A sewage treatment equipment fault monitoring method based on cloud computing

Abstract

The present invention discloses a sewage treatment equipment fault monitoring method based on cloud computing, which specifically relates to the technical field of cloud computing. The specific steps include collecting water signals and electrical signals of water treatment equipment through sensors and transmitting them to a cloud computing platform, using a distributed database of the cloud computing platform to store the water signals and electrical signals and pre-process the data, processing and analyzing the water signals and electrical signals, using a machine learning algorithm to build a fault detection model, using historical fault data and regular operation data of the water treatment equipment for label training, evaluating the performance of the fault detection model through cross-validation, using the monitoring service of the cloud computing platform to monitor the equipment operation status in real time, predicting the operation data of the water treatment equipment according to the fault detection model, and triggering an alarm mechanism when an abnormal fault of the water treatment equipment is predicted, thereby effectively solving the problem of shortage of data processing capacity in large-scale high-standard sewage treatment plants.

Description

Technical Field

[0001] The present invention relates to the technical field of cloud computing, and more specifically, to a sewage treatment equipment fault monitoring method based on cloud computing. Background Art

[0002] Traditional sewage treatment processes usually use PID control to adjust key parameters to ensure the stability and effectiveness of the treatment process. PID control measures process variables in real time through sensors and feeds the measured values ​​back to the PID controller in real time. The PID controller calculates the control quantity based on the error between the set value and the actual value. The PID controller outputs a control signal based on the control quantity and transmits it to the actuator to adjust the control variable to reduce the error. The PID control method relies on the traditional experience of the staff and has limited effect on complex systems with nonlinear time-varying. For small-scale sewage treatment plants, it can be optimized by combining fuzzy control or adaptive control. However, in large-scale sewage treatment plants with high standards, the accumulated iterative error of sensor signals will interfere with the accuracy and efficiency of PID control. In order to ensure the safe and stable operation of water treatment, signal detection and verification work will take a lot of time, and the data processing volume exceeds the limit that the water system can support. Traditional experience support is not effective, and the response of control efficiency is too slow. Fault monitoring and maintenance of water treatment equipment are out of the effective range of experience support, and it is difficult to meet the high dynamic performance requirements of large-scale, high-standard real-time water data processing.

[0003] In order to solve the above defects, a technical solution is now proposed. Summary of the invention

[0004] The purpose of the present invention is to provide a sewage treatment equipment fault monitoring method based on cloud computing to solve the problems in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solution: a sewage treatment equipment fault monitoring method based on cloud computing, the specific steps comprising:

[0006] The water signal and electrical signal of the water treatment equipment are collected through sensors, and the water signal and electrical signal are transmitted to the cloud computing platform through the network;

[0007] The distributed database of the cloud computing platform is used to store water signals and electrical signals, and the data is pre-processed. The real-time processing framework of the cloud computing platform is used to process and analyze the water signals and electrical signals.

[0008] Use machine learning algorithms to build fault detection models, use historical fault data and regular operation data of water treatment equipment for label training, and evaluate the performance of fault detection models through cross-validation;

[0009] The monitoring service of the cloud computing platform is used to monitor the equipment operation status in real time, and the operating data of the water treatment equipment is predicted based on the fault detection model. When an abnormal fault of the water treatment equipment is predicted, an alarm mechanism is triggered.

[0010] Preferably, the water signals collected by the sensor include flow rate, liquid level, pH value, dissolved oxygen concentration, and temperature, and the electrical signals include voltage, current, and frequency.

[0011] Preferably, the method for processing data using the real-time processing framework of the cloud computing platform is:

[0012] Fill missing data or delete data records containing missing values, identify and handle data outliers;

[0013] Use low-pass filtering, high-pass filtering or band-pass filtering to remove noise from the data, decompose the signal through wavelet transform, and remove the high-frequency noise part of the signal;

[0014] Transform the data to a standard normal distribution, usually by subtracting the mean and dividing by the standard deviation so that the data has a distribution with mean 0 and standard deviation 1. Scale the data to a specific range, the specific range is [ 0,1 ] or [ -1,1 ] ;

[0015] Select model features and extract features from the original data, including frequency domain features and time domain features;

[0016] Smooth the data by moving average or exponential smoothing, and discretize the continuous data.

[0017] Preferably, the method of using a machine learning algorithm to construct a fault detection model for fault detection is:

[0018] Connect sensor data to the data stream processing system through the IoT gateway, and convert data formats of different sensors into a standardized format;

[0019] Use buffers to temporarily store data, smooth data traffic, and use message queues to manage data flows;

[0020] Use the real-time processing framework to process and analyze data streams in real time, group data streams by time windows or sliding windows, and perform aggregation operations;

[0021] Use algorithms in the real-time processing framework to perform anomaly detection, identify abnormal data points, and input real-time data into the trained machine learning model for fault prediction and early warning;

[0022] The processed real-time data is stored in the real-time database, and the data is stored in hot and cold tiers according to the real-time requirements of the data.

[0023] Preferably, the method for generating the fault detection model is:

[0024] Obtain water signal and electrical signal data collected by sensors, as well as operation log data of water treatment equipment, pre-process missing values ​​and abnormal values ​​in the data, and normalize or standardize the data;

[0025] According to the operation log of the water treatment equipment, the fault points in the data are marked, the data is divided into normal and fault states, and the data is divided into training set, verification set, and test set;

[0026] The time series data collected by the sensor is used as the original feature, and the derived features are calculated based on the original features. The derived features include the mean, extreme value and variability of the original data;

[0027] Use sliding window technology to divide time series data into multiple overlapping subsequences, each subsequence as a sample;

[0028] Construct a fault detection model based on long short-term memory network. The input layer is time series data. Each time step corresponds to a data point of the sensor. LSTM is stacked with multiple LSTM units. The fully connected layer connects the LSTM layer and the output layer. The number of nodes and activation function of the output layer are set according to the prediction target, and the fault prediction is output.

[0029] Use cross entropy as the loss function of the fault detection model and apply optimization algorithms to accelerate model convergence;

[0030] Divide the data into small batches for training, detect the loss value of the validation set, stop training when the validation loss no longer decreases, and use L2 regularization or dropout technology to reduce overfitting;

[0031] Use evaluation metrics and confusion matrices to evaluate the fault detection model, adjust the model's hyperparameters through grid search or random search, and adjust the model structure based on the verification results.

[0032] Preferably, the calculation method of the derived feature is:

[0033] The vibration frequency of the calibration equipment is Vf, the water quality optimization efficiency coefficient is Op, and the data return coefficient is Ec. The derived characteristic is De, and the calculation expression is: In the formula, the equipment vibration frequency is the vibration frequency of the water treatment equipment, the water quality optimization efficiency coefficient is the ratio of the water quality index after physical and chemical treatment to the preset target water quality index, the data return coefficient is the ratio of missing values, abnormal values ​​to the total data value in the data, α and β are the weight coefficients of the ratio of equipment vibration frequency, water quality optimization efficiency coefficient and data return coefficient, and α and β are both positive numbers.

[0034] Preferably, the construction logic of the alarm mechanism is:

[0035] A preset threshold level is set, a first alarm threshold and a second alarm threshold are calibrated, and the failure prediction probability of the first alarm threshold is less than the second alarm threshold. When it is detected that the failure probability of the water treatment equipment exceeds the first alarm threshold and is lower than the second alarm threshold, the water treatment equipment is marked to prompt the staff to pay attention to the water treatment equipment and record the alarm log;

[0036] When it is detected that the failure probability of the water treatment equipment exceeds the second alarm threshold, an alarm notification is issued to inform the maintenance team to perform emergency treatment on the water treatment equipment and record the alarm log.

[0037] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0038] The present invention solves the problem of shortage of data processing capacity for large-scale sewage treatment plants with high-standard water discharge by processing the status data of water treatment equipment on the cloud. By constructing a long-short term memory network and applying a fault detection model to predict faults of water treatment equipment, the problem of production damage and profit loss caused by alarm after equipment damage is effectively avoided. The artificial intelligence algorithm solves the problem that the existing PID control is difficult to fully cover the water system and respond efficiently, further improving the level of automated management of the water system and reducing the management burden of staff. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0040] Figure 1 The present invention is a flow chart of the method. DETAILED DESCRIPTION

[0041] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0042] Example 1: Please refer to Figure 1 As shown, the present invention is a sewage treatment equipment fault monitoring method based on cloud computing, and the specific steps include:

[0043] The water signal and electrical signal of the water treatment equipment are collected through sensors, and the water signal and electrical signal are transmitted to the cloud computing platform through the network;

[0044] The distributed database of the cloud computing platform is used to store water signals and electrical signals, and the data is pre-processed. The real-time processing framework of the cloud computing platform is used to process and analyze the water signals and electrical signals.

[0045] Use machine learning algorithms to build fault detection models, use historical fault data and regular operation data of water treatment equipment for label training, and evaluate the performance of fault detection models through cross-validation;

[0046] The monitoring service of the cloud computing platform is used to monitor the equipment operation status in real time, and the operating data of the water treatment equipment is predicted based on the fault detection model. When an abnormal fault of the water treatment equipment is predicted, an alarm mechanism is triggered.

[0047] Traditional sewage treatment processes usually use PID control to adjust key parameters. PID controller is an important tool in automatic control systems, which is used to adjust various physical parameters, such as flow, pressure, liquid level, temperature, etc. Take dissolved oxygen control as an example. In the biological treatment stage, the dissolved oxygen concentration in the aeration tank must be controlled to ensure the normal growth and metabolism of microorganisms. Dissolved oxygen control first measures the dissolved oxygen concentration in the water with a dissolved oxygen sensor. The PID controller adjusts the operating status of the aeration equipment according to the error between the set target dissolved oxygen concentration and the actual concentration. The aeration equipment includes blowers, aeration heads and other equipment. The blowers and aeration heads adjust the aeration volume according to the output signal of the PID controller. The PID control algorithm is simple to implement and easy to adjust. It is widely used in industry and has a low application base cost. However, the parameter adjustment of the PID controller depends largely on the experience of the staff. Improper adjustment may cause system instability or slow response. For rapidly changing processes, the response speed of PID control may be slow and cannot meet high dynamic performance requirements.

[0048] The water signals and electrical signals of the water treatment equipment are collected through sensors, and the water signals and electrical signals are transmitted to the cloud computing platform through the network. The water signals collected by the sensors include flow, liquid level, pH value, dissolved oxygen concentration, and temperature, and the electrical signals include voltage, current, and frequency. The acquired water signals and electrical signals are uploaded to the cloud computing platform through the network, and the data are transmitted using standardized data transmission protocols to ensure the reliability and security of data transmission;

[0049] Standardized data transmission protocols include MQTT, HTTP, CoAP and other protocols;

[0050] The cloud computing platform has many characteristics and advantages. Through the cloud computing platform, the operating status of equipment can be monitored in real time, and data can be quickly acquired and processed. When equipment fails, the cloud platform can trigger the alarm mechanism in real time and notify the management personnel in time. The cloud computing platform provides large-scale data storage capabilities and can store massive data from various sensors. With the powerful computing power of cloud computing, it can efficiently process and analyze big data and support the training and operation of complex machine learning and deep learning models. By using artificial intelligence technology on the cloud platform, an intelligent fault detection model can be built to improve the accuracy and efficiency of fault detection. The model can be continuously optimized according to new data and feedback to improve detection accuracy. Cloud computing resources are allocated on demand, and computing and storage resources can be dynamically expanded or reduced according to demand. It can be flexibly deployed in different regions and environments to support cross-regional water management.

[0051] Compared with traditional water management methods, cloud computing platforms support on-demand payment and do not require high upfront investment. Traditional methods require a large investment in hardware equipment and infrastructure. Cloud services are provided and maintained by a third party, and users do not need to manage and maintain servers and infrastructure by themselves, which can reduce operation and maintenance costs. Cloud computing platforms can be quickly deployed and used, avoiding the complex installation and configuration process in traditional methods. System updates and function upgrades can be quickly completed through the cloud platform, avoiding the tedious on-site upgrade work in traditional methods. Cloud computing platforms have high availability and fault tolerance mechanisms to ensure the stable operation of the system. Cloud platforms also provide data backup and disaster recovery services to ensure data security and system reliability. The platform can centrally manage and integrate data from different sources, support comprehensive data analysis and decision-making. Different regions and departments can share data and work together through the cloud platform to improve work efficiency and collaboration capabilities. The cloud computing platform supports various new technologies and applications, such as the Internet of Things, edge computing, AI, etc., to promote the innovative development of water systems. The cloud platform provides rich APIs and services, supports flexible functional expansion and customized development. By using cloud computing technology, water systems can achieve more efficient and intelligent equipment fault detection and early warning, improve the system's operating efficiency and reliability, and reduce management costs and risks. These advantages make cloud computing technology have broad application prospects and significant competitive advantages in modern water management.

[0052] The distributed database of the cloud computing platform is used to store water signals and electrical signals, and the data is pre-processed. The real-time processing framework of the cloud computing platform is used to process and analyze the water signals and electrical signals.

[0053] The cloud computing platform provides distributed databases and data lakes for storing massive sensor data, including data storage services such as Amazon S3 and Azure Blob Storage. The process of preprocessing data includes data cleaning, data denoising, data standardization and normalization, feature engineering, and data changes. The specific steps are as follows:

[0054] Fill missing data or delete data records containing missing values, identify and handle data outliers;

[0055] Use low-pass filtering, high-pass filtering or band-pass filtering to remove noise from the data, decompose the signal through wavelet transform, and remove the high-frequency noise part of the signal;

[0056] Transform the data to a standard normal distribution, usually by subtracting the mean and dividing by the standard deviation so that the data has a distribution with mean 0 and standard deviation 1. Scale the data to a specific range, the specific range is [ 0,1 ] or [ -1,1 ] ;

[0057] Select model features and extract features from the original data, including frequency domain features and time domain features;

[0058] Smooth the data by moving average or exponential smoothing, and discretize the continuous data.

[0059] Common methods for dealing with missing values ​​include mean filling, median filling, interpolation, etc. Outlier detection and processing can use statistical methods or machine learning-based anomaly detection methods;

[0060] Statistical methods include Z-score and interquartile range, and anomaly detection methods based on machine learning include isolation forest and LOF algorithms;

[0061] Data normalization can improve the comparability between different features. Commonly used methods include minimum-maximum normalization and quantile normalization.

[0062] Common methods of feature selection include correlation analysis, principal component analysis, LASSO regression, etc. The features extracted by feature extraction include the mean, standard deviation, kurtosis, and kurtosis of the vibration signal;

[0063] Smoothing data can reduce the impact of random fluctuations, discretize continuous data into several intervals or categories, and improve the stability and robustness of the model.

[0064] The real-time processing framework of the cloud computing platform is used to process and analyze water signals and electrical signals. The continuously flowing data streams are processed and analyzed in real time to detect anomalies and faults in a timely manner. The specific steps include:

[0065] Connect sensor data to the data stream processing system through the IoT gateway, and convert data formats of different sensors into a standardized format;

[0066] Use buffers to temporarily store data, smooth data traffic, and use message queues to manage data flows;

[0067] Use the real-time processing framework to process and analyze data streams in real time, group data streams by time windows or sliding windows, and perform aggregation operations;

[0068] Use algorithms in the real-time processing framework to perform anomaly detection, identify abnormal data points, and input real-time data into the trained machine learning model for fault prediction and early warning;

[0069] The processed real-time data is stored in the real-time database, and the data is stored in hot and cold tiers according to the real-time requirements of the data.

[0070] Buffers temporarily store data to prevent sudden data volumes from putting pressure on the system. Message queues include Apache Kafka, RabbitMQ, etc.

[0071] Real-time processing frameworks include Apache Flink, Apache Storm, Apache Spark Streaming, and other frameworks;

[0072] Sliding window grouping, such as calculating the average, maximum, minimum, etc. within the sliding window;

[0073] The algorithms in the real-time processing framework include sliding average, exponential smoothing, Z-score and other algorithms, and the real-time databases include InfluxDB, TimescaleDB and other algorithms;

[0074] Hot and cold tiered storage means real-time data is stored in a high-performance database, and historical data is stored in a data lake or distributed storage system;

[0075] Through data preprocessing and real-time data stream processing steps, an efficient and reliable intelligent fault monitoring system can be built to ensure the stable operation of the sewage treatment process and the efficient management of the equipment.

[0076] Use machine learning algorithms to build fault detection models, use historical fault data and regular operation data of water treatment equipment for label training, and evaluate the performance of fault detection models through cross-validation;

[0077] In the process of inputting real-time data into the trained machine learning model for fault prediction and early warning, the method for constructing a fault detection model based on the machine learning model is as follows:

[0078] Obtain water signal and electrical signal data collected by sensors, as well as operation log data of water treatment equipment, pre-process missing values ​​and abnormal values ​​in the data, and normalize or standardize the data;

[0079] According to the operation log of the water treatment equipment, the fault points in the data are marked, the data is divided into normal and fault states, and the data is divided into training set, verification set, and test set;

[0080] The time series data collected by the sensor is used as the original feature, and the derived features are calculated based on the original features. The derived features include the mean, extreme value and variability of the original data;

[0081] Use sliding window technology to divide time series data into multiple overlapping subsequences, each subsequence as a sample;

[0082] Construct a fault detection model based on long short-term memory network. The input layer is time series data. Each time step corresponds to a data point of the sensor. LSTM is stacked with multiple LSTM units. The fully connected layer connects the LSTM layer and the output layer. The number of nodes and activation function of the output layer are set according to the prediction target, and the fault prediction is output.

[0083] Use cross entropy as the loss function of the fault detection model and apply optimization algorithms to accelerate model convergence;

[0084] Divide the data into small batches for training, detect the loss value of the validation set, stop training when the validation loss no longer decreases, and use L2 regularization or dropout technology to reduce overfitting;

[0085] Use evaluation metrics and confusion matrices to evaluate the fault detection model, adjust the model's hyperparameters through grid search or random search, and adjust the model structure based on the verification results.

[0086] The operation log of the water treatment equipment records the time when the fault occurred, the cause of the fault and the performance information;

[0087] Optimization algorithms such as Adam optimizer or RMSprop optimizer, small batch training can speed up training and stability, and monitoring the loss value of the validation set can prevent overfitting;

[0088] Evaluation metrics include accuracy, precision, recall, F1 score, AUC, etc., and hyperparameters include the number of LSTM units, learning rate, batch size, etc.

[0089] The calculation method of the derived characteristic is: the vibration frequency of the calibration equipment is Vf, the water quality optimization efficiency coefficient is Op, and the data return coefficient is Ec. Then the derived characteristic is De, and the calculation expression is In the formula, the equipment vibration frequency is the vibration frequency of the water treatment equipment, the water quality optimization efficiency coefficient is the ratio of the water quality index after physical and chemical treatment to the preset target water quality index, the data return coefficient is the ratio of missing values, abnormal values ​​to the total data value in the data, α and β are the weight coefficients of the ratio of equipment vibration frequency, water quality optimization efficiency coefficient and data return coefficient, and α and β are both positive numbers.

[0090] Use the monitoring service of the cloud computing platform to monitor the equipment operation status in real time, predict the operation data of the water treatment equipment based on the fault detection model, and trigger the alarm mechanism when abnormal faults of the water treatment equipment are predicted;

[0091] When the fault detection model predicts that the failure probability of the water treatment equipment exceeds the preset threshold, the alarm mechanism is triggered, the preset threshold level is set, the first alarm threshold and the second alarm threshold are calibrated, and the failure prediction probability of the first alarm threshold is less than the second alarm threshold. When it is detected that the failure probability of the water treatment equipment exceeds the first alarm threshold and is lower than the second alarm threshold, the water treatment equipment is marked to prompt the staff to pay attention to the water treatment equipment and record the alarm log;

[0092] When it is detected that the failure probability of the water treatment equipment exceeds the second alarm threshold, an alarm notification is issued to inform the maintenance team to perform emergency treatment on the water treatment equipment and record the alarm log.

[0093] The present invention solves the problem of shortage of data processing capacity for large-scale sewage treatment plants with high-standard water discharge by processing the status data of water treatment equipment on the cloud. By constructing a long-short term memory network and applying a fault detection model to predict faults of water treatment equipment, the problem of production damage and profit loss caused by alarm after equipment damage is effectively avoided. The artificial intelligence algorithm solves the problem that the existing PID control is difficult to fully cover the water system and respond efficiently, further improving the level of automated management of the water system and reducing the management burden of staff.

[0094] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0095] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of computer program goods. The computer program goods include one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0096] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0097] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0098] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0099] If the functions are implemented in the form of software functional units and sold or used as independent goods, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can essentially or in other words, the part that contributes to the prior art or the part of the technical solution can be embodied in the form of software goods, and the computer software goods are stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0100] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A sewage treatment equipment fault monitoring method based on cloud computing, characterized in that: The specific steps include: The water signal and electrical signal of the water treatment equipment are collected through sensors, and the water signal and electrical signal are transmitted to the cloud computing platform through the network; The distributed database of the cloud computing platform is used to store water signals and electrical signals, and the data is pre-processed. The real-time processing framework of the cloud computing platform is used to process and analyze the water signals and electrical signals. Use machine learning algorithms to build fault detection models, use historical fault data and regular operation data of water treatment equipment for label training, and evaluate the performance of fault detection models through cross-validation; Use the monitoring service of the cloud computing platform to monitor the equipment operation status in real time, predict the operation data of the water treatment equipment based on the fault detection model, and trigger the alarm mechanism when abnormal faults of the water treatment equipment are predicted; The method of processing data using the real-time processing framework of the cloud computing platform is: Fill missing data or delete data records containing missing values, identify and handle data outliers; Use low-pass filtering, high-pass filtering or band-pass filtering to remove noise from the data, decompose the signal through wavelet transform, and remove the high-frequency noise part of the signal; Transform the data to a standard normal distribution, usually by subtracting the mean and dividing by the standard deviation so that the data has a distribution with mean 0 and standard deviation 1. Scale the data to a specific range, the specific range is [ 0,1 ] or [ -1,1 ] ; Select model features and extract features from the original data, including frequency domain features and time domain features; Smooth the data by moving average or exponential smoothing and discretize the continuous data; The method of using machine learning algorithms to build a fault detection model for fault detection is: Connect sensor data to the data stream processing system through the IoT gateway, and convert data formats of different sensors into a standardized format; Use buffers to temporarily store data, smooth data traffic, and use message queues to manage data flows; Use the real-time processing framework to process and analyze data streams in real time, group data streams by time windows or sliding windows, and perform aggregation operations; Use algorithms in the real-time processing framework to perform anomaly detection, identify abnormal data points, and input real-time data into the trained machine learning model for fault prediction and early warning; The processed real-time data is stored in a real-time database, and the data is stored in a hot and cold tier according to the real-time requirements of the data; The fault detection model is generated as follows: Obtain water signal and electrical signal data collected by sensors, as well as operation log data of water treatment equipment, pre-process missing values ​​and abnormal values ​​in the data, and normalize or standardize the data; According to the operation log of the water treatment equipment, the fault points in the data are marked, the data is divided into normal and fault states, and the data is divided into training set, verification set, and test set; The time series data collected by the sensor is used as the original feature, and the derived features are calculated based on the original features. The derived features include the mean, extreme value and variability of the original data; Use sliding window technology to divide time series data into multiple overlapping subsequences, each subsequence as a sample; Construct a fault detection model based on long short-term memory network. The input layer is time series data. Each time step corresponds to a data point of the sensor. LSTM is stacked with multiple LSTM units. The fully connected layer connects the LSTM layer and the output layer. The number of nodes and activation function of the output layer are set according to the prediction target, and the fault prediction is output. Use cross entropy as the loss function of the fault detection model and apply optimization algorithms to accelerate model convergence; Divide the data into small batches for training, detect the loss value of the validation set, stop training when the validation loss no longer decreases, and use L2 regularization or dropout technology to reduce overfitting; Use evaluation metrics and confusion matrices to evaluate fault detection models, adjust model hyperparameters through grid search or random search, and adjust model structure based on verification results; The derived features are calculated as: The vibration frequency of the calibration equipment is Vf, the water quality optimization efficiency coefficient is Op, and the data return coefficient is Ec. The derived characteristic is De, and the calculation expression is: In the formula, the equipment vibration frequency is the vibration frequency of the water treatment equipment, the water quality optimization efficiency coefficient is the ratio of the water quality index after physical and chemical treatment to the preset target water quality index, the data return coefficient is the ratio of missing values, abnormal values ​​to the total data value in the data, α and β are the weight coefficients of the ratio of equipment vibration frequency, water quality optimization efficiency coefficient and data return coefficient, and α and β are both positive numbers.

2. A sewage treatment equipment fault monitoring method based on cloud computing according to claim 1, characterized in that: The water signals collected by the sensor include flow, liquid level, pH value, dissolved oxygen concentration, and temperature, and the electrical signals include voltage, current, and frequency.

3. A sewage treatment equipment fault monitoring method based on cloud computing according to claim 2, characterized in that: The construction logic of the alarm mechanism is: A preset threshold level is set, a first alarm threshold and a second alarm threshold are calibrated, and the failure prediction probability of the first alarm threshold is less than the second alarm threshold. When it is detected that the failure probability of the water treatment equipment exceeds the first alarm threshold and is lower than the second alarm threshold, the water treatment equipment is marked to prompt the staff to pay attention to the water treatment equipment and record the alarm log; When it is detected that the failure probability of the water treatment equipment exceeds the second alarm threshold, an alarm notification is issued to inform the maintenance team to perform emergency treatment on the water treatment equipment and record the alarm log.

Images