algorithms. Specifically, it uses lightweight neural networks (such as TinyML) to efficiently run in resource-constrained environments to predict water quality trends. A high-performance embedded computer installed in the PLC control box, running a dedicated operating system (such as Linux) and a real-time computing environment (such as RT-Linux), integrates a TinyML framework (such as TensorFlowLite for Microcontrollers), and pre-trains a water quality prediction model and deploys it to the PLC. Water quality parameters are collected in real time and fed into the TinyML model for prediction. For example, if a continuous decrease in pH is detected, the TinyML model predicts that an exceedance is likely within the next few hours, issuing an early warning. TinyML enables efficient execution of complex models in resource-constrained environments, enabling real-time prediction of water quality trends and ensuring rapid response to anomalies. Edge computing enables data processing to be performed locally, reducing transmission time and cloud processing latency, thereby improving system responsiveness.

[0039] Edge computing technology combined with FPGA accelerator cards can rapidly process large amounts of real-time data streams locally, reducing latency. These data are then encrypted and transmitted to the cloud or remote servers via built-in security protocols (such as TLS) to ensure information security. By integrating an FPGA accelerator card (such as the Xilinx Zynq series) within the PLC control box, the FPGA is responsible for rapidly processing large amounts of real-time data streams, reducing latency. The data is encrypted with TLS and transmitted to the cloud server via a 5G module. For example, the FPGA processes high-frequency sampled data from multiple sensors, ensuring initial analysis within milliseconds, and then uploads the results to the cloud for in-depth analysis.

[0040] Blockchain technology is used to build a decentralized data logging mechanism. This not only records operation logs and water quality monitoring data, but also stores equipment maintenance history, fault reports, and other information. All records are timestamped and hashed to form an unalterable data link. Blockchain node software (such as Hyperledger Fabric) is installed in the PLC control box. Each operation log and water quality monitoring data is timestamped and hashed, and data blocks are linked sequentially to form an unalterable data link. When a pump starts, this event is recorded as a block, and all subsequent operations are linked to this block, ensuring that historical records cannot be tampered with.

[0041] Among them, when the water quality exceeds the standard, the intelligent decision-making module automatically selects the optimal emergency plan based on the preset rule base and dynamic adjustment strategy, including but not limited to adjusting the treatment process parameters, starting the backup treatment unit or notifying the relevant responsible persons, and using the Bayesian optimization algorithm to continuously improve the response strategy to achieve accurate and efficient emergency treatment. The intelligent decision-making module has a built-in rule base (such as IF-THEN rule set) and a dynamic adjustment strategy. When the water quality exceeds the standard, the module selects the optimal emergency plan (such as closing the water inlet valve and starting the backup treatment unit) according to the current situation. The Bayesian optimization algorithm continuously updates the rule base to improve the response accuracy. If the COD exceeds the standard, the system immediately closes the water inlet valve and starts the backup treatment unit, and notifies the on-duty personnel.

[0042] In this embodiment, a remote monitoring platform based on the Internet of Things (IoT) architecture is built in the system. In addition to conventional functions, this platform also integrates an AI-driven image recognition module, which can analyze abnormal conditions (such as equipment leakage and foreign object intrusion) in the video stream and provide visual guidance to on-site workers through AR glasses;

[0043] The remote monitoring platform adopts a microservice architecture design, and each functional module is independently deployed and combined through the API gateway. It supports the on-demand expansion of new features, such as access to new operation and maintenance methods such as drone inspections and robot inspections, which greatly improves the flexibility and adaptability of the system; the remote monitoring platform server is installed with image recognition software (such as OpenCV combined with YOLOv5), and the on-site camera captures the video stream and transmits it to the platform through the 5G network. The platform analyzes abnormal conditions in the video stream (such as equipment leakage, foreign object intrusion), and sends guidance information to the staff through AR glasses. When a liquid leak is detected in a certain pipeline, the platform generates an AR guidance map to instruct maintenance personnel on how to seal the leak point. The remote monitoring platform uses Docker containers to containerize various functional modules (such as video analysis, alarm notification, and data analysis), uses Kubernetes clusters to manage containers, supports the on-demand expansion of new features, and the API gateway uniformly manages external requests to ensure flexible combination of different modules. When it is necessary to add a drone inspection function, you only need to deploy a new Docker container and integrate it into the existing system through the API gateway.

[0044] The target detection in the image recognition module uses the YOLOv5 algorithm, and its formula is:

[0045] Among them, λ coord and λ noobj are the weights of coordinate loss and non-object confidence loss respectively; S is the number of grids; B is the number of bounding boxes per grid; and Indicator functions for the presence and absence of an object, respectively; x i ,y i are the center coordinates of the predicted bounding box; is the center coordinate of the ground-truth bounding box; C i and are the predicted and true confidence scores respectively.

[0047] In this embodiment, the system has an intelligent analysis unit that uses a distributed deep learning framework to train a water quality prediction model. It not only uses large-scale historical data as training samples, but also incorporates a real-time feedback mechanism. That is, the actual results after each model prediction are re-input into the training set, enabling the model to continuously self-optimize. The model is deployed on edge cloud nodes and connected to PLC control boxes in various locations via a 5G low-latency network to ensure that all locations can obtain the latest model updates and service support. At the same time, federated learning technology is used to protect user privacy and prevent sensitive data from being directly uploaded to the central server. For the model, the FedAvg algorithm is used:

[0048] Among them, w t represents the value of the global model parameters after the tth iteration, K is the set of clients participating in the learning, and n k is the number of samples for client k, are the local model parameters of client k after the tth iteration.

[0049] Specifically, the edge cloud node server is installed with a distributed deep learning framework (such as TensorFlowFederated), and each wastewater treatment site regularly uploads local water quality data to the edge cloud node. The edge cloud node aggregates the data and trains the water quality prediction model, while also adding a real-time feedback mechanism. For example, at 2 a.m. every day, each site uploads the previous day's water quality data, and the edge cloud node uses this data to update the water quality prediction model; the edge cloud node server is installed with federated learning client software (such as TFF-TensorFlow Federated), and each wastewater treatment site only uploads model updates rather than original data to protect user privacy. The 5G low-latency network connects to PLC control boxes in various locations to ensure that all locations can obtain the latest model updates and service support. For example, the model update of site A only includes weight changes and does not involve specific water quality data. After uploading to the edge cloud node, it is merged with updates from other sites to form a global model.

[0050] In this embodiment, the system also has an adaptive sound and light alarm module based on acoustic fingerprint recognition. This module establishes a database based on the sound characteristics of different areas. When an abnormality occurs, the module will automatically match the corresponding alarm mode according to the current environmental soundscape to ensure the effective communication of the alarm information. It also has a two-way voice communication device and an emotion analysis engine, which can not only realize human-computer dialogue, but also understand the emotional state of the on-duty personnel, adjust the communication method in time, and improve the user experience. In addition, the system supports multi-language instant translation to facilitate international team collaboration.

[0051] Specifically, the sound and light alarm device is equipped with a microphone array to collect environmental sounds in real time. It has a built-in acoustic fingerprint database and builds models based on the sound characteristics of different areas. When an abnormality occurs, the system matches the most appropriate alarm mode to ensure that the alarm information is effectively conveyed. For example, when a sudden increase in noise is detected in the pump room, the system determines it is a mechanical failure, issues a high-pitched alarm, and flashes a red light. The alarm system is equipped with a two-way voice communication module (such as an IP phone or intercom interface), and a sentiment analysis engine (such as using natural language processing (NLP) technology) to analyze the emotional state of the on-duty personnel in real time, adjust the communication method based on the emotional state, and enhance the user experience. For example, when the on-duty personnel speak in a hurried tone, the system infers that they are in an emergency state and quickly switches to priority processing mode to provide more direct assistance.

[0052] In this embodiment, the system includes a mobile application for receiving alarm information and managing devices, and incorporates a blockchain-based points incentive strategy. Users who complete specific tasks (such as regular inspections and report submissions) earn points, which can be redeemed for value-added services or product discounts, enhancing user engagement. The application also includes a built-in knowledge graph-based intelligent assistant that not only answers frequently asked questions but also recommends personalized training courses and technical documentation based on user behavior patterns, acting as a personal mentor for the user. The application also incorporates a built-in knowledge graph database (such as Neo4j) to store frequently asked questions and technical documentation. The intelligent assistant then recommends personalized training courses and technical documentation based on user behavior patterns. If a user frequently searches for solutions to a specific type of problem, the intelligent assistant recommends relevant training courses, helping the user acquire additional skills.

[0053] In this embodiment, the PLC control box is equipped with an uninterruptible power supply (UPS) and a solar auxiliary power supply module. The solar panel is integrated with a maximum power point tracking (MPPT) controller to ensure that the highest efficiency power is always output under different lighting conditions. The system's redundant power supply module is designed with hot-swappable capabilities, allowing faulty components to be replaced without interrupting system operation, greatly improving the system's availability and reliability. At the same time, the system has automatic fault detection and early warning functions to prevent potential problems in advance. The operating formula of the MPPT controller in the solar auxiliary power supply module is:

[0054] Among them, V mppt is the maximum power point voltage, V pv is the open circuit voltage of the photovoltaic array, P max is the power at the maximum power point, I pv is the maximum current of the photovoltaic array.

[0055] Specifically, a UPS is installed in the PLC control box to provide short-term power support. Solar panels are installed outdoors and equipped with an MPPT controller to optimize energy conversion efficiency. The MPPT controller adjusts the operating voltage of the photovoltaic panels in real time to ensure maximum power output. During daytime hours when there is ample sunlight, the solar panels power the system; at night or on cloudy days, the UPS takes over to ensure stable system operation. The redundant power module design allows for the replacement of faulty components without downtime, and automatic fault detection and early warning functions prevent potential problems in advance. If the main power module detects an anomaly, it automatically switches to the backup power module and triggers an alarm to notify maintenance personnel to replace the faulty component promptly.

[0056] In this embodiment, the wastewater treatment station is provided with an intermediate water tank. The submersible drainage pump in the intermediate water tank adopts magnetic levitation bearing technology and liquid cooling, which effectively reduces mechanical wear and heat accumulation and extends the service life. Combined with the dual protection measures of the float liquid level switch and the ultrasonic liquid level meter, a laser rangefinder is introduced as the third protection to achieve more precise liquid level control, and the Kalman filter algorithm is used to fuse the data from the three sources to further improve the measurement accuracy. When the liquid level approaches the set upper limit, the system integrates the data from the three sensors to ensure precise control of the drainage pump startup and avoid overflow accidents.

[0057] In this embodiment, the online water quality monitoring device features a self-cleaning function, using high-pressure water or airflow to regularly clean the sensor surface to maintain sensor sensitivity. Furthermore, the device's housing is coated with a nano-coating material, offering excellent corrosion resistance and suitability for use in harsh environments. The device supports wireless updates, but unlike traditional OTA (over-the-air) technology, it uses differential OTA (over-the-air) technology, transmitting only the differences between the previous and next versions. This significantly reduces data transmission, speeds up updates, and reduces bandwidth usage.

[0058] In this embodiment, the system is equipped with a soft starter and a variable frequency drive (VFD). The VFD is equipped with an energy feedback device. When the motor decelerates, the generated electrical energy is recovered and stored in a supercapacitor. It is then released to other loads when needed, forming a closed-loop energy management system. The VFD has a built-in preset fault prediction algorithm. By monitoring parameters such as current and voltage in real time, it can identify potential faults in advance and take preventive measures such as reducing the speed or shutting down for maintenance, thereby improving the stability and safety of the entire system. When the motor current increases abnormally, the VFD predicts the potential fault and takes measures (such as reducing the speed or shutting down for maintenance) in advance to prevent the fault from escalating. When the current fluctuation frequency exceeds the normal range, the system issues an early warning and recommends preventive maintenance to avoid sudden failures that affect production.

[0059] In this embodiment, the system also includes modular design and plug-and-play architecture, zero-trust architecture, and digital twin technology;

[0060] Specifically, the modular design and plug-and-play architecture: The system utilizes a modular design, with components interconnected via standardized interfaces and hot-swappable operation. This allows for the replacement or addition of new devices without shutting down the entire system, simplifying installation, maintenance, and upgrades. This design not only improves system flexibility and scalability, but also reduces downtime and enhances system reliability. When a sensor needs to be replaced, there's no need to shut down the entire system; simply disconnect the corresponding interface and insert the new sensor to continue operation.

[0061] Specifically, the Zero Trust Architecture: To safeguard network security, the system integrates a Zero Trust Architecture, abandoning the traditional internal and external network boundary protection model and subjecting every access request to strict identity authentication and permission checks. This architecture continuously monitors and analyzes network traffic, detecting abnormal behavior in real time, effectively preventing unauthorized access and potential security threats, and ensuring the security of data transmission and storage. When remote users attempt to log in to the system, they must pass two-factor authentication to ensure that only authorized personnel can access sensitive data.

[0062] Specifically, the system incorporates digital twin technology, creating a virtual model of each physical device and synchronizing its operating status in real time on the cloud. These virtual models can simulate and predict changes in device performance, proactively identify potential failures, optimize maintenance plans, and test new control strategies in a virtual environment, reducing trial-and-error costs. Furthermore, digital twin technology supports remote monitoring and management, improving operational efficiency and decision-making accuracy. When a water pump is predicted to be nearing its maintenance cycle, the system notifies maintenance personnel to prepare the necessary parts in advance, optimizing maintenance plans.

[0063] It should be noted that, in this article, relational terms such as one and two are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions. The sentence "including an element defined by ... does not exclude the presence of other identical elements in the process, method, article or device that includes the element."

[0064] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An alarm processing system for the outlet of an industrial wastewater treatment station, characterized in that: include: The online water quality monitoring device integrates multiple sensor fusion settings and has an adaptive calibration algorithm. By comparing the reading differences between different sensors in real time, it automatically corrects the measurement deviation caused by environmental factors to ensure data accuracy; The PLC control box has a built-in embedded computer that supports the execution of mathematical models and machine learning algorithms to predict water quality trends; Utilize edge computing technology combined with FPGA acceleration cards to quickly process large amounts of real-time data streams locally, reduce latency, and encrypt and transmit them to the cloud or remote servers through built-in security protocols to ensure information security; Blockchain technology is used to build a decentralized data recording mechanism. All records are time-stamped and hashed to form an unalterable data link. Among them, when the water quality exceeds the standard, the intelligent decision-making module automatically selects the emergency plan based on the preset rule base and dynamic adjustment strategy, and uses the Bayesian optimization algorithm to continuously improve the response strategy to achieve emergency handling.

2. The alarm processing system for the water outlet of an industrial wastewater treatment station according to claim 1, characterized in that: The system builds a remote monitoring platform based on the Internet of Things architecture, which also integrates an image recognition module to analyze abnormal situations in video streams and provide visual guidance to on-site staff through AR glasses; The remote monitoring platform adopts a microservice architecture design. Each functional module is deployed independently and combined through an API gateway to support on-demand expansion of new features. The target detection in the image recognition module uses the YOLOv5 algorithm, and its formula is: Among them, λ coord and λ noobj are the weights of coordinate loss and non-object confidence loss respectively; S is the number of grids; B is the number of bounding boxes per grid; and Indicator functions for the presence and absence of an object, respectively; x i ,y i are the center coordinates of the predicted bounding box; is the center coordinate of the ground-truth bounding box; C i and are the predicted and true confidence scores respectively.

3. The alarm processing system for the water outlet of an industrial wastewater treatment station according to claim 2, characterized in that: The system has an intelligent analysis unit, which uses a distributed deep learning framework to train a water quality prediction model. The model is deployed on edge cloud nodes and connected to PLC control boxes in various locations through a 5G low-latency network to ensure that all locations can obtain the latest model updates and service support. At the same time, federated learning technology is used to protect user privacy and avoid uploading sensitive data directly to the central server. For the model, the FedAvg algorithm is used: Among them, w t represents the value of the global model parameter after the tth iteration, K is the set of clients participating in the learning, and n k is the number of samples of client k, are the local model parameters of client k after the tth iteration.

4. The alarm processing system for the water outlet of an industrial wastewater treatment station according to claim 3 is characterized in that: The system also has an adaptive sound and light alarm module based on acoustic fingerprint recognition. The module establishes a database based on the sound characteristics of different areas. When an abnormality occurs, the module will automatically match the corresponding alarm mode according to the current environmental soundscape to ensure the effective communication of the alarm information.

5. The alarm processing system for the water outlet of an industrial wastewater treatment station according to claim 4, characterized in that: A mobile application is provided in the system for receiving alarm information and managing equipment, and a blockchain points incentive strategy is introduced. The application has a built-in intelligent assistant based on a knowledge graph.

6. The alarm processing system for the water outlet of an industrial wastewater treatment station according to claim 5, characterized in that: The PLC control box is provided with an uninterruptible power supply and a solar auxiliary power supply module. The redundant power supply module of the system is designed with hot-swap capability, allowing the replacement of faulty components without interrupting the operation of the system. The operating formula of the MPPT controller in the solar auxiliary power supply module is: Among them, V mppt is the maximum power point voltage, V pv is the open circuit voltage of the photovoltaic array, P max is the power at the maximum power point, I pv is the maximum current of the PV array.

7. The alarm processing system for the water outlet of an industrial wastewater treatment station according to claim 6, characterized in that: The wastewater treatment station is provided with an intermediate water tank. The submersible drainage pump in the intermediate water tank adopts magnetic bearing technology and liquid cooling, combined with double protection measures of float liquid level switch and ultrasonic liquid level meter, and introduces a laser rangefinder as the third protection.

8. The alarm processing system for the water outlet of an industrial wastewater treatment station according to claim 7, characterized in that: The water quality online monitoring device has a self-cleaning function and uses high-pressure water flow or air flow to regularly clean the sensor surface. The device shell is made of nano-coating material.

9. The alarm processing system for the water outlet of an industrial wastewater treatment station according to claim 8, characterized in that: The system is equipped with a soft starter and a variable frequency speed regulator, and the VFD is equipped with an energy feedback device. The VFD has a built-in preset fault prediction algorithm to identify possible faults in advance and take preventive measures.

10. The alarm processing system for the water outlet of an industrial wastewater treatment station according to claim 9, characterized in that: The system also includes: modular design and plug-and-play architecture, zero-trust architecture, and digital twin technology.

Images