Industrial-grade water quality telemetering method and terminal

Abstract

The invention relates to the technical field of water quality monitoring, in particular to an industrial-grade water quality telemetering method and terminal.The method comprises the steps that multi-modal water quality data in a target monitoring area are obtained, canonical correlation analysis operation is conducted on the multi-modal water quality data, and a projection matrix of the multi-modal water quality data is obtained; according to the projection matrix, mapping the multi-modal water quality data to a common subspace to obtain a fusion feature vector; and pre-training the fusion feature vector, obtaining a dynamic change value through time sequence prediction, and outputting target water quality data based on the dynamic change value, thereby effectively solving the problem that the existing industrial water quality telemetering architecture cannot provide a high-reliability evaluation basis for the dynamic change of water quality.

Description

Technical Field

[0001] This invention relates to the field of water quality monitoring technology, specifically to an industrial-grade water quality remote sensing method and terminal. Background Technology

[0002] Industrial water quality remote sensing is an important means of enabling real-time and continuous monitoring of water quality in the industrial environment, and it is of great significance for ensuring water environment safety and early warning of pollution incidents.

[0003] Currently, water quality remote sensing in industrial environments typically involves deploying multi-parameter sensing devices to simultaneously collect water quality data such as pH, conductivity, dissolved oxygen, and biochemical oxygen demand. Based on remote transmission technology, the water quality data is directly sent to back-end equipment, which then needs to integrate and analyze these water quality data with different characteristics to form a unified water quality status assessment.

[0004] However, due to the inherent differences in the working principles, response characteristics, and data formats of different parameter sensing devices, there is a serious heterogeneity problem among the water quality data received by the back-end system. It is difficult to effectively explore the intrinsic correlation between these water quality data and form an accurate water quality status characterization, resulting in a decrease in the final analysis accuracy and failing to provide a highly reliable assessment basis for dynamic changes in water quality. Summary of the Invention

[0005] To address the technical problem that existing industrial water quality telemetry architectures cannot provide a highly reliable assessment basis for dynamic changes in water quality, this application provides an industrial-grade water quality telemetry method and terminal.

[0006] This application provides an industrial-grade water quality remote sensing method and terminal, which adopts the following technical solution: An industrial-grade water quality remote sensing method includes: Acquire multimodal water quality data in the target monitoring area, perform canonical correlation analysis on the multimodal water quality data, and obtain the projection matrix of the multimodal water quality data; Based on the projection matrix, multimodal water quality data are mapped to a common subspace to obtain a fused feature vector; The fused feature vector is pre-trained, and the dynamic change value is obtained through time series prediction. The target water quality data is then output based on the dynamic change value.

[0007] Furthermore, the steps for acquiring multimodal water quality data in the target monitoring area include: According to the preset collection time period, the initial water quality data sequence corresponding to each target parameter is obtained by multi-parameter sensing devices deployed in the target monitoring area; Based on the initial water quality data sequence, the spatiotemporal alignment error between each target parameter is calculated. Based on the spatiotemporal alignment error, the initial water quality data sequence is dynamically time-normalized and calibrated to generate multimodal water quality data.

[0008] Furthermore, the steps for performing canonical correlation analysis on the multimodal water quality data to obtain the projection matrix of the multimodal water quality data include: A real-time observation data matrix containing the parameters of each target is constructed using a sliding time window mechanism; Canonical correlation analysis was performed on the multimodal water quality data to obtain the baseline projection matrix; Canonical correlation analysis is performed on the real-time observation data matrix and the reference projection matrix to obtain the projection matrix.

[0009] Furthermore, based on the projection matrix, the steps of mapping multimodal water quality data to a common subspace to obtain fused feature vectors include: Based on the projection matrix, multimodal water quality data are mapped to a common subspace to generate an initial fusion feature vector; Environmental interference factor analysis is performed on the initial fused feature vector, and the real-time signal-to-noise ratio compensation coefficient is calculated. The initial fused feature vector is optimized by using a real-time signal-to-noise ratio compensation coefficient to obtain the fused feature vector.

[0010] Furthermore, the steps of pre-training the fused feature vector and obtaining dynamically changing values ​​through time series prediction include: Based on the fused feature vector, time series prediction of multimodal water quality data is performed through pre-training to obtain an initial prediction sequence; Obtain the measured sequence of water quality parameters, and calculate the prediction deviation between the initial prediction sequence and the measured sequence of water quality parameters through residual analysis; The initial prediction sequence is compensated based on the prediction deviation to obtain the dynamic change value.

[0011] Furthermore, the steps for outputting target water quality data based on dynamically changing values ​​include: Confidence assessment is performed on dynamically changing values ​​to calculate real-time confidence levels. The system combines real-time confidence levels with multi-level early warning thresholds to make state decisions and output water quality early warning levels. The dynamic change values ​​are encapsulated according to the water quality early warning level, and the target water quality data is output.

[0012] Furthermore, before the step of combining real-time confidence levels with multi-level early warning thresholds to make state decisions and output the water quality early warning level, the following steps are also included: Historical water quality data is acquired, and the periodic variation patterns of the historical water quality data are analyzed through time-series clustering algorithms to obtain threshold clusters; Redundant threshold points are removed from the threshold cluster to obtain the initial multi-level early warning thresholds; Based on the fluctuation characteristics of water quality data during the preset time period, the initial multi-level early warning thresholds are calibrated in real time through a sliding window mechanism to generate multi-level early warning thresholds.

[0013] The industrial-grade water quality remote sensing terminal proposed in this application includes: The data acquisition and analysis module is used to acquire multimodal water quality data in the target monitoring area, perform canonical correlation analysis on the multimodal water quality data, and obtain the projection matrix of the multimodal water quality data. The feature fusion module is used to map multimodal water quality data to a common subspace based on the projection matrix to obtain a fused feature vector; The prediction output module is used to pre-train the fused feature vector, obtain dynamic change values ​​through time series prediction, and output target water quality data based on the dynamic change values.

[0014] Beneficial effects achieved: This application provides an industrial-grade water quality remote sensing method, comprising: acquiring multimodal water quality data in a target monitoring area; performing canonical correlation analysis on the multimodal water quality data to obtain a projection matrix of the multimodal water quality data; mapping the multimodal water quality data to a common subspace based on the projection matrix to obtain a fused feature vector; pre-training the fused feature vector; obtaining dynamic change values ​​through time series prediction; and outputting target water quality data based on the dynamic change values.

[0015] In this application, canonical correlation analysis is performed directly on the acquired multimodal water quality data at the front end, effectively solving the problem of decreased analysis accuracy caused by data heterogeneity during centralized back-end processing. Canonical correlation analysis can uncover the intrinsic correlation between multimodal water quality data with different characteristics. By calculating the projection matrix, the multimodal water quality data is mapped to a common subspace, thereby eliminating data heterogeneity between multi-parameter sensing devices and generating a unified fusion feature vector representing the water quality state. This fusion feature vector retains the key correlation information of the multimodal water quality data, providing a highly consistent input foundation for subsequent analysis. Then, by pre-training the fusion feature vector, the system can output accurate target water quality data, ultimately significantly improving the reliability of water quality dynamic change assessment. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the overall process of an industrial-grade water quality remote sensing method according to this application; Figure 2 This is a flowchart illustrating step S10 in an industrial-grade water quality remote sensing method of this application. Figure 3 This is a flowchart illustrating step S20 in an industrial-grade water quality remote sensing method according to this application. Figure 4 This is a flowchart illustrating step S30 in an industrial-grade water quality remote sensing method according to this application. Figure 5 This is a schematic diagram of a module of an industrial-grade water quality remote sensing terminal according to this application.

[0017] Explanation of icon numbers: 10. Data Acquisition and Analysis Module; 20. Feature Fusion Module; 30. Prediction Output Module; 40. Early Warning Threshold Generation Module. Detailed Implementation

[0018] The following is in conjunction with the appendix Figure 1-5 This application will be described in further detail.

[0019] In the description of this application, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the present invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0020] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0021] This application discloses an industrial-grade water quality remote sensing method and terminal.

[0022] Please refer to Figure 1 The industrial-grade water quality remote sensing method proposed in this embodiment includes steps S10 to S30: Step S10: Obtain multimodal water quality data in the target monitoring area, perform canonical correlation analysis on the multimodal water quality data, and obtain the projection matrix of the multimodal water quality data.

[0023] It should be noted that, compared with conventional industrial water quality remote sensing schemes, this embodiment proposes to process multimodal water quality data at the front end of water quality remote sensing, so as to preserve and utilize the spatiotemporal consistency and environmental context when the multimodal water quality data is generated to the greatest extent.

[0024] Conventional industrial water quality telemetry solutions involve acquiring water quality data corresponding to various target parameters at the front end, encoding, transmitting, and decoding the data before sending it to the back end, where the target water quality data is output. However, because various sensors are deployed in industrial settings, and these sensors have different data formats, response characteristics, and sampling frequencies, coupled with network latency and jitter during transmission, the water quality data arrives at the back end at different times, disrupting its simultaneity. This results in deviations in the final calculated target water quality data, making it impossible to provide a highly reliable assessment basis for dynamic changes in water quality.

[0025] Therefore, based on the aforementioned problems, this embodiment proposes to directly process the water quality data at the front end after collecting the water quality data corresponding to each target parameter. Since there are no abnormalities such as network delays or jitters during transmission, it can provide truly synchronized calculation of the target water quality data, thereby effectively eliminating the existing calculation deviations.

[0026] Specifically, after acquiring multimodal water quality data in the target monitoring area in this step, the statistical correlation patterns hidden among multimodal water quality data such as pH, conductivity, dissolved oxygen, and biochemical oxygen demand are explored in depth through canonical correlation analysis. Based on this, a projection matrix that can eliminate the heterogeneity between multi-parameter sensing devices is calculated. This process can map multimodal water quality data with different characteristics and difficult to directly fuse into a common subspace, laying the foundation for the subsequent generation of fused feature vectors with consistent characterization capabilities.

[0027] It should be noted that multimodal water quality data refers to a collection of unfused multi-source sensor data.

[0028] Step S20: Based on the projection matrix, the multimodal water quality data is mapped to a common subspace to obtain the fused feature vector.

[0029] By utilizing the transformation properties of the projection matrix, multimodal water quality data that were originally scattered in different feature spaces are mapped to a common subspace that can reflect the maximum correlation between target parameters. This eliminates the impact of differences in features between multi-parameter sensing devices on data consistency, effectively transforming heterogeneous multimodal water quality data into fusion feature vectors with a unified scale and retaining key correlation information, thus providing a consistent data foundation for subsequent water quality analysis.

[0030] It should be noted that the target parameters refer to pH value, conductivity, dissolved oxygen, biochemical oxygen demand, etc.

[0031] Step S30: Pre-train the fused feature vector, obtain the dynamic change value through time series prediction, and output the target water quality data based on the dynamic change value.

[0032] This method utilizes fused feature vectors to predict and make decisions about water quality status. Its function lies in using deep learning to study the water quality patterns reflected in the fused feature vectors through pre-training, and capturing the dynamic trends of water quality data using time series prediction. This yields dynamic change values ​​representing future water quality status. This step transforms the fused feature vectors into forward-looking dynamic change values, overcoming the limitation of conventional monitoring that only reflects the current water quality status. Finally, based on the dynamic change values, it outputs target water quality data containing prediction results and decision-making information, achieving a leap from passive monitoring to proactive early warning. This provides highly reliable trend assessment and decision-making basis for water quality safety management, significantly improving the reliability of water quality dynamic change assessment.

[0033] It should be noted that the water quality data refers to the values ​​corresponding to the target parameters such as pH, conductivity, dissolved oxygen, and biochemical oxygen demand.

[0034] Specifically, refer to Figure 2 As shown, in one feasible implementation, step S10 may include steps S11 to S16: Step S11: According to the preset collection time period, the initial water quality data sequence corresponding to each target parameter is obtained by using multi-parameter sensing devices deployed on the target monitoring area.

[0035] By pre-setting the acquisition time period in the multi-parameter sensing devices, all sensing devices are controlled to start the water quality data acquisition process synchronously at exactly the same acquisition time point, ensuring that the water quality data of various target parameters such as pH value, conductivity, dissolved oxygen and biochemical oxygen demand have strict time synchronization.

[0036] Each sensor deployed in the target monitoring area continuously collects raw signals according to this unified time reference, and converts the analog signals into digital water quality data values ​​through the built-in analog-to-digital conversion module. Finally, the initial water quality data sequence corresponding to the water quality data is generated in chronological order, which can fundamentally ensure that the timestamps of the water quality data are completely aligned. This provides a data foundation with strict temporal consistency for subsequent calculation of spatiotemporal alignment errors, thereby effectively solving the problem of data temporal disorder caused by differences in the response speed of sensor devices.

[0037] It should be noted that the preset data collection time period is the time period during which the control sensor device collects water quality data; the initial water quality data sequence consists of parameter sequences corresponding to each target parameter, such as pH value sequence, conductivity sequence, dissolved oxygen sequence, and biochemical oxygen demand sequence.

[0038] Step S12: Based on the initial water quality data sequence, calculate the spatiotemporal alignment error between each target parameter.

[0039] From the initial water quality data sequence, one is selected as the reference parameter sequence, such as the conductivity sequence. The parameter sequences corresponding to other target parameters are dynamically time-warped and matched with the reference parameter sequence. By finding the warping path that minimizes the cumulative distance between each parameter sequence, the time offset and morphological distortion of each parameter sequence relative to the reference parameter sequence are calculated. Together, they constitute the spatiotemporal alignment error, which accurately quantifies the degree of temporal asynchrony and waveform distortion caused by the inherent characteristics of different sensing devices such as response delay and sampling jitter. This provides an accurate basis for subsequent data calibration, thereby effectively eliminating the spatiotemporal heterogeneity of multimodal water quality data.

[0040] Step S13: Dynamically time-normalize and calibrate the initial water quality data sequence based on the spatiotemporal alignment error to generate multimodal water quality data.

[0041] Based on the time offset in the spatiotemporal alignment error, a nonlinear scaling transformation is performed on each parameter sequence to align the time axis of the reference sequence with the time axes of other parameter sequences. Simultaneously, waveform amplitude is compensated and corrected according to morphological distortion to eliminate phase deviation and amplitude distortion caused by differences in the response characteristics of sensing devices. Then, a linear calibration model based on the least squares method is used to correct the amplitude values ​​of other aligned parameter sequences with the reference sequence, generating time-synchronized and amplitude-accurate multimodal water quality data. This operation can effectively eliminate the spatiotemporal representation differences between multi-parameter sensing devices, achieve data isomorphism, and provide spatiotemporally consistent multimodal water quality data for subsequent typical correlation analysis operations.

[0042] Step S14: Construct a real-time observation data matrix containing the parameters of each target using a sliding time window mechanism.

[0043] By setting a fixed-length sliding time window, such as including the most recent 60 sampling points, the time series data of each water quality data point within the sliding time window is extracted backward from the latest acquisition time. These parameter sequences arranged in chronological order are used as column vectors, and real-time observation data matrices are constructed by combining them according to data types. Each row can be set to represent a sampling time, and each column can represent all observation values ​​of a water quality data point. This transforms the dynamically changing streaming data into a structured matrix representation, which not only preserves the temporal correlation characteristics between multiple target parameters but also eliminates random fluctuations during the data acquisition process. This provides data input that reflects changes in water quality status for subsequent typical correlation analysis operations.

[0044] Step S15: Perform canonical correlation analysis on the multimodal water quality data to obtain the baseline projection matrix.

[0045] Multimodal water quality data are divided into two sets of variables based on parameter type. For example, pH and conductivity are used as the first set of variables, and dissolved oxygen and biochemical oxygen demand are used as the second set. By calculating the covariance matrix between the two sets of variables and the cross-set covariance matrix, the characteristic equation is solved to obtain the eigenvector corresponding to the largest eigenvalue. This eigenvector constitutes the initial projection matrix. Then, the initial projection matrix is ​​orthogonalized using the singular value decomposition algorithm to finally obtain the reference projection matrix that satisfies the orthogonality constraint conditions.

[0046] The above operations can extract the most stable statistical correlation patterns among water quality data, establish a benchmark projection matrix to eliminate inherent differences in sensing devices, and provide data support for subsequent mining of essential correlations among multi-source water quality data in the feature space.

[0047] Step S16: Perform canonical correlation analysis on the real-time observation data matrix and the reference projection matrix to obtain the projection matrix.

[0048] The real-time observation data matrix is ​​grouped according to the target parameters to maintain structural consistency with the reference projection matrix. Then, using the reference projection matrix as a constraint, the covariance matrix of the real-time observation data matrix and its cross-group covariance matrix with the reference projection matrix are calculated. By solving the generalized characteristic equation, the feature vector set reflecting the current data characteristics is obtained. Then, the feature vector set is linearly weighted and fused with the reference projection matrix. Finally, the fused matrix is ​​orthogonalized using the singular value decomposition algorithm to obtain the projection matrix.

[0049] It can achieve high-precision spatiotemporal alignment and deep fusion of multi-source water quality data, providing an accurate and consistent data foundation for subsequent water quality status assessment, and effectively improving the reliability and prediction accuracy of water quality telemetry.

[0050] Specifically, refer to Figure 3 As shown, in one feasible implementation, step S20 may include steps S21 to S23: Step S21: Based on the projection matrix, the multimodal water quality data is mapped to a common subspace to generate an initial fusion feature vector.

[0051] Multimodal water quality data is used as the input matrix and matrix multiplication is performed with the projection matrix. Through linear transformation, the water quality data are transformed from their independent feature spaces to a common subspace with the greatest correlation. In this process, each weight coefficient of the projection matrix corresponds to the contribution of different water quality data in the fusion process. By weighted combination, the dimensional differences and scale fluctuations between different water quality data are eliminated. Finally, an initial fusion feature vector with unified dimension and consistent representation is output, thus solving the heterogeneity problem of different water quality data.

[0052] Step S22: Perform environmental interference factor analysis on the initial fused feature vector and calculate the real-time signal-to-noise ratio compensation coefficient.

[0053] A baseline eigenvalue distribution model of the initial fused feature vector under a standard environment is established. Then, the Mahalanobis distance between the eigenvalues ​​of each dimension of the initial fused feature vector and the baseline model is calculated in real time. This distance value quantifies the degree of feature shift caused by environmental interference. Simultaneously, based on a sliding window mechanism, short-time fluctuation components of the initial fused feature vector are extracted. These short-time fluctuation components are decomposed into intrinsic mode functions of different frequencies using the Hilbert-Huang transform. High-frequency mode components that match the frequency characteristics of environmental interference are selected, and their high-frequency energy proportion is calculated. Finally, the feature shift degree and the high-frequency energy proportion are weighted and fused to generate a real-time signal-to-noise ratio (SNR) compensation coefficient. This real-time SNR compensation coefficient can accurately characterize the distortion degree caused by the current environmental interference to the initial fused feature vector, providing a precise compensation basis for subsequent optimization of the initial fused feature vector.

[0054] Step S23: Optimize the initial fused feature vector using the real-time signal-to-noise ratio compensation coefficient to obtain the fused feature vector.

[0055] The real-time signal-to-noise ratio compensation coefficient is used as the weight index of the initial fused feature vector. By establishing the mapping relationship between the real-time signal-to-noise ratio compensation coefficient and the initial fused feature vector, the components of each dimension in the initial fused feature vector are adaptively weighted. When the real-time signal-to-noise ratio compensation coefficient shows that a certain dimension is severely disturbed, that is, the signal-to-noise ratio is low, its weight is reduced. When the disturbance is slight, that is, the signal-to-noise ratio is high, its weight is increased.

[0056] Simultaneously, based on the inverse projection reconstruction algorithm, the weighted components of each dimension are projected onto the noise subspace and the signal subspace. The main water quality features are retained in the signal subspace, while frequency components related to environmental interference are filtered out in the noise subspace. Finally, the optimized fusion feature vector is reconstructed. This operation can dynamically adjust the contribution rate of each dimension component in the initial fusion feature vector according to the real-time environmental interference level, effectively suppressing feature distortion caused by environmental factors such as temperature fluctuations and water turbidity. This makes the output fusion feature vector more essentially represent changes in water quality status, providing a data foundation with strong anti-interference ability and high reliability for subsequent time series prediction.

[0057] Specifically, refer to Figure 4 As shown, in one feasible implementation, step S30 may include steps S31 to S36: Step S31: Based on the fused feature vector, perform time series prediction on multimodal water quality data through pre-training to obtain the initial prediction sequence.

[0058] The LSTM model is trained on a historical dataset consisting of multimodal water quality data accumulated over historical monitoring cycles and calibrated in a spatiotemporal manner, along with their corresponding fused feature vectors. This dataset learns the nonlinear mapping relationship between the fused feature vectors and the multimodal water quality data. The LSTM model captures the time-dependent characteristics of the multimodal water quality data through its internal gating mechanism and cell state.

[0059] In the actual prediction stage, the real-time generated fusion feature vector sequence is input into the LSTM model in chronological order. The LSTM model outputs the predicted values ​​of multimodal water quality data for multiple future time steps through forward computation, thereby forming the initial prediction sequence.

[0060] By utilizing the multi-parameter collaborative change information contained in the fused feature vectors and extracting the dynamic patterns of water quality changes through the time-series modeling capability of the deep learning model, namely the LSTM model, we can ultimately achieve accurate prediction of the future change trends of multiple water quality data such as pH, conductivity, dissolved oxygen, and biochemical oxygen demand, providing a time window for water quality early warning that is ahead of actual changes.

[0061] Step S32: Obtain the measured sequence of water quality parameters, and calculate the prediction deviation between the initial prediction sequence and the measured sequence of water quality parameters through residual analysis.

[0062] After generating the initial prediction sequence in step S31, the measured sequence of water quality parameters with the same environmental factors, such as water temperature and pH baseline, and the same meteorological conditions, such as air temperature and rainfall, is extracted from the historical database as a verification benchmark to ensure data comparability.

[0063] The numerical difference between the initial predicted sequence and the measured sequence of water quality parameters is calculated by the point-by-point difference algorithm. After forming the residual sequence, the root mean square error algorithm is used for statistical analysis to quantify the prediction deviation between the predicted sequence and the measured sequence. This can verify the accuracy of the prediction in different time dimensions with similar water quality change patterns, and provide objective prediction deviation for subsequent compensation and correction of dynamic change values, thereby significantly improving the reliability of the prediction results in practical applications.

[0064] Step S33: Compensate the initial prediction sequence based on the prediction deviation to obtain the dynamic change value.

[0065] The least squares method is used to fit the prediction deviation into an error function that varies with time. This error function quantifies the inherent deviation pattern. Then, a corresponding compensation sequence is generated based on this error function. The compensation sequence is then superimposed point by point with the initial prediction sequence to obtain the dynamic change value. This corrects the initial prediction sequence and eliminates systematic errors caused by environmental changes or equipment aging. The corrected dynamic change value is closer to the true trend of water quality data, thus significantly improving the accuracy and reliability of time series prediction results.

[0066] Step S34: Evaluate the confidence level of the dynamically changing values ​​and calculate the real-time confidence level.

[0067] First, the residuals between the initial predicted sequence and the measured water quality parameter sequence are calculated to form a historical prediction deviation sequence. Then, a sliding window mechanism is used to calculate the mean and variance of the historical prediction deviation sequence. The historical deviation range is constructed with the mean as the center and twice the variance as the interval. The deviation between the dynamic change value and the historical change value is calculated, and the probability that the deviation falls within the historical deviation range is determined. The probability is converted into a real-time confidence level through a fuzzy logic algorithm. This real-time confidence level can objectively reflect the reliability level of the dynamic change value relative to the historical performance, providing a quantitative basis for credibility for subsequent early warning decisions, thereby effectively reducing the risk of false alarms caused by unreliable prediction results.

[0068] It should be noted that historical change values ​​are those in the historical database that share the same environmental factors and meteorological conditions as dynamic change values.

[0069] Step S35: Combine real-time confidence level with multi-level early warning threshold to make state decisions and output water quality early warning level.

[0070] The real-time confidence level is used as an adaptive weighting factor to dynamically adjust the multi-level early warning thresholds obtained in advance through training with historical water quality data. That is, when the confidence level is high, the original threshold is maintained and strictly judged, and when the confidence level is low, the threshold range is widened proportionally, thus forming an adaptive threshold range.

[0071] Next, a decision model based on state transition logic is adopted. According to the distribution of dynamic change values ​​in different adaptive threshold ranges, combined with its real-time confidence, a comprehensive state judgment is made, and the water quality warning level is output from normal, attention, warning to alarm. In this way, the reliability quantification index of the prediction results is directly integrated into the decision-making process, so as to realize the dynamic adjustment of the warning strategy according to the data credibility. Ultimately, the intelligent warning effect can not only capture water quality anomalies in a timely manner but also effectively avoid false alarms, significantly improving the decision reliability of the water quality telemetry system.

[0072] Step S36: Encapsulate the dynamic change values ​​according to the water quality early warning level and output the target water quality data.

[0073] The urgency identifier of the data encapsulation format is determined based on the water quality early warning level. The dynamic change value is used as the core data field, and metadata such as early warning level, real-time confidence level, timestamp, and monitoring area coordinates are embedded. Then, the data frame structure specified by the HJ212 protocol is used for encoding and assembly. Through the field separator and check code rules specified by the protocol, various data are combined into a transmission message that conforms to industry standards. Finally, the target water quality data containing complete early warning decision information and accurate prediction results is output. This encapsulation method can integrate the multi-dimensional information generated in the analysis process into a standardized data product that can be directly identified by the monitoring system, realizing a seamless connection from predictive analysis to decision execution.

[0074] It should be noted that steps S37 to S39 are also included between step S35: Step S37: Obtain historical water quality data, analyze the periodic variation pattern of historical water quality data through time-series clustering algorithm, and obtain threshold clusters.

[0075] By extracting long-term accumulated historical water quality data from historical databases deployed in the target monitoring area, a time-series clustering algorithm based on dynamic time warping is used to analyze the periodic variation patterns of historical water quality data. Specifically, the historical water quality data is first divided into equal-length periods according to the annual cycle, and water quality feature vectors for each period are extracted. After calculating the similarity distance matrix between water quality feature vectors of different periods using the time-series clustering algorithm, the K-shapes clustering algorithm is used to group periods with similar morphological variation patterns into the same category, forming data clusters that represent different hydrological characteristics. Finally, the quantile characteristics of the statistical distribution of water quality data are calculated for each data cluster, and each quantile point is used as the warning threshold under the corresponding hydrological conditions, ultimately forming threshold clusters corresponding to different periodic variation patterns.

[0076] Step S38: Remove redundant threshold points from the threshold cluster to obtain the initial multi-level early warning threshold.

[0077] All threshold points in the threshold cluster are mapped to the numerical space to form a data point set. A density peak search algorithm based on kernel density estimation is used to calculate the local density and relative distance of each threshold point in the numerical space. Threshold points with both high local density and high relative distance are selected as core threshold points. Redundant threshold points with low density and too close to the peak point are removed. The selected core threshold points are sorted by numerical value and classified into levels to form a simplified initial multi-level warning threshold.

[0078] Step S39: Based on the water quality data fluctuation characteristics during the preset time acquisition period, the initial multi-level early warning threshold is calibrated in real time through a sliding window mechanism to generate multi-level early warning thresholds.

[0079] From the water quality data collected in real time by multi-parameter sensing devices deployed in the target monitoring area, continuous monitoring values ​​of water quality data within a preset time period are extracted to form a data sequence. The ratio of the standard deviation to the mean of this data sequence is calculated to obtain the water quality data fluctuation characteristics. Then, the water quality data fluctuation characteristics are used as an adjustment factor and weighted and fused with the initial multi-level early warning threshold. Specifically, when the water quality data fluctuation characteristics are large, the threshold range is expanded accordingly to adapt to the data fluctuations; when the water quality data fluctuation characteristics are small, the threshold range is kept relatively tight. Finally, based on the weighted calculation results, a multi-level early warning threshold matching the water quality fluctuation characteristics is generated, so that the early warning threshold can be adaptively adjusted with the dynamic changes in water quality characteristics, effectively avoiding false alarms caused by normal fluctuations in water quality while ensuring early warning sensitivity.

[0080] This application also provides an industrial-grade water quality remote sensing terminal, referring to... Figure 5 As shown, this industrial-grade water quality remote sensing terminal includes: The data acquisition and analysis module 10 is used to acquire multimodal water quality data in the target monitoring area, perform canonical correlation analysis on the multimodal water quality data, and obtain the projection matrix of the multimodal water quality data. The feature fusion module 20 is used to map multimodal water quality data to a common subspace based on the projection matrix to obtain a fused feature vector; The prediction output module 30 is used to pre-train the fused feature vector, obtain the dynamic change value through time series prediction, and output the target water quality data based on the dynamic change value.

[0081] Optionally, the data acquisition and analysis module 10 is also used for: According to the preset collection time period, the initial water quality data sequence corresponding to each target parameter is obtained by multi-parameter sensing devices deployed in the target monitoring area; Based on the initial water quality data sequence, the spatiotemporal alignment error between each target parameter is calculated. Based on the spatiotemporal alignment error, the initial water quality data sequence is dynamically time-normalized and calibrated to generate multimodal water quality data.

[0082] Optionally, the data acquisition and analysis module 10 is also used for: A real-time observation data matrix containing the parameters of each target is constructed using a sliding time window mechanism; Canonical correlation analysis was performed on the multimodal water quality data to obtain the baseline projection matrix; Canonical correlation analysis is performed on the real-time observation data matrix and the reference projection matrix to obtain the projection matrix.

[0083] Optionally, the feature fusion module 20 is also used for: Based on the projection matrix, multimodal water quality data are mapped to a common subspace to generate an initial fusion feature vector; Environmental interference factor analysis is performed on the initial fused feature vector, and the real-time signal-to-noise ratio compensation coefficient is calculated. The initial fused feature vector is optimized by using a real-time signal-to-noise ratio compensation coefficient to obtain the fused feature vector.

[0084] Optionally, the prediction output module 30 is also used for: Based on the fused feature vector, time series prediction of each target parameter is performed through pre-training to obtain the initial prediction sequence; Obtain the measured sequence of water quality parameters within a preset time period, and calculate the prediction deviation between the initial prediction sequence and the measured sequence of water quality parameters through residual analysis; The initial prediction sequence is compensated based on the prediction deviation to obtain the dynamic change value.

[0085] Optionally, the prediction output module 30 is also used for: Confidence assessment is performed on dynamically changing values ​​to calculate real-time confidence levels. The system combines real-time confidence levels with multi-level early warning thresholds to make state decisions and output water quality early warning levels. The dynamic change values ​​are encapsulated according to the water quality early warning level, and the target water quality data is output.

[0086] The industrial-grade water quality remote sensing terminal also includes an early warning threshold generation module 10, used for: Historical water quality data is acquired, and the periodic variation patterns of the historical water quality data are analyzed through time-series clustering algorithms to obtain threshold clusters; Redundant threshold points are removed from the threshold cluster to obtain the initial multi-level early warning thresholds; Based on the fluctuation characteristics of water quality data during the preset time period, the initial multi-level early warning thresholds are calibrated in real time through a sliding window mechanism to generate multi-level early warning thresholds.

[0087] The above are all preferred embodiments of this application, and are not intended to limit the scope of protection of this application. Therefore, all equivalent changes made in accordance with the structure, shape and principle of this application should be covered within the scope of protection of this application.

Claims

1. An industrial-grade water quality remote sensing method, characterized in that, include: Acquire multimodal water quality data in the target monitoring area, perform canonical correlation analysis on the multimodal water quality data, and obtain the projection matrix of the multimodal water quality data; Based on the projection matrix, the multimodal water quality data is mapped to a common subspace to obtain a fused feature vector; The fused feature vector is pre-trained, and dynamic change values ​​are obtained through time series prediction. Target water quality data is then output based on these dynamic change values.

2. The industrial-grade water quality remote sensing method according to claim 1, characterized in that, The steps for acquiring multimodal water quality data in the target monitoring area include: According to the preset collection time period, the initial water quality data sequence corresponding to each target parameter is obtained by multi-parameter sensing devices deployed on the target monitoring area; Based on the initial water quality data sequence, the spatiotemporal alignment error between each of the target parameters is calculated; The initial water quality data sequence is dynamically time-normalized and calibrated based on the spatiotemporal alignment error to generate the multimodal water quality data.

3. The industrial-grade water quality remote sensing method according to claim 2, characterized in that, The step of performing canonical correlation analysis on the multimodal water quality data to obtain the projection matrix of the multimodal water quality data includes: A real-time observation data matrix containing the parameters of each target is constructed using a sliding time window mechanism; Canonical correlation analysis was performed on the multimodal water quality data to obtain the baseline projection matrix; The projection matrix is ​​obtained by performing canonical correlation analysis on the real-time observation data matrix and the reference projection matrix.

4. The industrial-grade water quality remote sensing method according to claim 1, characterized in that, The step of mapping the multimodal water quality data to a common subspace based on the projection matrix to obtain the fused feature vector includes: Based on the projection matrix, the multimodal water quality data is mapped to the common subspace to generate an initial fusion feature vector; Environmental interference factor analysis is performed on the initial fused feature vector to calculate the real-time signal-to-noise ratio compensation coefficient; The initial fused feature vector is optimized using the real-time signal-to-noise ratio compensation coefficient to obtain the fused feature vector.

5. The industrial-grade water quality remote sensing method according to claim 2, characterized in that, The step of pre-training the fused feature vector and obtaining dynamically changing values ​​through time series prediction includes: Based on the fused feature vector, time series prediction is performed on the multimodal water quality data through pre-training to obtain an initial prediction sequence; Obtain the measured sequence of water quality parameters, and calculate the prediction deviation between the initial prediction sequence and the measured sequence of water quality parameters through residual analysis; The initial prediction sequence is compensated based on the prediction deviation to obtain the dynamic change value.

6. The industrial-grade water quality remote sensing method according to claim 5, characterized in that, The step of outputting target water quality data based on the dynamically changing values ​​includes: The confidence level of the dynamically changing values ​​is evaluated, and the real-time confidence level is calculated. The real-time confidence level is combined with multi-level early warning thresholds to make state decisions and output the water quality early warning level. The dynamic change value is encapsulated according to the water quality early warning level, and the target water quality data is output.

7. The industrial-grade water quality remote sensing method according to claim 6, characterized in that, Before the step of combining the real-time confidence level with the multi-level early warning threshold to make a state decision and output the water quality early warning level, the following steps are also included: Historical water quality data is acquired, and the periodic variation patterns of the historical water quality data are analyzed using a time-series clustering algorithm to obtain threshold clusters; Redundant threshold points are removed from the threshold cluster to obtain the initial multi-level early warning thresholds; Based on the fluctuation characteristics of water quality data during a preset time period, the initial multi-level early warning threshold is calibrated in real time using a sliding window mechanism to generate the multi-level early warning threshold.

8. An industrial-grade water quality remote sensing terminal, characterized in that, The industrial-grade water quality remote sensing terminal is applied to the industrial-grade water quality remote sensing method as described in any one of claims 1 to 7, wherein the industrial-grade water quality remote sensing terminal comprises: The data acquisition and analysis module is used to acquire multimodal water quality data in the target monitoring area, perform canonical correlation analysis on the multimodal water quality data, and obtain the projection matrix of the multimodal water quality data. The feature fusion module is used to map the multimodal water quality data to a common subspace based on the projection matrix to obtain a fused feature vector; The prediction output module is used to pre-train the fused feature vector, obtain dynamic change values ​​through time series prediction, and output target water quality data based on the dynamic change values.

Images