An electrochemical signal data analysis method and system for early warning of water entry toxicity

Abstract

The application discloses an electrochemical signal data analysis method and system for water inflow toxicity early warning, relates to the technical field of water quality monitoring and early warning, and is used for improving the timeliness, accuracy and intelligent level of water inflow toxicity early warning and comprises the following steps: performing preliminary processing on electrochemical signal data to generate simplified data and local abnormal information; identifying the normal fluctuation range of the electrochemical signal data under different situations according to the simplified data; adjusting the normal fluctuation range of the electrochemical signal data according to the change of water quality, obtaining the current normal behavior definition, identifying abnormal behavior, and analyzing the characteristics of the abnormal behavior; when the characteristics of the abnormal behavior are verified through multidimensional correlation, the characteristics of the abnormal behavior are confirmed as the characteristics of a toxicity mode; when the electrochemical signal data matches any toxicity mode in a toxicity behavior mode rule set and deviates from the current normal behavior definition, the toxicity event is determined, and early warning information is generated.

Description

Technical Field

[0001] This invention relates to the field of water quality monitoring and early warning technology, and in particular to an electrochemical signal data analysis method and system for early warning of influent toxicity. Background Technology

[0002] In the daily operations of industrial wastewater treatment, timely and accurate detection of toxic substances in the influent is crucial for protecting biological treatment systems and ensuring their stable operation. Failure to provide timely warnings can lead to severe damage to the microbial community, or even its complete collapse. As modern industrial facilities increasingly rely on automation and data-driven approaches, various online water quality monitoring devices, especially sensor systems utilizing the principles of microbial electrochemistry, are widely used. These devices generate vast amounts of electrochemical signal data every moment, providing a foundation for more intelligent early warning and judgment based on data.

[0003] However, existing monitoring systems still have some significant shortcomings in fully utilizing this data and making real-time decisions. First, traditional systems typically use fixed thresholds for toxicity assessment, but the electrical signals generated by microbial fuel cells drift slowly due to factors such as ambient temperature, nutrient solution consumption, and biofilm aging. If the baseline cannot be updated adaptively, it can lead to false alarms (mistaking normal drift for toxicity) or missed alarms (failing to trigger an alarm when toxicity occurs due to baseline shift). Existing baseline correction often relies on periodic manual calibration, which is inefficient and lagging.

[0004] Secondly, existing systems are not sensitive enough to slowly evolving toxic events (such as the long-term continuous inflow of low-concentration toxins), because a slow decline in power generation can easily be mistaken for a normal baseline drift. Furthermore, existing systems cannot effectively distinguish between toxicity inhibition and nutrient promotion—when nutrients flow in, power generation increases, which is also a change in water quality, but traditional methods often only focus on the direction of decline and lack the ability to identify bidirectional changes.

[0005] Finally, traditional systems have slow response times, often requiring a considerable amount of time from water sample introduction to early warning output, which cannot meet the needs of real-time early warning. Therefore, there is an urgent need for an electrochemical signal data analysis method that can dynamically adapt to the baseline, rapidly identify bidirectional anomalies, and be sensitive to slow toxicity evolution. Summary of the Invention

[0006] This application discloses an electrochemical signal data analysis method and system for influent toxicity early warning, aiming to solve the technical problems in existing industrial wastewater treatment monitoring systems, such as low efficiency when processing large-scale, multi-source real-time electrochemical signal data, baseline drift leading to false alarms or missed alarms, insensitivity to slow toxicity, and slow response speed.

[0007] In a first aspect, this application discloses an electrochemical signal data analysis method for influent toxicity early warning, comprising the following steps: Electrochemical signal data from multiple monitoring points are acquired, and the electrochemical signal data are preliminarily processed to generate simplified data and local anomaly information, which are then transmitted to the central processing platform. The central processing platform receives simplified data, identifies the normal fluctuation range of electrochemical signal data under different scenarios based on the simplified data, and adjusts the normal fluctuation range of electrochemical signal data according to changes in water quality to obtain the current definition of normal behavior. When local abnormal information deviates from the normal fluctuation range of data corresponding to the current normal behavior definition, or when simplified data deviates from the normal fluctuation range of data corresponding to the current normal behavior definition, the abnormal behavior is identified and its characteristics are analyzed. When the features of abnormal behavior are verified through multi-dimensional association, the features of abnormal behavior are confirmed as features of toxicity patterns, and the features of toxicity patterns are added to the toxicity behavior pattern rule set, which includes features of different toxicity patterns. Based on the set of rules for toxic behavior patterns and the definition of current normal behavior, the continuously flowing electrochemical signal data is matched; when the electrochemical signal data matches any toxicity pattern in the set of rules for toxic behavior patterns and deviates from the definition of current normal behavior, or when the absolute value of the inhibition rate is greater than or equal to a preset inhibition rate threshold, it is determined to be a toxic event and an early warning message is generated; wherein, the formula for calculating the inhibition rate is TR = (U - U0) / U0×100%, TR is the inhibition rate, U is the power generation after the water sample to be tested is introduced, and U0 is the baseline power generation before the water sample to be tested is introduced.

[0008] Optionally, generate simplified data, including: Determine the mean and standard deviation of electrochemical signal data at different time periods; The mean and standard deviation are used as simplified data.

[0009] Optionally, generate local anomaly information, including: Determine the magnitude of the change in the average value of the electrochemical signal data; When the change magnitude exceeds the magnitude threshold, local anomaly information is generated; the local anomaly information includes the time of occurrence, duration, and magnitude of change of the anomaly.

[0010] Optionally, the characteristics of anomalous behavior are determined through multi-dimensional correlation verification, including: The characteristics of the anomalous behavior are matched with known toxic behavior patterns; if the characteristics of the anomalous behavior do not match any known toxic behavior pattern, the following three parameters of the electrochemical signal are obtained within a preset time period: The decrease amplitude ΔI = I0 - I_min, where I0 is the steady-state current value before the abnormal behavior occurs, and I_min is the lowest current value within the preset time. The rate of descent v = ΔI / Δt, where Δt is the time elapsed from the start of the anomaly to reaching I_min; Recovery time is the time required to recover from I_min to 90% of I0; When the decrease magnitude is greater than a first preset threshold, the decrease rate is greater than a second preset threshold, and the recovery time is greater than a third preset threshold, the characteristics of the abnormal behavior are determined to have passed the multi-dimensional correlation verification.

[0011] Optionally, the central processing platform receives simplified data and identifies the normal fluctuation range of electrochemical signal data under different scenarios based on the simplified data, including: The central processing platform receives simplified data and performs signal decomposition on the simplified data to separate the long-term trend component, periodic fluctuation component and residual component in the simplified data. Based on long-term trend components, periodic fluctuation components, and residual components, the normal fluctuation range of electrochemical signal data is constructed.

[0012] Optionally, identify anomalous behavior, including: Signal decomposition is performed on local anomalies or simplified data to separate long-term trend components, periodic fluctuation components, and residual components from the local anomalies or simplified data. When the rate of change of a long-term trend component exceeds the preset normal drift range, it is marked as a trend anomaly. Determine whether the trend anomalies in electrochemical signal data are consistent with known non-toxic factors; When anomalies in electrochemical signal data are not consistent with known non-toxic factors, the anomalies are marked as potential long-term evolutionary toxicity indicators. Anomaly detection is performed on residual components to identify whether the residual components are specific; if the instantaneous amplitude or duration of a specific residual component exceeds a preset threshold. Anomalous behavior is identified when long-term trend anomalies in electrochemical signal data are flagged as potential long-term evolutionary toxicity indicators, or when residual components exhibit toxicity pattern specificity.

[0013] Optionally, when anomalies in the trend of electrochemical signal data are not consistent with known non-toxic factors, the anomalies are marked as potential long-term evolutionary toxicity indicators, including: Electrochemical signal data before and after the occurrence of trend anomalies are sampled, and time-dimensional change features are extracted from the sampled electrochemical signal data; the change features include the instantaneous slope of the signal, the distribution of local extreme points, and the energy distribution within a preset frequency range; The fluctuations and mutual influences of auxiliary sensor data related to electrochemical signal data during preset time periods before and after the occurrence of trend anomalies were analyzed to obtain the synchronous fluctuation characteristics of auxiliary sensor data. The variation characteristics of electrochemical signal data are correlated with the synchronous fluctuation characteristics of auxiliary sensor data; When the variation characteristics of electrochemical signal data are consistent with the synchronous fluctuation characteristics of auxiliary sensor data in time, and the variation characteristics and synchronous fluctuation characteristics differ from known short-term random fluctuation patterns, the confidence level of trend anomalies as potential long-term evolutionary toxicity indicators is increased. Based on the increased confidence level, trend anomalies are marked as potential indicators of long-term evolutionary toxicity.

[0014] Optionally, after generating the warning information, the method further includes: Send a close command to the inlet valve; Increase the aeration rate to improve the dissolved oxygen concentration.

[0015] Secondly, this application also discloses an electrochemical signal data analysis system for influent toxicity early warning, the system comprising: The data acquisition and preprocessing module is used to acquire electrochemical signal data from multiple monitoring points, perform preliminary processing on the electrochemical signal data to generate simplified data and local anomaly information, and transmit the simplified data and local anomaly information to the central processing platform. The normal behavior definition module is used to receive simplified data on the central processing platform, identify the normal fluctuation range of electrochemical signal data under different scenarios based on the simplified data, and adjust the normal fluctuation range of electrochemical signal data according to changes in water quality to obtain the current normal behavior definition. The abnormal behavior identification module is used to identify and analyze the characteristics of abnormal behavior when local abnormal information deviates from the normal fluctuation range of the data corresponding to the current normal behavior definition, or when simplified data deviates from the normal fluctuation range of the data corresponding to the current normal behavior definition. The toxicity pattern learning module is used to identify the features of abnormal behavior as features of toxicity patterns when the features of abnormal behavior are verified through multi-dimensional association, and to add the features of toxicity patterns to the toxicity behavior pattern rule set, which includes features of different toxicity patterns. The early warning and response module is used to match the continuously flowing electrochemical signal data based on the set of toxic behavior patterns and the definition of current normal behavior. When the electrochemical signal data matches any toxicity pattern in the set of toxic behavior patterns and deviates from the definition of current normal behavior, or when the absolute value of the inhibition rate is greater than or equal to a preset inhibition rate threshold, it is determined to be a toxic event and an early warning message is generated. The inhibition rate is calculated as TR = (U - U0) / U0 × 100%, where TR is the inhibition rate, U is the power generation after the water sample is introduced, and U0 is the baseline power generation before the water sample is introduced.

[0016] Beneficial effects This application discloses an electrochemical signal data analysis method for influent toxicity early warning. By acquiring and preliminarily processing electrochemical signal data (i.e., electrical signals generated by microbial fuel cells) from multiple monitoring points, it generates simplified data and local anomaly information, effectively solving the problem of low efficiency in existing systems when processing large-scale, multi-source real-time data. This method dynamically identifies and adjusts the normal fluctuation range of electrochemical signal data on a central processing platform, forming a current definition of normal behavior. This overcomes the drawbacks of traditional fixed threshold methods, which suffer from false alarms or missed alarms due to baseline drift. It can adapt to slow changes in water quality, significantly improving the accuracy of early warning. When local anomaly information or simplified data deviates from the definition of normal behavior, the system can identify the abnormal behavior and analyze its characteristics, including rapid instantaneous declines (acute toxicity) and slow trend declines (chronic toxicity). More importantly, this application introduces a multi-dimensional correlation verification mechanism based on the magnitude of the generated electrical signal decline, which can effectively distinguish between different scenarios such as toxicity inhibition and nutrient promotion, avoiding misjudgments caused by nutrient inflow. Finally, by matching the continuously flowing electrochemical signal data with a dynamically updated set of toxic behavior pattern rules and the current definition of normal behavior, toxic events can be determined in real time and early warning information can be generated. In summary, the method of this application effectively solves the problems of baseline drift, slow toxicity insensitivity, slow response speed, and inability to distinguish bidirectional changes in the prior art, and significantly improves the timeliness, accuracy and intelligence level of influent toxicity early warning, thereby effectively ensuring the stable operation of the biological treatment system. Attached Figure Description

[0017] Figure 1 This is a schematic flowchart of an electrochemical signal data analysis method for influent toxicity early warning provided by an embodiment of the present invention; Figure 2 This is a schematic diagram of another electrochemical signal data analysis method for influent toxicity early warning provided by an embodiment of the present invention; Figure 3 This is a schematic diagram of an electrochemical signal data analysis system for influent toxicity early warning provided in an embodiment of the present invention. Detailed Implementation

[0018] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0019] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0020] To better understand the electrochemical signal data analysis method for influent toxicity early warning proposed in this application, the following will elaborate on some key terms and implementation environments involved.

[0021] "Electrochemical signal data" refers to electrical signals, such as voltage or current signals, generated in real time by sensors in a microbial fuel cell (MFC). This signal is produced by the metabolic activity of electrogenic microorganisms. When toxic substances enter, the activity of these microorganisms decreases, and the amount of electricity produced drops; when nutrients enter, the amount of electricity produced increases. By monitoring changes in this signal, toxicity (inhibition rate) can be calculated to characterize the overall toxicity level of the water.

[0022] "Simplified data" refers to the simplified information extracted from the raw electrochemical signal data after preliminary processing, which represents the core characteristics of the data. The aim is to reduce the burden of data transmission and processing while retaining key information.

[0023] "Local anomaly information" refers to data points or data segments that deviate significantly from the normal pattern within a short period of time or a local area, which are identified in the initial data processing stage. This information may indicate potential anomalies.

[0024] The "central processing platform" refers to the core computing unit responsible for receiving simplified data and local anomaly information from multiple monitoring points, and conducting centralized analysis, decision-making, and early warning.

[0025] The "normal behavior definition" refers to the normal fluctuation range and pattern of electrochemical signal data under different situations, established by learning and analyzing historical simplified data. It serves as a benchmark for judging whether the data is abnormal.

[0026] The "Toxic Behavior Pattern Rule Set" refers to a database that stores the characteristics of various known and newly identified toxic events, used to match and identify potential toxic events. This rule set does not distinguish between specific types of toxic substances, but rather classifies them based on characteristics such as the magnitude, rate, and duration of the decline in the generated electrical signal.

[0027] The following specific embodiments will provide a detailed introduction and explanation of an electrochemical signal data analysis method for influent toxicity early warning provided in this application.

[0028] Reference Figure 1 This invention provides an electrochemical signal data analysis method for influent toxicity early warning, comprising the following steps: S1 acquires electrochemical signal data from multiple monitoring points, performs preliminary processing on the electrochemical signal data to generate simplified data and local anomaly information, and transmits the simplified data and local anomaly information to the central processing platform.

[0029] The electrochemical signal data includes electrical signals (such as voltage) generated by the microbial fuel cell. This signal is produced by the metabolic activity of electrogenic microorganisms. When toxic substances enter, microbial activity is inhibited, and the electrical signal decreases; when nutrients enter, the electrical signal increases. By monitoring the dynamic changes in this electrical signal, fluctuations in water toxicity can be captured in real time.

[0030] Electrochemical signal data can be acquired in real time by multiple sensors deployed in front of different influent networks or treatment units. These sensors can be microbial fuel cell sensors, which continuously monitor key electrochemical parameters in the water. In the preliminary processing stage, the raw data can be processed in various ways to generate simplified data and local anomaly information. For example, the raw electrochemical signal data can be averaged and its standard deviation calculated within a time window, and these statistics can be used as simplified data. Alternatively, the data can be downsampled, extracting a data point at regular time intervals as simplified data. For the generation of local anomaly information, a threshold for the magnitude of change can be set. When the change in electrochemical signal data exceeds this threshold within a short period, local anomaly information is generated, which can include the time of occurrence, duration, and magnitude of the change. The simplified data and local anomaly information after preliminary processing are then transmitted to a central processing platform for further centralized analysis.

[0031] S2. On the central processing platform, receive simplified data, identify the normal fluctuation range of electrochemical signal data under different scenarios based on the simplified data, and adjust the normal fluctuation range of electrochemical signal data according to changes in water quality to obtain the current normal behavior definition.

[0032] For example, based on historical simplified data, statistical analysis methods (such as calculating the mean and standard deviation) can be used to determine the normal fluctuation range over a specific time period (such as different times of day or different dates of week) or under specific operating conditions (such as different influent flow rates or different seasons). Furthermore, to adapt to dynamic changes in water quality, the system will adjust these normal fluctuation ranges to arrive at the current definition of normal behavior. This adjustment can be periodic, such as recalculating the normal fluctuation range daily or weekly based on the latest simplified data; or it can be adaptive, such as automatically updating the normal fluctuation range when a long-term trend in water quality is detected.

[0033] S3. When local abnormal information deviates from the normal fluctuation range of the data corresponding to the current normal behavior definition, or when simplified data deviates from the normal fluctuation range of the data corresponding to the current normal behavior definition, identify the abnormal behavior and analyze its characteristics.

[0034] For example, if local anomaly information at a monitoring point shows a sharp drop in the electrical signal within a short period of time, and this drop exceeds the fluctuation range allowed by the current definition of normal behavior, it will be identified as abnormal behavior. When analyzing the characteristics of abnormal behavior, parameters such as the time of occurrence, duration, magnitude of change, and rate of change can be extracted.

[0035] S4. When the features of abnormal behavior are verified through multi-dimensional association, the features of abnormal behavior are confirmed as features of toxicity patterns, and the features of toxicity patterns are added to the toxic behavior pattern rule set.

[0036] The set of rules for toxic behavior patterns includes the characteristics of different toxic patterns.

[0037] Multi-dimensional correlation verification can include matching the characteristics of the current abnormal behavior with known toxic behavior patterns. If no match can be found, further judgment is made by considering the changing trends of electrical signals. For example, when the decrease magnitude, decrease rate, and recovery time of the electrochemical signal within a preset time period are all greater than their respective preset thresholds, the abnormal behavior can be considered to have the characteristics of a toxic pattern. The toxic behavior pattern rule set is a dynamically updated database that contains the characteristics of various known toxic patterns, as well as the characteristics of newly identified toxic patterns learned from other sources.

[0038] S5. Based on the set of rules for toxic behavior patterns and the current definition of normal behavior, the continuously flowing electrochemical signal data is matched. When the electrochemical signal data matches any toxicity pattern in the set of rules for toxic behavior patterns and deviates from the current definition of normal behavior, or when the absolute value of the inhibition rate is greater than or equal to the preset inhibition rate threshold, it is determined to be a toxic event and an early warning message is generated.

[0039] The inhibition rate is calculated using the formula TR = (U - U0) / U0 × 100%, where TR is the inhibition rate, U is the power generation after the water sample is introduced, and U0 is the baseline power generation before the water sample is introduced.

[0040] For example, if newly incoming streamlined data exhibits characteristics that are highly consistent with a known toxicity pattern in the set of toxic behavior pattern rules in terms of electrical signal features, and these data points exceed the fluctuation range allowed by the current definition of normal behavior, the system will immediately determine it as a toxic event and generate corresponding warning information, such as sending an alarm to the operator or displaying it on the monitoring interface.

[0041] In some embodiments, after generating the warning information, the method further includes: sending a shut-off command to the inlet valve; and increasing the aeration rate to increase the dissolved oxygen concentration.

[0042] Specifically, sending a shut-off command to the inlet valve means that when the system detects a toxic event and generates an early warning message, it automatically or semi-automatically sends a shut-off signal to the valve connected to the inlet pipeline through the control system. This valve can be an electric valve, a pneumatic valve, or other types of automatically controlled valve. Its purpose is to quickly cut off or significantly reduce the inflow of contaminated water, thereby protecting subsequent water treatment units from the impact of toxic substances. Increasing aeration to raise dissolved oxygen concentration can be understood as the system activating aeration equipment, such as blowers and aeration heads, to operate at higher power or for longer periods after detecting a toxic event, thereby injecting more air or pure oxygen into the water to increase the dissolved oxygen concentration. The purpose is that some toxic substances may be more easily oxidized and decomposed in an oxygen-rich environment, or that increasing the dissolved oxygen concentration can provide a better living environment for microorganisms, enhancing their ability to degrade toxic substances, thus mitigating the toxic effects to some extent.

[0043] Sending a shut-off command to the inlet valve can directly prevent or significantly reduce the entry of influent containing toxic substances into subsequent treatment stages, thereby avoiding damage to water treatment facilities and buying valuable time for emergency response. Simultaneously, increasing aeration to raise dissolved oxygen concentration creates a high-oxygen environment to promote the biodegradation or chemical oxidation of certain toxic substances, reducing their toxicity and mitigating negative impacts on aquatic organisms and the treatment process, especially in situations where toxic substances may have already entered or are about to enter the treatment system. This dual response mechanism works synergistically to comprehensively enhance the toxicity early warning system's response capabilities from both source control and process mitigation perspectives.

[0044] The electrochemical signal data analysis method for influent toxicity early warning proposed in this application effectively solves the problems of baseline drift leading to false alarms or missed alarms, insensitivity to slow toxicity, slow response speed, and inability to distinguish between toxicity inhibition and nutrient promotion in existing technologies through a series of collaborative technical features. Specifically, the method first achieves distributed preliminary processing and efficient transmission of massive multi-source data by "acquiring electrochemical signal data from multiple monitoring points, performing preliminary processing on the electrochemical signal data to generate simplified data and local anomaly information, and transmitting the simplified data and local anomaly information to a central processing platform." This approach avoids the bottleneck of traditional centralized processing, ensures timely acquisition and preliminary screening of real-time data, and lays the foundation for subsequent in-depth analysis.

[0045] Next, the process of "receiving simplified data on the central processing platform, identifying the normal fluctuation range of electrochemical signal data under different scenarios based on the simplified data, and adjusting the normal fluctuation range of electrochemical signal data according to changes in water quality to obtain the current definition of normal behavior" constructs a dynamically adaptive baseline for normal behavior. Unlike traditional static threshold settings, this application can adjust the normal fluctuation range in real time according to the actual trend of water quality changes, making anomaly identification more accurate, effectively avoiding false alarms and missed alarms, and significantly improving the accuracy of early warning.

[0046] Subsequently, the step of "identifying abnormal behavior and analyzing its characteristics when local abnormal information deviates from the normal fluctuation range of data corresponding to the current definition of normal behavior, or when simplified data deviates from the normal fluctuation range of data corresponding to the current definition of normal behavior" enables early detection and feature extraction of potential toxic events. By simultaneously monitoring simplified data and local abnormal information, this application can more comprehensively detect anomalies, effectively identifying both slow trend changes and sudden instantaneous fluctuations.

[0047] Most importantly, the innovative mechanism of "confirming the features of abnormal behavior as features of toxicity patterns when the features of abnormal behavior are verified through multi-dimensional correlation, and adding the features of toxicity patterns to the toxicity behavior pattern rule set, which includes features of different toxicity patterns" enables the system to possess self-learning and adaptive capabilities. It breaks through the limitations of traditional reliance on human experience and static feature libraries, and can proactively identify and learn newly generated electrical signal change patterns, greatly enhancing the system's ability to cope with threats from unknown toxic substances and ensuring the continuous effectiveness of the early warning system.

[0048] Finally, the step of "matching continuously flowing electrochemical signal data based on the set of toxic behavior pattern rules and the current definition of normal behavior; when the electrochemical signal data matches any toxicity pattern in the set of toxic behavior pattern rules and deviates from the current definition of normal behavior, it is determined as a toxic event, and an early warning message is generated" completes the final determination and early warning of the toxic event. By combining dynamic definitions of normal behavior and a constantly updated set of toxic behavior pattern rules, this application can accurately match and judge continuously flowing data, and immediately generate early warning information once a toxic event is detected, providing valuable time for subsequent emergency response.

[0049] In summary, this application, through distributed data preliminary processing, dynamically adaptive definition of normal behavior, comprehensive abnormal behavior identification, and an innovative toxicity pattern learning mechanism, has formed an efficient, intelligent, and adaptive influent toxicity early warning system. Compared with existing technologies, this application has achieved significant progress in data processing efficiency, early warning accuracy, sensitive identification of slow toxicity evolution, and system adaptability, providing a more reliable guarantee for the stable operation of industrial wastewater treatment systems.

[0050] like Figure 2 As shown, generating simplified data specifically includes: S101. Determine the average value and standard deviation of electrochemical signal data at different time periods.

[0051] Determining the average and standard deviation of electrochemical signal data across different time periods involves averaging the collected electrochemical signal data within a pre-defined time window to reflect the overall signal strength during that period, and calculating the standard deviation to measure the volatility or dispersion of the electrochemical signal data within that time frame. By calculating the average and standard deviation, the original, high-frequency electrochemical signal data can be effectively compressed and summarized, resulting in simplified data. Simplified data is an abstraction of the core characteristics of the original data, aiming to significantly reduce the data volume while retaining key information, facilitating subsequent transmission and processing.

[0052] S102. Use the mean and standard deviation as simplified data.

[0053] The proposed solution effectively captures the core characteristics of raw electrochemical signal data by calculating the average and standard deviation over different time periods. The average reflects the central trend of the signal, while the standard deviation quantifies its volatility. This processing method transforms the raw, high-dimensional electrochemical signal data into low-dimensional, more easily analyzable, simplified data. After being transmitted to a central processing platform, this simplified data serves as the basis for identifying the normal fluctuation range of electrochemical signal data. Furthermore, it, along with local anomaly information, is used to identify abnormal behavior. This approach ensures efficient data transmission while providing necessary and sufficient information for subsequent toxicity warning analysis.

[0054] In some of the embodiments described above in this application, when acquiring electrochemical signal data from multiple monitoring points and performing preliminary processing, it is necessary to generate local anomaly information.

[0055] Specifically, generating local anomaly information includes: S201. Determine the variation range of the average value of the electrochemical signal data.

[0056] Electrochemical signal data can be understood as continuous measurements collected within a specific time window, such as voltage or current signals. The magnitude of the average value refers to the degree of difference between the average values ​​of electrochemical signal data over consecutive or adjacent time periods. For example, one can calculate the absolute value of the difference between the average value of the current time period and the average value of the previous time period, or calculate the rate of change of the average value within a certain sliding window.

[0057] S202. When the change amplitude is greater than the amplitude threshold, generate local anomaly information.

[0058] Local anomaly information includes the time of occurrence, duration, and magnitude of change of the anomaly.

[0059] The amplitude threshold is a pre-defined value used to define the significance of changes in the average value of electrochemical signal data. If the amplitude exceeds this threshold, it indicates that a rapid and significant change in water quality may have occurred, which is often an early sign of a potential toxic event. The generated local anomaly information is not a simple Boolean value, but contains several key parameters, including the time of anomaly occurrence, the duration of anomaly, and the amplitude of anomaly change. The time of anomaly occurrence records the specific moment the anomalous event began, which is helpful for subsequent time series analysis and event tracing. The duration indicates the time span from the start to the end of the anomalous state, which is crucial for assessing the severity and nature of the anomalous event. The amplitude of change quantifies the degree to which the signal deviates from the normal state, providing a direct measure of the anomaly intensity.

[0060] This application's solution monitors the fluctuation range of the average value of electrochemical signal data and compares it with a preset amplitude threshold, thereby enabling timely detection of rapid fluctuations in water quality parameters. The core of this method lies in the fact that sudden toxic substances in water often cause a significant and sharp drop in the generated electrical signal within a short period. By calculating the fluctuation range of the average value, this instantaneous change can be effectively quantified, while the amplitude threshold provides a benchmark for judging anomalies. Once the fluctuation range exceeds this threshold, it indicates a possible abnormal water quality disturbance. Furthermore, by integrating detailed information such as the time of anomaly occurrence, duration, and fluctuation range into local anomaly information, the system can not only identify the existence of anomalies but also preliminarily characterize their features, providing crucial input data for subsequent anomaly behavior identification and toxicity pattern learning.

[0061] In some embodiments described above, it is proposed to identify anomalous behavior features as toxicity patterns when these features pass multi-dimensional correlation verification. However, in practical applications, how to effectively perform multi-dimensional correlation verification to identify and confirm toxicity patterns when the features of anomalous behavior do not match any known toxic behavior patterns is a problem that needs to be solved. If this problem is not addressed, the system may be unable to detect and learn new toxic threats in a timely manner, thus affecting the accuracy and timeliness of early warnings. Therefore, this application further proposes a method for determining the features of anomalous behavior through multi-dimensional correlation verification, aiming to improve the ability to identify and confirm toxicity patterns.

[0062] The aforementioned features for identifying anomalous behavior are verified through multi-dimensional correlation, including: S301. Match the characteristics of abnormal behavior with known toxic behavior patterns.

[0063] This matching process aims to quickly identify patterns similar to historical toxicity events. Known toxic behavior patterns refer to electrochemical signal change patterns with clear toxicity indications, determined through historical data analysis, expert experience, or simulation experiments.

[0064] S302. When the characteristics of the abnormal behavior do not match any known toxic behavior pattern, obtain the following parameters of the electrochemical signal within a preset time period: The decrease amplitude ΔI = I0 - I_min, where I0 is the steady-state current value before the abnormal behavior occurs, and I_min is the lowest current value within the preset time. The rate of descent v = ΔI / Δt, where Δt is the time elapsed from the start of the anomaly to reaching I_min; Recovery time is the time required to recover from I_min to 90% of I0; S303. When the decrease magnitude is greater than the first preset threshold, the decrease rate is greater than the second preset threshold, and the recovery time is greater than the third preset threshold, the characteristics of the abnormal behavior are determined to be verified by the multi-dimensional correlation.

[0065] Among them, the preset duration, the first preset threshold, the second preset threshold, and the third preset threshold are thresholds that are preset based on water characteristics, historical data analysis, and typical impact patterns of toxic substances.

[0066] In some preferred embodiments, suppose that the continuously flowing electrochemical signal data at a monitoring point shows abnormal behavior characteristics, but these characteristics do not match any existing toxicity patterns in the toxicity behavior pattern rule set. In this case, the system will further analyze the trend of the electrical signal change. Specifically, if the electrical signal decreases from an initial 0.500V to 0.300V within a preset duration of 30 minutes, then: The decrease in voltage ΔI = 0.500V - 0.300V = 0.200V; The rate of descent v = 0.200V / 30min = 0.0067V / min; The recovery time is 45 minutes (i.e., the time required to recover from the lowest point of 0.300V to the steady-state value of 0.450V, which is 90% of I0).

[0067] If the pre-set first threshold (decline magnitude) is 0.100V, the second preset threshold (decline rate) is 0.005V / min, and the third preset threshold (recovery time) is 30min, then the above three dimensions of indicators satisfy the following conditions simultaneously: 0.200V>0.100V, 0.0067V / min>0.005V / min, and 45min>30min. Based on this, the system determines that the characteristics of this abnormal behavior have passed the multi-dimensional correlation verification and confirms it as a characteristic of the toxicity pattern.

[0068] Subsequently, the features of this toxicity pattern will be added to the toxic behavior pattern rule set for subsequent matching and early warning. In this way, even the first occurrence of a toxic event can be effectively identified by the system and included in its learning scope, thereby enhancing the system's adaptability and early warning capabilities.

[0069] This application's solution effectively addresses the problem of accurately determining whether an event is toxic when anomalous behavioral characteristics do not match known toxic patterns by introducing a multi-dimensional correlation verification mechanism for toxicity patterns. First, by matching the characteristics of anomalous behavior with known toxic behavior patterns, known toxic threats can be efficiently identified. When such direct matching is not possible, i.e., when facing potential toxic patterns, this solution uses the magnitude, rate, and recovery time of the decrease in the generated electrical signal as the basis for judgment. A significant decrease in the electrical signal usually indicates inhibition of electrogenic microbial activity, a common phenomenon in toxic contamination events. By setting reasonable decrease magnitude, rate, and recovery time, changes in electrochemical signals that differ from known patterns but possess typical toxic characteristics can be captured, thus enabling reliable confirmation of toxicity patterns even in the absence of direct matching rules.

[0070] In some embodiments described above, this application proposes identifying the normal fluctuation range of electrochemical signal data under different scenarios based on simplified data. However, in practical applications, electrochemical signal data often contains complex long-term trends, periodic fluctuations, and random noise. Simple identification methods may not accurately capture these complex patterns, leading to an imprecise definition of the normal fluctuation range and affecting the accuracy of toxicity warnings. To address this, this application further proposes a step of receiving simplified data on a central processing platform and identifying the normal fluctuation range of electrochemical signal data under different scenarios based on the simplified data. Specifically, this includes signal decomposition of the simplified data to separate long-term trend components, periodic fluctuation components, and residual components, and constructing the normal fluctuation range of the electrochemical signal data based on these components.

[0071] Specifically, the central processing platform receives simplified data and identifies the normal fluctuation range of electrochemical signal data under different scenarios based on the simplified data, including: S401. On the central processing platform, receive simplified data and perform signal decomposition on the simplified data to separate the long-term trend component, periodic fluctuation component and residual component in the simplified data.

[0072] Signal decomposition refers to breaking down a complex time-series signal into several simpler, more physically meaningful components. Specifically, methods such as Seasonal Trend Decomposition (STL), Wavelet Decomposition, or Empirical Mode Decomposition (EMD) can be used. The aim is to effectively distinguish different patterns of change in the original, simplified data, so that each pattern can be analyzed and modeled independently in subsequent processing.

[0073] Long-term trend components can be understood as the slow, continuous direction of change in electrochemical signal data over a longer timescale, such as signal changes caused by environmental changes, equipment aging, or long-term drift of water quality background values. Specifically, they can be extracted from simplified data using techniques such as moving averages, polynomial fitting, or low-pass filtering. The aim is to capture the fundamental evolutionary direction of the data.

[0074] Periodic fluctuation components refer to patterns in electrochemical signal data that repeat with a fixed or near-fixed period, such as fluctuations caused by diurnal variations, seasonal variations, or production cycles. These can be identified and extracted using methods such as Fourier transform, wavelet analysis, or autocorrelation analysis. The aim is to reflect predictable, repetitive behavior in the data.

[0075] The residual component refers to the portion remaining after removing long-term trend components and periodic fluctuation components from simplified data. It is generally considered to be a manifestation of random noise or anomalous events. Specifically, it can be obtained by subtracting the trend and periodic components from the original data. Its purpose is to isolate non-trend, non-periodic signals that may indicate anomalies.

[0076] S402. Based on long-term trend components, periodic fluctuation components, and residual components, construct the normal fluctuation range of electrochemical signal data.

[0077] Constructing the normal fluctuation range of electrochemical signal data refers to establishing normal fluctuation boundaries or models for each component after separating the three components mentioned above, based on their characteristics. For example, a reasonable drift range can be set for the long-term trend component; a prediction model based on historical data can be established for the periodic fluctuation component, and a prediction error range can be set; for the residual component, statistical methods (such as the three-standard-deviation rule or the IQR rule) can be used to define its normal fluctuation range. The purpose is to comprehensively consider the multiple dynamic characteristics of the data to form a comprehensive and accurate definition of normal behavior.

[0078] This application's solution decomposes simplified data into signal components, breaking down complex electrochemical signal data into long-term trend components, periodic fluctuation components, and residual components. This allows for a more refined understanding and modeling of the data's intrinsic structure. Traditional methods, when identifying normal fluctuation ranges, may struggle to distinguish between trend drift caused by normal environmental changes, periodic fluctuations caused by daily activities, and genuine anomalous events. Signal decomposition allows for independent analysis and modeling of the characteristics of each component. For example, the long-term trend component reflects slow changes in water quality background, the periodic fluctuation component reflects regular daily changes, and the residual component more directly reflects instantaneous anomalies or noise. This divide-and-conquer strategy makes the definition of normal behavior more precise and robust, avoiding misjudging normal data fluctuations as anomalies and improving sensitivity to genuine anomalies.

[0079] In some preferred embodiments, it is assumed that the central processing platform receives abbreviated electrochemical signal data from monitoring points, which may contain the mean and standard deviation of the generated electrical signals. To accurately identify the normal fluctuation range of this data, the system first performs signal decomposition on this abbreviated data. Specifically, the STL decomposition method can be used to decompose the abbreviated data of the generated electrical signals into a long-term trend component, a periodic fluctuation component (e.g., daily or weekly periodic variations), and a residual component. For the long-term trend component, an autoregressive model can be built based on historical data, and a confidence interval can be set as its normal drift range. For the periodic fluctuation component, its periodic pattern can be fitted using Fourier series, and an allowable fluctuation amplitude can be set. For the residual component, its mean and standard deviation can be calculated, and an outlier threshold based on three times the standard deviation can be set. In this way, when new abbreviated data flows in, its long-term trend, periodic fluctuation, and residual components are compared with their respective normal ranges. For example, if the instantaneous amplitude of the residual component of the generated electrical signal suddenly exceeds a preset threshold, it may indicate an instantaneous anomaly, even if the overall trend and periodicity are still within the normal range, thus more accurately identifying situations that deviate from the current definition of normal behavior.

[0080] In some of the embodiments described above in this application, abnormal behavior is identified when local abnormal information or simplified data deviates from the normal fluctuation range corresponding to the current definition of normal behavior. However, identifying abnormal behavior solely based on data deviation from the normal fluctuation range may not effectively distinguish between real toxic events caused by water quality changes and false anomalies caused by other non-toxic factors (such as environmental disturbances, sensor drift, or fluctuations in routine operations), which may lead to false alarms or missed alarms, affecting the reliability and accuracy of the early warning system.

[0081] In this regard, this application further proposes steps for identifying anomalous behavior, including: S501. Perform signal decomposition on local anomaly information or simplified data to separate the long-term trend component, periodic fluctuation component and residual component in the local anomaly information or simplified data.

[0082] Specifically, signal decomposition refers to breaking down complex electrochemical signal data (including local anomalies or simplified data) into several simpler, more interpretable components. The long-term trend component reflects the baseline or average level of the data as it slowly changes over time, such as a long-term, slow decline or rise in the generated electrical signal. The periodic fluctuation component captures recurring patterns in the data, such as periodic fluctuations in the electrical signal caused by diurnal temperature variations or seasonal changes. The residual component represents random noise or sudden events after removing trends and periodicity, such as signal spikes caused by instantaneous sensor malfunctions or sudden pollutant emissions. Signal decomposition can be achieved using various mature algorithms, such as Seasonal Trend Decomposition (STL), wavelet decomposition, or Empirical Mode Decomposition (EMD), with the aim of decoupling different types of anomalous behavior from the overall signal for targeted detection.

[0083] S502. When the rate of change of a long-term trend component exceeds the preset normal drift range, it is marked as a trend anomaly.

[0084] Furthermore, the rate of change of the long-term trend component refers to its slope or magnitude of change over a period of time. The preset normal drift range is a threshold set based on historical normal data and expert experience, used to define the normal fluctuation boundary of the long-term trend. When the rate of change of the long-term trend component exceeds this range, it indicates that there may be abnormal, persistent changes, such as the continuous and slow release of a certain pollutant leading to a continuous deterioration of water quality indicators. In this case, it is marked as an abnormal trend.

[0085] S503. Determine whether the trend anomalies in electrochemical signal data are consistent with known non-toxic factors.

[0086] The purpose of determining whether anomalies in electrochemical signal data are consistent with known non-toxic factors is to rule out trend changes caused by non-toxic reasons. Known non-toxic factors may include, but are not limited to, changes in ambient temperature, fluctuations in water flow velocity, sensor aging and drift, equipment maintenance operations, or routine process adjustments. Consistency determination can be achieved by comparing the change patterns of auxiliary sensor data (such as temperature sensors and flow meters) during the period in which the trend anomaly occurred, or by matching it with a historical database of non-toxic events. For example, if the generated electrical signal decreases slowly over a long period, but at the same time the temperature sensor shows a continuous decrease in water temperature, and it is known that this temperature decrease leads to a decrease in the generated electrical signal, then this trend anomaly may be considered consistent with non-toxic factors.

[0087] S504. When anomalies in electrochemical signal data are not consistent with known non-toxic factors, the anomalies are marked as potential long-term evolutionary toxicity indicators.

[0088] S505. Perform anomaly detection on residual components to identify whether the residual components are specific.

[0089] Among them, the instantaneous amplitude or duration of the specific residual component exceeds a preset threshold.

[0090] Specifically, anomaly detection of residual components aims to identify sudden, transient anomalous events. The specificity of a residual component refers to its characteristics that significantly differ from normal noise or random fluctuations, such as abnormally high transient amplitudes (i.e., sudden spikes or drops) or abnormally long durations. Preset thresholds are set based on the statistical properties of historical residual components and empirical characteristics of toxic events. When the transient amplitude or duration of a residual component exceeds these thresholds, it indicates a possible acute toxic event or sudden sensor malfunction.

[0091] S506. Identify anomalous behavior when long-term trend anomalies in electrochemical signal data are flagged as potential long-term evolutionary toxicity indicators, or when residual components exhibit toxicity pattern specificity.

[0092] This application's solution decomposes complex electrochemical signal data into long-term trend components, periodic fluctuation components, and residual components by analyzing local anomaly information or simplified data. This enables refined detection of different types of anomaly patterns. Specifically, detecting the rate of change of the long-term trend component can capture gradual water quality deterioration caused by the slow accumulation or continuous discharge of toxic substances, while the specific detection of the residual component can promptly detect instantaneous and drastic fluctuations caused by the sudden introduction of acutely toxic substances. By determining whether the trend anomaly is consistent with known non-toxic factors, this application can effectively filter out false anomalies caused by environmental changes or equipment malfunctions, ensuring that the identified trend anomalies more accurately point to potential long-term toxicity. This hierarchical, multi-dimensional anomaly identification mechanism allows the system to more comprehensively and accurately capture toxic events of different time scales and manifestations, significantly improving the accuracy and reliability of anomaly behavior identification.

[0093] In some preferred embodiments, it is assumed that a monitoring point continuously receives electrochemical signal data such as electrical signals.

[0094] Specifically, after the system receives the simplified data, it first performs signal decomposition. For example, using the STL decomposition method, the generated electrical signal data is decomposed into long-term trend components, periodic fluctuation components, and residual components.

[0095] In one scenario, the long-term trend component of the generated electrical signal is observed to exhibit a sustained rate of decline exceeding the preset normal drift range over the past 24 hours. The system then flags this as an anomaly. Subsequently, the system examines concurrent auxiliary sensor data, such as water temperature and flow rate. If both water temperature and flow rate remain stable, and the declining trend of the generated electrical signal does not match known non-toxic factors (such as seasonal acidification or routine chemical dosing), then this anomaly indicating an unknown toxic substance threat will be identified as a potential long-term toxicity indicator. This may suggest that a toxic substance is slowly and continuously entering the water body.

[0096] In another scenario, the long-term trend and periodic fluctuations of the generated electrical signal values ​​were within the normal range, but its residual component suddenly exhibited a spike with an instantaneous amplitude far exceeding the preset threshold, and its duration was extremely short. The system identified this specificity of the residual component. This may indicate that a large amount of acutely toxic substance entered the system in a short period of time, causing a drastic instantaneous change in the generated electrical signal.

[0097] In both cases, whether the long-term trend anomaly is flagged as a potential indicator of long-term evolutionary toxicity or the residual component is specific, the system will identify the anomalous behavior and further analyze its characteristics, providing a basis for subsequent toxicity pattern learning and early warning.

[0098] In some of the embodiments described above in this application, the judgment process for determining whether an anomaly in the trend of electrochemical signal data is inconsistent with known non-toxic factors may face complexity and uncertainty. For example, multiple factors in the aquatic environment may fluctuate simultaneously, making it difficult for a single-dimensional judgment to accurately distinguish between a true toxic event and ordinary non-toxic disturbances, thereby potentially affecting the accuracy and reliability of the early warning.

[0099] To address this, this application further proposes a more refined and multi-dimensional verification method to improve the confidence that trend anomalies are potential long-term toxicity indicators, thereby ensuring the accuracy and reliability of the early warning. When trend anomalies in electrochemical signal data do not align with known non-toxic factors, the trend anomalies are marked as potential long-term toxicity indicators, including: S601. Sample the electrochemical signal data before and after the occurrence of the trend anomaly, and extract the time dimension change characteristics from the sampled electrochemical signal data.

[0100] The variation characteristics include the instantaneous slope of the signal, the distribution of local extrema, and the energy distribution within a preset frequency range.

[0101] Specifically, sampling electrochemical signal data before and after the occurrence of trend anomalies refers to the system automatically and proactively collecting electrochemical signal data within a specific time window before and after the anomaly occurs, after detecting the anomaly. From these sampled data, temporal change characteristics can be extracted, such as the instantaneous slope of the signal, which reflects the rate and direction of signal change; the distribution of local extrema, which reveals the peak and trough characteristics and periodicity of signal fluctuations; and the energy distribution within a preset frequency range, which can indicate the presence of a specific oscillation mode or noise component in the signal. The extraction of these features helps to comprehensively characterize the dynamic behavior of electrochemical signals before and after the anomaly.

[0102] S602. Analyze the fluctuations and mutual influences of auxiliary sensor data related to electrochemical signal data within a preset time period before and after the occurrence of trend anomalies, and obtain the synchronous fluctuation characteristics of auxiliary sensor data.

[0103] This involves analyzing auxiliary sensor data related to electrochemical signal data, such as water temperature, dissolved oxygen, and turbidity, and their fluctuations and interactions within a preset time period before and after the occurrence of anomalies. The aim is to obtain the synchronous fluctuation characteristics of the auxiliary sensor data. This auxiliary data can provide broader information about the aquatic environment, and its co-changing patterns with electrochemical signal data are crucial for determining the nature of the anomaly. For example, if an abnormal trend is observed in the electrochemical signal, and water temperature or dissolved oxygen also shows significant and abnormal fluctuations, it may indicate an environmental change rather than a toxic event.

[0104] S603. Correlate the variation characteristics of electrochemical signal data with the synchronous fluctuation characteristics of auxiliary sensor data.

[0105] In practical applications, correlating the variation characteristics of electrochemical signal data with the synchronous fluctuation characteristics of auxiliary sensor data can be achieved through various data fusion and pattern recognition techniques, such as multivariate time series analysis, correlation analysis, or machine learning models. The purpose of this correlation analysis is to identify whether synchronous, non-random correlation patterns exist between different types of data.

[0106] S604. When the variation characteristics of electrochemical signal data are consistent with the synchronous fluctuation characteristics of auxiliary sensor data in time, and the variation characteristics and synchronous fluctuation characteristics differ from known short-term random fluctuation patterns, increase the confidence that the trend anomaly is a potential long-term evolutionary toxicity indicator.

[0107] Known transient random fluctuation patterns typically refer to harmless and predictable short-term signal changes caused by normal environmental disturbances (such as rainfall, daily emission fluctuations, etc.). By excluding these known patterns, we can more accurately focus on anomalies that are truly likely caused by toxic substances.

[0108] S605. Based on the increased confidence level, abnormal trends are marked as potential indicators of long-term evolutionary toxicity.

[0109] Therefore, based on the increased confidence level, the system ultimately flagged this trend anomaly as a potential long-term evolutionary toxicity indicator. This means that the anomaly is considered to be highly likely to be caused by a toxic substance and requires further attention and treatment.

[0110] This application's solution addresses the ambiguity and uncertainty inherent in basic solutions when determining whether trend anomalies are inconsistent with known non-toxic factors by introducing a multi-dimensional correlation verification mechanism. Specifically, firstly, by extracting refined temporal features from electrochemical signal data, including instantaneous slope, local extreme point distribution, and energy distribution within a preset frequency range, it can more comprehensively capture the dynamic details of signal changes, rather than relying solely on the overall trend. Secondly, by synchronously analyzing auxiliary sensor data and acquiring its synchronous fluctuation characteristics, it provides crucial environmental background information for anomalies in electrochemical signals. For example, if the electrochemical signal's power generation value continuously decreases, while auxiliary sensor data shows abnormal fluctuations in water temperature or dissolved oxygen, this multi-dimensional data correlation helps distinguish between normal fluctuations caused by environmental factors and potential toxic events. Finally, correlating these multi-dimensional features and comparing them with known transient random fluctuation patterns can effectively eliminate false anomalies caused by non-toxic factors (such as transient equipment disturbances or instantaneous environmental changes). This progressive verification mechanism makes the judgment that trend anomalies are potential indicators of long-term evolutionary toxicity more rigorous and reliable, thereby avoiding false alarms or omissions that may be caused by a single-dimensional judgment.

[0111] In some preferred embodiments, a specific example is given below. Suppose that at a certain inlet monitoring point, the electrochemical signal data shows a slow but continuous downward trend over the past few hours, which is initially flagged as an anomaly by the system. To verify whether this anomaly is a potential long-term evolutionary toxicity indicator, the system initiates a multi-dimensional correlation verification process.

[0112] First, the system samples electrochemical signal data before and after the occurrence of the abnormal power generation signal trend. For example, it collects power generation signal data from 24 hours before the anomaly to the period of its occurrence. From this data, the instantaneous slope of the power generation signal is extracted, revealing that it is consistently negative and the absolute value of the slope gradually increases. Analysis of the distribution of local extreme points shows that the signal fluctuation amplitude decreases and the periodicity weakens. Simultaneously, within a specific low-frequency range, the energy distribution exhibits an abnormal enhancement.

[0113] Secondly, the system simultaneously analyzes auxiliary sensor data related to the electrochemical signal data, such as water temperature, dissolved oxygen, and turbidity. Within the preset time period, the system found that the water temperature remained stable and the dissolved oxygen concentration did not change significantly. However, the turbidity data increased slightly after the abnormal trend began, but the increase was much smaller than the typical turbidity fluctuation pattern known to be caused by non-toxic factors such as sediment erosion.

[0114] Next, the system correlates the temporal variation characteristics of the power generation signal (a continuously decreasing slope, weakening fluctuations, and enhanced low-frequency energy) with the synchronous fluctuation characteristics of auxiliary sensor data (stable water temperature and dissolved oxygen, and a slight increase in turbidity). Through a machine learning model, the system found that this combined pattern is significantly different from known transient random fluctuation patterns (e.g., instantaneous pressure fluctuations caused by pump station start-up and shutdown).

[0115] Because the changing characteristics of the electrochemical signal data and the synchronous fluctuation characteristics of the auxiliary sensor data exhibit temporal consistency, and this combination of multi-dimensional features differs significantly from known short-term random fluctuation patterns, the system significantly increases its confidence that the trend of the electrochemical signal is a potential long-term toxicity indicator. For example, the confidence level increases from an initial 60% to 95%. Based on this increased confidence level, the system ultimately marks the downward trend of the electrochemical signal as a potential long-term toxicity indicator and triggers subsequent toxicity pattern learning and early warning processes. This multi-dimensional verification ensures the accuracy of the early warning and avoids false alarms caused by fluctuations in a single indicator.

[0116] like Figure 3 As shown in the figure, this invention also provides an electrochemical signal data analysis system for influent toxicity early warning. The system includes: The data acquisition and preprocessing module is used to acquire electrochemical signal data from multiple monitoring points, perform preliminary processing on the electrochemical signal data to generate simplified data and local anomaly information, and transmit the simplified data and local anomaly information to the central processing platform. The normal behavior definition module is used to receive simplified data on the central processing platform, identify the normal fluctuation range of electrochemical signal data under different scenarios based on the simplified data, and adjust the normal fluctuation range of electrochemical signal data according to changes in water quality to obtain the current normal behavior definition. The abnormal behavior identification module is used to identify and analyze the characteristics of abnormal behavior when local abnormal information deviates from the normal fluctuation range of the data corresponding to the current normal behavior definition, or when simplified data deviates from the normal fluctuation range of the data corresponding to the current normal behavior definition. The toxicity pattern learning module is used to identify the features of abnormal behavior as features of toxicity patterns when the features of abnormal behavior are verified through multi-dimensional association, and to add the features of toxicity patterns to the toxicity behavior pattern rule set, which includes features of different toxicity patterns. The early warning and response module is used to match the continuously flowing electrochemical signal data based on the set of toxic behavior patterns and the definition of the current normal behavior. When the electrochemical signal data matches any toxicity pattern in the set of toxic behavior patterns and deviates from the definition of the current normal behavior, or when the absolute value of the inhibition rate is greater than or equal to a preset inhibition rate threshold, it is determined to be a toxic event and an early warning message is generated. The inhibition rate is calculated as TR = (U-U0) / U0×100%, where TR is the inhibition rate, U is the power generation after the water sample is introduced, and U0 is the baseline power generation before the water sample is introduced.

[0117] This application also provides a computer-readable storage medium. All or part of the processes in the above method embodiments can be executed by a computer program instructing related hardware. This program can be stored in the computer-readable storage medium, and when executed, it can include the processes of the above method embodiments. The computer-readable storage medium can be an internal storage unit of the task execution device (including a data sending end and / or a data receiving end) of any of the foregoing embodiments, such as the hard disk or memory of the task execution device. The computer-readable storage medium can also be an external storage device of the terminal device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the terminal device. Further, the computer-readable storage medium can include both the internal storage unit of the task execution device and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by the task execution device. The computer-readable storage medium can also be used to temporarily store data that has been output or will be output.

[0118] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0119] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a readable storage medium. Based on this understanding, the technical solutions of the embodiments of this application, essentially, or the parts that contribute to the prior art, or all or part of the technical solutions, can be embodied in the form of a software product. This software product is stored in a storage medium and includes several instructions to cause a device (which may be a microcontroller, chip, etc.) or processor to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, ROM, RAM, magnetic disks, or optical disks.

[0120] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be covered within the scope of protection of this application.

Claims

1. An electrochemical signal data analysis method for influent toxicity early warning, characterized in that, include: Acquire simplified data and local anomaly information of electrochemical signal data from multiple monitoring points, identify the normal fluctuation range of electrochemical signal data under different scenarios based on the simplified data, and adjust the normal fluctuation range of electrochemical signal data according to changes in water quality to obtain the current normal behavior definition; When the local anomaly information or the simplified data deviates from the current normal behavior definition, the abnormal behavior is identified and the characteristics of the abnormal behavior are analyzed. When the features of the abnormal behavior are verified through multi-dimensional association, the features of the abnormal behavior are confirmed as features of the toxicity pattern, and the features of the toxicity pattern are added to the toxicity behavior pattern rule set, which includes features of different toxicity patterns. When the continuously flowing electrochemical signal data matches any toxicity pattern in the set of toxicity behavior patterns and deviates from the current normal behavior definition, or when the absolute value of the inhibition rate is greater than or equal to a preset inhibition rate threshold, a toxicity event is determined, and an early warning message is generated. in, Generating simplified data includes: determining the average and standard deviation of the electrochemical signal data at different time periods, and using the average and standard deviation as the simplified data; Generating local anomaly information includes: determining the variation range of the average value of the electrochemical signal data; if the variation range is greater than the amplitude threshold, generating local anomaly information, the local anomaly information including the time of anomaly occurrence, duration, and variation range. The electrochemical signal data includes electrical signals generated by the microbial fuel cell; The inhibition rate is calculated as TR=(U-U0) / U0×100%, where TR is the inhibition rate, U is the power generation after the water sample is introduced, and U0 is the baseline power generation before the water sample is introduced. The characteristics of the abnormal behavior are determined through the multi-dimensional association verification, including: The characteristics of the abnormal behavior are matched with known toxic behavior patterns; When the characteristics of the anomalous behavior do not match any known toxic behavior pattern, the following parameters of the electrochemical signal are obtained within a preset time period: The decrease in magnitude ΔI = I0 - I min Where I0 is the steady-state current value before the abnormal behavior occurs, I min The minimum current value within the preset time period; The rate of descent v = ΔI / Δt, where Δt is the rate from the start of the anomaly to reaching I. min The time elapsed; Recovery time, i.e., from I min The time required to recover to 90% of I0; When the decrease magnitude is greater than a first preset threshold, the decrease rate is greater than a second preset threshold, and the recovery time is greater than a third preset threshold, the characteristics of the abnormal behavior are determined to pass the multi-dimensional correlation verification. The identification of abnormal behavior includes: Signal decomposition is performed on the local anomaly information or the simplified data to separate the long-term trend component, periodic fluctuation component and residual component in the local anomaly information or the simplified data; When the rate of change of the long-term trend component exceeds the preset normal drift range, it is marked as a trend anomaly; Determine whether the trend anomalies in the electrochemical signal data are consistent with known non-toxic factors; When the trend anomaly of the electrochemical signal data is not consistent with the known non-toxic factors, the trend anomaly is marked as a potential long-term evolutionary toxicity indicator. Anomaly detection is performed on the residual components to identify whether the residual components have specificity; if the instantaneous amplitude or duration of the residual components with specificity exceeds a preset threshold; Abnormal behavior is identified when the long-term trend anomalies in the electrochemical signal data are flagged as potential long-term evolutionary toxicity indicators, or when the residual components exhibit specificity to the toxicity pattern.

2. The electrochemical signal data analysis method for influent toxicity early warning according to claim 1, characterized in that, The step of identifying the normal fluctuation range of electrochemical signal data under different scenarios based on the simplified data includes: The simplified data is subjected to signal decomposition to separate the long-term trend component, periodic fluctuation component and residual component in the simplified data; Based on the long-term trend component, the periodic fluctuation component, and the residual component, the normal fluctuation range of the electrochemical signal data is constructed.

3. The electrochemical signal data analysis method for influent toxicity early warning according to claim 1, characterized in that, When the trend anomaly in the electrochemical signal data is not consistent with the known non-toxic factors, the trend anomaly is marked as a potential long-term evolutionary toxicity indicator, including: Electrochemical signal data before and after the occurrence of the trend anomaly are sampled, and time-dimensional change features are extracted from the sampled electrochemical signal data; the change features include the instantaneous slope of the signal, the distribution of local extreme points, and the energy distribution within a preset frequency range; The fluctuations and mutual influences of the auxiliary sensor data related to the electrochemical signal data during a preset period before and after the occurrence of the trend anomaly are analyzed to obtain the synchronous fluctuation characteristics of the auxiliary sensor data. The variation characteristics of the electrochemical signal data are correlated with the synchronous fluctuation characteristics of the auxiliary sensor data; When the variation characteristics of the electrochemical signal data are consistent with the synchronous fluctuation characteristics of the auxiliary sensor data in time, and the variation characteristics and the synchronous fluctuation characteristics differ from known short-term random fluctuation patterns, the confidence that the trend anomaly is a potential long-term evolutionary toxicity indicator is increased. Based on the increased confidence level, the trend anomaly is marked as a potential long-term evolutionary toxicity indicator.

4. The electrochemical signal data analysis method for influent toxicity early warning according to claim 1, characterized in that, After generating the warning information, the method further includes: Send a close command to the inlet valve; Increase the aeration rate to improve the dissolved oxygen concentration.

5. An electrochemical signal data analysis system for influent toxicity early warning, characterized in that, The system includes: The data acquisition and preprocessing module is used to acquire simplified data and local anomaly information of electrochemical signal data from multiple monitoring points; The normal behavior definition module is used to identify the normal fluctuation range of electrochemical signal data under different scenarios based on the simplified data, and adjust the normal fluctuation range of the electrochemical signal data according to the changes in water quality to obtain the current normal behavior definition; An abnormal behavior identification module is used to identify abnormal behavior and analyze the characteristics of the abnormal behavior when the local abnormal information or the simplified data deviates from the current normal behavior definition. The identification of abnormal behavior includes: Signal decomposition is performed on the local anomaly information or the simplified data to separate the long-term trend component, periodic fluctuation component and residual component in the local anomaly information or the simplified data; When the rate of change of the long-term trend component exceeds the preset normal drift range, it is marked as a trend anomaly; Determine whether the trend anomalies in the electrochemical signal data are consistent with known non-toxic factors; When the trend anomaly of the electrochemical signal data is not consistent with the known non-toxic factors, the trend anomaly is marked as a potential long-term evolutionary toxicity indicator. Anomaly detection is performed on the residual components to identify whether the residual components have specificity; if the instantaneous amplitude or duration of the residual components with specificity exceeds a preset threshold; Abnormal behavior is identified when the long-term trend anomaly of the electrochemical signal data is flagged as a potential long-term evolutionary toxicity indicator, or when the residual component has a specificity for the toxicity pattern. The toxicity pattern learning module is used to identify the features of the abnormal behavior as features of the toxicity pattern when the features of the abnormal behavior are verified through multi-dimensional association, and to add the features of the toxicity pattern to the toxicity behavior pattern rule set, which includes features of different toxicity patterns. The characteristics of the abnormal behavior are determined through the multi-dimensional association verification, including: The characteristics of the abnormal behavior are matched with known toxic behavior patterns; When the characteristics of the anomalous behavior do not match any known toxic behavior pattern, the following parameters of the electrochemical signal are obtained within a preset time period: The decrease in magnitude ΔI = I0 - I min Where I0 is the steady-state current value before the abnormal behavior occurs, I min The minimum current value within the preset time period; The rate of descent v = ΔI / Δt, where Δt is the rate from the start of the anomaly to reaching I. min The time elapsed; Recovery time, i.e., from I min The time required to recover to 90% of I0; When the decrease magnitude is greater than a first preset threshold, the decrease rate is greater than a second preset threshold, and the recovery time is greater than a third preset threshold, the characteristics of the abnormal behavior are determined to pass the multi-dimensional correlation verification. The early warning and response module is used to determine a toxic event and generate early warning information when the continuously flowing electrochemical signal data matches any toxicity pattern in the set of toxicity behavior patterns and deviates from the current normal behavior definition, or when the absolute value of the inhibition rate is greater than or equal to a preset inhibition rate threshold. The electrochemical signal data includes electrical signals generated by the microbial fuel cell; The inhibition rate is calculated using the formula TR=(U-U0) / U0×100%, where TR is the inhibition rate, U is the power generation after the water sample is introduced, and U0 is the baseline power generation before the water sample is introduced.

Images