A dynamic calibration method and system for environmental monitoring instrument

Abstract

The present application relates to the field of data processing technology, and provides a method and system for dynamic calibration of environmental monitoring instruments. The method includes: determining the target industrial zone and constructing an environmental pollution database; obtaining standard environmental attributes based on the environmental pollution database and configuring a standard environment; obtaining basic environmental calibration data and water quality pollution calibration data using a standard monitoring instrument under a standard environment; obtaining target basic environmental data and target water quality pollution data using a target monitoring instrument; introducing a Bayesian optimization mechanism and constructing two calibration models; inputting the target data into the two calibration models to obtain the target calibration coefficient; calibrating the target data in real time based on the target calibration coefficient to obtain calibration data. The present application solves the technical problem of fluctuations in measurement accuracy of environmental monitoring instruments under different environmental conditions due to environmental changes, and achieves the effect of reducing calibration errors caused by environmental changes and improving the reliability and accuracy of monitoring data by constructing a dynamic calibration model.

Description

Technical Field

[0001] The present application relates to the field of data processing technology, and in particular to a dynamic calibration method and system for an environmental monitoring instrument. Background Art

[0002] With the development of industrial areas and the increase in production activities, water pollution monitoring has become particularly important. However, with the expansion of the scale of industrial areas and the complexity of pollution sources, the calibration process of water pollution monitoring instruments is also facing increasingly severe challenges. Traditional environmental monitoring instrument calibration methods usually rely on fixed parameter settings and static environmental conditions, which are difficult to cope with the complex and changeable water pollution environment in industrial areas. This makes it difficult to ensure the accuracy and reliability of monitoring data, especially during the peak period of pollutant emissions. The problems of low control accuracy, slow response speed and poor stability of traditional calibration methods are more prominent, and cannot meet the high requirements of data accuracy and real-time performance for water pollution monitoring in modern industrial areas. Summary of the invention

[0003] The present application provides a dynamic calibration method and system for an environmental monitoring instrument, aiming to solve the technical problem of fluctuations in measurement accuracy of an environmental monitoring instrument caused by environmental changes under different environmental conditions.

[0004] In view of the above problems, the present application provides a method and system for dynamic calibration of environmental monitoring instruments.

[0005] The first aspect disclosed in the present application provides a method for dynamic calibration of an environmental monitoring instrument, the method comprising: determining a target industrial zone, retrieving and constructing an environmental pollution database based on the target industrial zone, the environmental pollution database comprising an environmental basic attribute library and a water quality attribute library; acquiring M standard environmental attributes based on the environmental basic attribute library and the water quality attribute library, and configuring a standard environment based on the M standard environmental attributes; under the standard environment, performing water quality environmental monitoring through a standard environmental monitoring instrument to obtain basic environmental calibration data and water quality pollution calibration data; performing real-time water quality environmental monitoring through a target environmental monitoring instrument to obtain a target Basic environmental data and target water quality pollution data; introducing a Bayesian optimization mechanism, constructing a dynamic basic calibration model based on the basic environmental calibration data, and constructing a dynamic pollution calibration model based on the water quality pollution calibration data; inputting the target basic environmental data into the dynamic basic calibration model to obtain a target basic calibration coefficient, and inputting the target water quality pollution data into the dynamic pollution calibration model to obtain a target pollution calibration coefficient; based on the target basic calibration coefficient and the target pollution calibration coefficient, calibrating the target basic environmental data and the target water quality pollution data in real time to obtain calibrated environmental data and calibrated water quality pollution data.

[0006] Another aspect disclosed in the present application provides a dynamic calibration system for environmental monitoring instruments, the system comprising: a database construction module, the database construction module is used to determine a target industrial zone, retrieve and construct an environmental pollution database based on the target industrial zone, the environmental pollution database comprises an environmental basic attribute library and a water quality attribute library; a standard environment configuration module, the standard environment configuration module is used to obtain M standard environmental attributes based on the environmental basic attribute library and the water quality attribute library, and configure a standard environment based on the M standard environmental attributes; a water quality environment monitoring module, the water quality environment monitoring module is used to perform water quality environment monitoring under the standard environment through a standard environment monitoring instrument, and obtain basic environment calibration data and water quality pollution calibration data; a real-time water quality environment monitoring module is used to perform water quality environment monitoring through a target environment monitoring instrument Real-time water quality environment monitoring, acquiring target basic environment data and target water quality pollution data; a calibration model construction module, the calibration model construction module is used to introduce a Bayesian optimization mechanism, construct a dynamic basic calibration model based on the basic environment calibration data, and construct a dynamic pollution calibration model based on the water quality pollution calibration data; a calibration coefficient acquisition module, the calibration coefficient acquisition module is used to input the target basic environment data into the dynamic basic calibration model to obtain the target basic calibration coefficient, and input the target water quality pollution data into the dynamic pollution calibration model to obtain the target pollution calibration coefficient; a real-time calibration module, the real-time calibration module is used to perform real-time calibration on the target basic environment data and the target water quality pollution data based on the target basic calibration coefficient and the target pollution calibration coefficient to obtain calibrated environment data and calibrated water quality pollution data.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0008] The above-mentioned dynamic calibration method of an environmental monitoring instrument first determines a specific industrial zone as a monitoring object, and constructs a comprehensive environmental pollution database around the industrial zone. This database is divided into an environmental basic attribute library and a water quality attribute library, which respectively record the basic information of the industrial zone environment and the pollution characteristics of water quality. Subsequently, M representative standard environmental attributes are obtained from these two databases for simulating and configuring a standard environment. Under this standard environment, a series of monitoring is carried out using a calibrated standard environmental monitoring instrument to obtain basic environmental calibration data and water quality pollution calibration data. These data are used as a benchmark for subsequent calibration and verification of the accuracy of other monitoring instruments. Afterwards, the target basic environmental data and target water quality pollution data of the industrial zone are collected using the target environmental monitoring instrument that needs to be calibrated. These data reflect the current real environmental conditions of the industrial zone. In order to improve the efficiency and accuracy of data processing, a Bayesian optimization mechanism is introduced, and a dynamic basic calibration model is constructed using the basic environmental calibration data to calibrate the target basic environmental data. Similarly, based on the water quality pollution calibration data, a dynamic pollution calibration model is constructed to calibrate the target water quality pollution data. During the real-time monitoring process, the data collected by the target environmental monitoring instrument are input into the two dynamic calibration models to obtain the target basic calibration coefficient and the target pollution calibration coefficient respectively. These coefficients reflect the difference between the target data and the standard data and are used to calibrate the target data in real time. Then, by applying these calibration coefficients, the target basic environmental data and the target water pollution data are calibrated in real time, and more accurate and reliable calibrated environmental data and calibrated water pollution data are obtained. These data provide strong support for the management and control of water pollution in industrial areas.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 The figure is a flow chart of a dynamic calibration method for an environmental monitoring instrument in an embodiment.

[0012] Figure 2The following is an architecture diagram of a dynamic calibration system for an environmental monitoring instrument in one embodiment.

[0013] Explanation of the accompanying drawings: database construction module 1, standard environment configuration module 2, water quality environment monitoring module 3, real-time water quality environment monitoring module 4, calibration model construction module 5, calibration coefficient acquisition module 6, real-time calibration module 7. DETAILED DESCRIPTION

[0014] The embodiments of the present application provide a method and system for dynamic calibration of an environmental monitoring instrument to solve the technical problem of fluctuations in measurement accuracy of an environmental monitoring instrument caused by environmental changes under different environmental conditions.

[0015] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0016] It should be noted that the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or inherent to these processes, methods, products or devices.

[0017] Embodiment 1, as Figure 1 As shown, the present application provides a method for dynamic calibration of an environmental monitoring instrument, the method comprising:

[0018] A target industrial zone is determined, and an environmental pollution database is retrieved and constructed based on the target industrial zone. The environmental pollution database includes an environmental basic attribute database and a water quality attribute database.

[0019] In the embodiment of the present application, the system terminal first determines the industrial area that needs to be monitored for water quality environment, and takes the industrial area as the target industrial area. Subsequently, a communication connection is made with the data monitoring center of the target industrial area, and data related to the water quality environment of the target industrial area is collected to build a comprehensive environmental pollution database. This environmental pollution database is divided into two parts, namely, the basic environmental attribute library and the water quality attribute library. Among them, the basic environmental attribute library focuses on recording the basic environmental information of the target industrial area, including but not limited to geographical location, temperature, humidity, etc. This information provides a solid foundation for understanding the environmental background of the target industrial area, and helps to more accurately assess the influencing factors and changing trends of water pollution. The water quality attribute library focuses on recording the detailed information of various water pollutants in the target industrial area. This includes but is not limited to the types and concentrations of various toxic and harmful substances. By analyzing these data, the water pollution status of the target industrial area can be more accurately understood, providing a basis for subsequent data calibration.

[0020] Based on the basic environment attribute library and the water quality attribute library, M standard environment attributes are acquired, and a standard environment is configured based on the M standard environment attributes.

[0021] In one embodiment, based on the already constructed basic environmental attribute library and water quality attribute library, the system terminal performs data mining and analysis to extract M representative standard environmental attributes. In this data mining and analysis process, the system terminal extracts M commonly existing environmental attributes from the basic environmental attribute library and water quality attribute library through multiple clustering, and uses these extracted environmental attributes as standard environmental attributes to comprehensively and accurately reflect the typical environmental characteristics of the target industrial area, including climate, topography, types of water pollutants and concentration levels. Subsequently, using the M standard environmental attributes as a benchmark, a standard environment is configured, providing a stable and reliable reference benchmark for subsequent monitoring instrument calibration and data calibration work.

[0022] Furthermore, the present application provides a method for obtaining M standard environmental attributes based on the basic environmental attribute library and the water quality attribute library, including:

[0023] Historical monitoring data is extracted from the environmental pollution database, the historical monitoring data including at least I basic environmental attributes and J water quality pollution attributes; environmental attribute clustering is performed on the historical monitoring data to obtain I basic environmental attribute clustering clusters and J water quality pollution attribute clustering clusters, the basic environmental attribute clustering clusters correspond one-to-one to the basic environmental attributes, and the water quality pollution attribute clustering clusters correspond one-to-one to the water quality pollution attributes.

[0024] Preferably, after building an environmental pollution database, the system terminal extracts a large amount of historical monitoring data from the basic environmental attributes library and water quality attribute library in the environmental pollution database. These historical data cover I basic environmental attributes and J water pollution attributes. Basic environmental attributes refer to natural environmental factors such as temperature and humidity, while water pollution attributes refer to indicators such as chemical oxygen demand (COD), ammonia nitrogen concentration and other indicators related to water pollution. Subsequently, the system terminal uses the maximum-minimum method to standardize the historical monitoring data, that is, the difference between the data of each attribute and the minimum value of the attribute and the difference between the maximum value and the minimum value of the attribute are used to perform ratio calculation to obtain standard historical monitoring data. After the standardization process is completed, the system terminal uses each standard data point in the standard historical monitoring data as a separate cluster cluster, and then uses the Euclidean distance to calculate the distance between every two cluster clusters to construct the basic environmental attribute distance matrix and the water pollution attribute distance matrix. After that, the two clusters with the smallest corresponding distance are extracted from the basic environmental attribute distance matrix and the water pollution attribute distance matrix to merge and form a new cluster, and then the basic environmental attribute distance matrix and the water pollution attribute distance matrix are updated to reflect the new cluster structure. Repeat the above steps until the distance between the clusters exceeds the predetermined distance threshold. At this time, the system terminal outputs the current cluster to generate I basic environmental attribute clusters and J water pollution attribute clusters. Each basic environmental attribute corresponds to a basic environmental attribute cluster, which reflects the change pattern and law of the attribute in different time periods or under different conditions; similarly, each water pollution attribute also corresponds to a water pollution attribute cluster, which reveals the change trend and characteristics of the water pollution status. By extracting historical monitoring data and performing cluster analysis on it, the complex environmental attribute data is successfully converted into a series of representative and regular clusters. This process not only helps to better understand the changing laws of environmental conditions, but also provides strong data support for the subsequent standard environment construction.

[0025] Based on the I basic environmental attribute clusters and the J water quality pollution attribute clusters, sub-cluster analysis is performed to obtain I basic sub-cluster sets and J water quality pollution sub-cluster sets, wherein the basic sub-cluster set includes multiple basic sub-cluster clusters, and the water quality pollution sub-cluster set includes multiple water quality pollution sub-cluster clusters; priority extraction is performed on the I basic sub-cluster sets and the J water quality pollution sub-cluster sets to construct the M standard environmental attributes, wherein I+J=M.

[0026] Preferably, after obtaining I basic environmental attribute clusters and J water pollution attribute clusters, the system terminal uses the same clustering method as above to perform a more in-depth sub-cluster analysis on I basic environmental attribute clusters and J water pollution attribute clusters. Inside each basic environmental attribute and water pollution attribute cluster, smaller cluster units are subdivided, namely basic sub-cluster clusters and water pollution sub-cluster clusters. This is done to capture the changes and patterns of environmental attributes and water pollution characteristics more finely. Subsequently, the system terminal performs priority extraction on these basic sub-cluster sets and water pollution sub-cluster sets based on the priority determination threshold. The purpose of this step is to identify the clusters that have the greatest impact on the overall environmental conditions or water pollution conditions and are the most representative from a large number of sub-clusters. Based on the results of priority extraction, the system terminal constructs M standard environmental attributes. Wherein, M is equal to I plus J, that is, each basic environmental attribute cluster and water pollution attribute cluster corresponds to a standard environmental attribute. These standard environmental attributes represent the key aspects of the environmental conditions and can be used for subsequent environmental data calibration.

[0027] Further, the present application provides a method for performing priority extraction on the I basic sub-cluster sets and the J water pollution sub-cluster sets, including:

[0028] Aggregate and count the I basic sub-clustering sets and the J water quality pollution sub-clustering sets to generate I basic quantity distribution matrices and J water quality pollution quantity distribution matrices; obtain a priority judgment threshold, perform sub-clustering cluster quantity threshold screening based on the I basic quantity distribution matrices, the J water quality pollution quantity distribution matrices and the priority judgment threshold, and identify N preferred sub-clustering clusters, where N≥I+J; perform same-attribute fusion on the N preferred sub-clustering clusters to obtain the M standard environmental attributes.

[0029] Optionally, the system terminal aggregates and counts I basic sub-cluster sets and J water pollution sub-cluster sets. The purpose of this step is to understand the distribution of data points in each sub-cluster cluster, thereby generating a basic quantity distribution matrix and a water pollution quantity distribution matrix. These matrices reflect the number and density of data points in different sub-cluster clusters, which helps to understand the importance and representativeness of each sub-cluster cluster. Subsequently, a priority judgment threshold is obtained, which is set based on business needs and historical experience, and is used to screen out sub-cluster clusters that meet certain standards in quantity or importance. Based on this threshold, the sub-cluster clusters in the basic quantity distribution matrix and the water pollution quantity distribution matrix are screened by quantity threshold. This screening process helps the system terminal identify N preferred sub-cluster clusters, where N may be greater than or equal to I+J, because each sub-cluster set may include multiple sub-cluster clusters, and the number of data points in each sub-cluster cluster may be greater than or equal to the priority judgment threshold. Afterwards, the N preferred sub-cluster clusters are fused with the same attributes. The purpose of this step is to merge sub-cluster clusters with the same attributes into a standard environmental attribute. Through fusion, the number of attributes is reduced, but the most important information is retained. In this way, the system terminal obtains M standard environmental attributes, which can comprehensively and accurately reflect the key aspects of the water quality environment.

[0030] Further, the present application provides for generating M calibration standard environment attributes, including:

[0031] An anomaly detection mechanism is introduced to perform dynamic mutation feature analysis on the multiple basic sub-clustering clusters and the multiple water quality pollution sub-clustering clusters, identify and label multiple mutation clustering clusters, wherein the anomaly detection mechanism includes an anomaly detection algorithm; density index evaluation is performed on the multiple mutation clustering clusters, and multiple key mutation clustering clusters whose density index is greater than or equal to a density index threshold are screened; environmental attribute authentication is performed on the multiple key mutation clustering clusters, and they are merged with the M standard environmental attributes to generate M calibration standard environmental attributes.

[0032] Optionally, in order to make the configured standard environment bursty, that is, to simulate the sudden increase or decrease of temperature, humidity, and certain polluted water quality, the system terminal introduces an anomaly detection mechanism to enhance the analysis ability of the basic sub-cluster clusters and water pollution sub-cluster clusters. The core of this mechanism is the anomaly detection algorithm, which can identify clusters that show anomalies by comparing the local density deviation of the data point with other data points in its neighborhood. These abnormal clusters may represent sudden changes in environmental conditions or sudden pollution events, so they are particularly important for environmental monitoring. First, the system terminal determines a suitable k value based on the distribution of the data and the needs of anomaly detection. This k value determines how many nearest neighbors each data point considers to calculate the local density. For each data point p in the sub-cluster cluster, find its k nearest neighbors (k-NN). This is achieved by calculating the distance between the data point p and all other data points through the Euclidean distance and selecting the k points with the smallest distance. For each k-NN data point o of the data point p, calculate the reachable distance of p about o, which is the larger value between the distance from p to o and the maximum distance in the k-NN of o. Based on the reachable distance, the system terminal estimates the local density of data point p using the inverse of the reachable distance. Data points with high local density represent that their k-NN data points are relatively close, while data points with low local density may be far away from their k-NN data points. For each k-NN data point o of data point p, calculate the local density of o and find the average of these local densities. Subsequently, the local density of data point p is calculated by comparing the local density of data point p with the average k-NN local density, and then the logarithm of the calculation result is taken to obtain the LOF value of data point p to avoid the influence of extreme values. If the local density of p is significantly lower than the average k-NN local density, the LOF value of p will be greater than 1. Afterwards, the above method is used to calculate the LOF values ​​of data points in multiple basic sub-clustering clusters and multiple water pollution sub-clustering clusters. If a sub-clustering cluster contains multiple data points with LOF values ​​significantly greater than 1, the sub-clustering cluster is marked as abnormal to obtain multiple mutation clustering clusters. Then, in order to determine which mutation clusters are common, the system terminal evaluates their density index, that is, counts the number of data points in each mutation cluster, and uses the counted number as the density index of the mutation cluster. The density index reflects the density of data points within the mutation cluster, which can help determine whether a mutation cluster is significant or representative enough. By screening mutation clusters with a density index greater than or equal to the density index threshold, the system terminal obtains multiple key mutation clusters. These clusters not only show significant changes in values, but also have a high density in space or time, and are therefore more likely to represent real environmental changes or pollution events.Finally, the system terminal authenticates the environmental attributes of these key mutation clusters, determines the environmental attributes corresponding to each key mutation cluster, and fuses the key mutation clusters with M standard environmental attributes according to the determined environmental attributes to generate M calibration standard environmental attributes. These calibrated attributes not only contain the original environmental information, but also incorporate the latest mutation characteristics, so they can more comprehensively reflect the current environmental conditions, better respond to sudden changes in the environment, and improve the accuracy and adaptability of calibration.

[0033] Under the standard environment, water quality environment monitoring is performed using standard environmental monitoring instruments to obtain basic environmental calibration data and water quality pollution calibration data; real-time water quality environment monitoring is performed using target environmental monitoring instruments to obtain target basic environmental data and target water quality pollution data.

[0034] In one embodiment, under a standard environment, the system terminal uses standard environmental monitoring instruments to perform comprehensive environmental monitoring to obtain basic environmental calibration data and water pollution calibration data. Among them, the basic environmental calibration data includes temperature, humidity, etc. The water pollution calibration data includes the concentration and type of pollutants, such as chemical oxygen demand and ammonia nitrogen concentration. In the target industrial area, the system terminal uses the target environmental monitoring instrument to perform real-time water quality environmental monitoring to obtain the current basic environmental data and water pollution data of the area, that is, the target basic environmental data and target water pollution data. These data reflect the environmental conditions of the target industrial area in real time, including the types and concentrations of possible water pollutants and real-time changes in environmental parameters.

[0035] A Bayesian optimization mechanism is introduced to construct a dynamic basic calibration model based on the basic environmental calibration data, and a dynamic pollution calibration model is constructed based on the water quality pollution calibration data.

[0036] In one embodiment, when constructing the model, the system terminal introduces a Bayesian optimization mechanism. Bayesian optimization is an optimization method based on a probability model. It uses Bayes' theorem to update the probability distribution of the objective function, thereby guiding the search process to move to the area where the optimal solution is most likely to be found. In environmental monitoring, the system terminal applies Bayesian optimization to the adjustment and optimization of model parameters, combining basic environmental calibration data and water pollution calibration data to build a more accurate dynamic calibration model, including a dynamic basic calibration model and a dynamic pollution calibration model. These models can not only accurately reflect the dynamic changes of environmental parameters and water pollutants, but also continuously optimize their own performance through learning, providing more reliable and powerful support for environmental monitoring.

[0037] Furthermore, the present application provides a method of introducing a Bayesian optimization mechanism to construct a dynamic basic calibration model based on the basic environmental calibration data, including:

[0038] Based on the historical monitoring data and the basic environmental calibration data, standardization processing and division are performed to obtain first training data and second training data; the machine learning model is initialized, and an optimization function with the goal of minimizing measurement errors is constructed; in a multi-dimensional hyperparameter space, based on the acquisition function and the optimization function, parameter search and optimization strategies are performed on the machine learning model according to the first training data, hyperparameter combinations are determined, and an initial dynamic basic calibration model is constructed; based on the second training data, the initial dynamic basic calibration model is deeply trained to construct a dynamic basic calibration model.

[0039] Optionally, the system terminal uses the same method as above to standardize the historical monitoring data and the basic environmental calibration data, obtain standard historical monitoring data and standard basic environmental calibration data, and then annotate the historical calibration coefficients of the standard historical monitoring data, and divide the annotated standard historical monitoring data, and combine the results of the division with the standard basic environmental calibration data to generate the first training data and the second training data. Subsequently, the system terminal initializes a deep learning model, including the number of network layers, the number of neurons, the activation function, etc. Then set the objective function to minimize the measurement error. The objective function is as follows: Where L is the objective function, which represents the error measure of the entire model. N is the number of samples, that is, the total number of samples used to calculate the objective function. true,i The actual error of the i-th sample is the actual measurement error. pred,i is the prediction error of the i-th sample, and is the error predicted by the model. After that, a multidimensional hyperparameter space of the deep learning model is defined. Bayesian optimization will find the optimal hyperparameter combination in this multidimensional hyperparameter space, such as the number of network layers, the number of neurons, the learning rate, the batch size, etc. Then, the Gaussian process is used as a proxy model for Bayesian optimization. The Gaussian process is a non-parametric Bayesian model used to simulate the relationship between hyperparameters and the objective function. Then, a set of initial hyperparameter combinations are randomly selected in the multidimensional hyperparameter space, and the deep learning model is trained using the first training data and these hyperparameters to calculate the objective function value. The Gaussian process model is updated using the calculation results to obtain the estimated distribution of the objective function in the hyperparameter space. According to the Gaussian process model, the acquisition function is used to select the next hyperparameter combination to be evaluated. The model is then trained using the selected hyperparameter combination and the objective function value is calculated. The above process is repeated until the objective function value is no longer significantly improved. Based on the results of Bayesian optimization, the system terminal determines the optimal hyperparameter combination and uses this combination to configure the deep learning model. Combined with the second training data, the deep learning model is deeply trained through forward propagation, loss calculation, back propagation, parameter update and other steps, and the optimal hyperparameter combination is fine-tuned to construct a dynamic basic calibration model to ensure that it still has good predictive ability on unseen data.

[0040] Furthermore, the present application provides the acquisition function, including:

[0041] The acquisition function is as follows: EI(x) = (μ(x) - f(x) best ))*Φ(Z)+σ(x)*ф(Z).

[0042] Optionally, in Bayesian optimization, the acquisition function is used to determine the evaluation position of the next hyperparameter combination, which can strike a balance between finding new areas and optimizing the current optimal area. The acquisition function is as follows: EI(x) = (μ(x) - f(x best ))*Φ(Z)+σ(x)*ф(Z); where EI(x) represents the expected improvement, i.e., the improvement relative to the current optimal value f(x) at the hyperparameter combination x. best ) is the expected improvement in the objective function value; x represents the current hyperparameter combination to be evaluated; μ(x) is the mean value of the objective function predicted by the Gaussian process model at the hyperparameter combination x; f(x best ) is the optimal value of the current objective function; Φ(Z) is the cumulative distribution function (CDF) of the standard normal distribution, which is used to calculate the probability that Z is less than or equal to a certain value; Z is a standardized variable used to measure the difference between the predicted value and the current optimal value. σ(x) is the square root of the variance of the Gaussian process model at the hyperparameter combination x, which reflects the uncertainty of the prediction. The larger the variance, the lower the accuracy of the prediction at that point and the higher the uncertainty. ф(Z) is the probability density function (PDF) of the standard normal distribution, which is used together with Φ(Z) to calculate the expected improvement. The acquisition function evaluates the potential value of making the next evaluation at a given hyperparameter combination x by comprehensively considering the predicted value, the uncertainty of the prediction, and the current optimal value. If the predicted value is lower than the current optimal value and the uncertainty of the prediction is moderate, then the expected improvement will be high, indicating that evaluating at this point is expected to achieve a large performance improvement. In this way, the Bayesian optimization mechanism can intelligently explore the search space to find the optimal hyperparameter combination.

[0043] The target basic environmental data are input into the dynamic basic calibration model to obtain the target basic calibration coefficient, and the target water quality pollution data are input into the dynamic pollution calibration model to obtain the target pollution calibration coefficient; based on the target basic calibration coefficient and the target pollution calibration coefficient, the target basic environmental data and the target water quality pollution data are calibrated in real time to obtain calibrated environmental data and calibrated water quality pollution data.

[0044] In one embodiment, the system terminal inputs the target basic environmental data into the dynamic basic calibration model, and inputs the target water quality pollution data into the dynamic pollution calibration model. Through the processing of the dynamic basic calibration model, the system terminal obtains a target basic calibration coefficient. These coefficients reflect the calibration or adjustment value of the basic environmental data under specific conditions. Similarly, after the dynamic pollution calibration model processes the target water quality pollution data, it outputs a target pollution calibration coefficient. This coefficient is used to calibrate or adjust the water quality pollution data to reflect a more realistic pollution condition. Subsequently, the system terminal uses the obtained target basic calibration coefficient and target pollution calibration coefficient to calibrate the target basic environmental data and the target water quality pollution data in real time, that is, multiplying the target basic calibration coefficient and the target basic environmental data, multiplying the target pollution calibration coefficient and the target water quality pollution data, and obtaining the calibration environmental data and the calibration water quality pollution data. These data are more accurate and reliable, and can better reflect the actual environmental conditions and water quality pollution levels.

[0045] Furthermore, the present application provides adaptive optimization of the dynamic basic calibration model, including:

[0046] Acquire an iteration cycle, collect the basic environmental calibration data and water pollution calibration data based on the iteration cycle, and synchronize them to a calibration database; based on the calibration database and in combination with a historical monitoring data update mechanism, adaptively optimize the dynamic basic calibration model.

[0047] Preferably, the system terminal obtains an iteration cycle according to business needs, and this iteration cycle defines how often the model optimization cycle is performed. In each iteration cycle, the system terminal collects basic environmental calibration data and water pollution calibration data under a standard environment, and synchronizes the collected data to the calibration database for subsequent analysis and processing. Subsequently, the latest data in the calibration database is used, combined with the update mechanism of historical monitoring data, to adaptively optimize the dynamic basic calibration model, that is, the aforementioned training process is performed again, and the model parameters and structure are continuously optimized to improve the prediction accuracy and generalization ability of the model, so that it can better adapt to the changes in the actual environment and water pollution. Among them, the historical monitoring data update mechanism refers to synchronizing the data for each calibration to the environmental pollution database after the processing is completed, and gradually improving the environmental pollution database. Through the above feedback mechanism, the accuracy and robustness of the calibration can be enhanced, and the adaptability and prediction ability of the calibration model can be continuously improved.

[0048] In summary, the embodiments of the present application have at least the following technical effects:

[0049] The embodiment of the present application constructs an environmental pollution database including an environmental basic attribute library and a water quality attribute library by determining the target industrial zone, so as to fully reflect the environmental and water quality pollution conditions in the industrial zone. Based on the database, M standard environmental attributes are extracted, and the standard environment is configured accordingly. Subsequently, under the standard environment, data collection is performed using a standard environmental monitoring instrument to obtain basic environmental calibration data and water quality pollution calibration data. Real-time water quality environmental monitoring is performed by the target environmental monitoring instrument, and target basic environmental data and target water quality pollution data are obtained respectively. In order to ensure the accuracy of calibration, a Bayesian optimization mechanism is introduced, and a dynamic basic calibration model and a dynamic pollution calibration model are constructed respectively using basic environmental calibration data and water quality pollution calibration data, and target basic calibration coefficients and target pollution calibration coefficients are obtained respectively according to these models. Finally, based on these calibration coefficients, the target basic environmental data and target water quality pollution data are calibrated in real time to generate calibrated environmental data and pollution data. The method also involves extracting, clustering and sub-clustering analysis of historical monitoring data from the environmental pollution database to construct accurate standard environmental attributes, and analyzing the mutation environment through anomaly detection mechanisms to ensure the adaptability and stability of the calibration model. The entire calibration process is continuously optimized through iterative cycles and data update mechanisms with the support of the Bayesian optimization mechanism to ensure that the monitoring data under complex environmental conditions in industrial areas has high accuracy and reliability. These technical effects jointly solve the technical problem of fluctuations in measurement accuracy of environmental monitoring instruments caused by environmental changes under different environmental conditions, and achieve the effect of reducing calibration errors caused by environmental changes and improving the reliability and accuracy of monitoring data by building a dynamic calibration model.

[0050] Embodiment 2, based on the same inventive concept as the method for dynamic calibration of an environmental monitoring instrument in the above embodiment, Figure 2As shown, the present application provides a dynamic calibration system for environmental monitoring instruments, and the system includes: a database construction module 1: the database construction module 1 is used to determine the target industrial zone, retrieve and construct an environmental pollution database based on the target industrial zone, and the environmental pollution database includes an environmental basic attribute library and a water quality attribute library; a standard environment configuration module 2: the standard environment configuration module 2 is used to obtain M standard environmental attributes based on the environmental basic attribute library and the water quality attribute library, and configure a standard environment based on the M standard environmental attributes; a water quality environment monitoring module 3: the water quality environment monitoring module 3 is used to perform water quality environment monitoring through a standard environment monitoring instrument under the standard environment, and obtain basic environment calibration data and water quality pollution calibration data; a real-time water quality environment monitoring module 4: the real-time water quality environment monitoring module 4 is used to perform real-time water quality environment monitoring through a target environment monitoring instrument Water quality environment monitoring, obtaining target basic environment data and target water quality pollution data; calibration model construction module 5: the calibration model construction module 5 is used to introduce the Bayesian optimization mechanism, construct a dynamic basic calibration model based on the basic environment calibration data, and construct a dynamic pollution calibration model based on the water quality pollution calibration data; calibration coefficient acquisition module 6: the calibration coefficient acquisition module 6 is used to input the target basic environment data into the dynamic basic calibration model to obtain the target basic calibration coefficient, and input the target water quality pollution data into the dynamic pollution calibration model to obtain the target pollution calibration coefficient; real-time calibration module 7: the real-time calibration module 7 is used to perform real-time calibration on the target basic environment data and the target water quality pollution data based on the target basic calibration coefficient and the target pollution calibration coefficient to obtain calibrated environment data and calibrated water quality pollution data.

[0051] Furthermore, the standard environment configuration module 2 is also used to execute the following method:

[0052] Historical monitoring data is extracted from the environmental pollution database, the historical monitoring data includes at least I basic environmental attributes and J water quality pollution attributes; environmental attribute clustering is performed on the historical monitoring data to obtain I basic environmental attribute clustering clusters and J water quality pollution attribute clustering clusters, the basic environmental attribute clustering clusters correspond to the basic environmental attributes one-to-one, and the water quality pollution attribute clustering clusters correspond to the water quality pollution attributes one-to-one; sub-clustering analysis is performed based on the I basic environmental attribute clustering clusters and the J water quality pollution attribute clustering clusters to obtain I basic sub-clustering sets and J water quality pollution sub-clustering sets, the basic sub-clustering set includes multiple basic sub-clustering clusters, and the water quality pollution sub-clustering set includes multiple water quality pollution sub-clustering clusters; priority extraction is performed on the I basic sub-clustering sets and the J water quality pollution sub-clustering sets to construct the M standard environmental attributes, wherein I+J=M.

[0053] Furthermore, the standard environment configuration module 2 is also used to execute the following method:

[0054] Aggregate and count the I basic sub-clustering sets and the J water quality pollution sub-clustering sets to generate I basic quantity distribution matrices and J water quality pollution quantity distribution matrices; obtain a priority judgment threshold, perform sub-clustering cluster quantity threshold screening based on the I basic quantity distribution matrices, the J water quality pollution quantity distribution matrices and the priority judgment threshold, and identify N preferred sub-clustering clusters, where N≥I+J; perform same-attribute fusion on the N preferred sub-clustering clusters to obtain the M standard environmental attributes.

[0055] Furthermore, the standard environment configuration module 2 is also used to execute the following method:

[0056] An anomaly detection mechanism is introduced to perform dynamic mutation feature analysis on the multiple basic sub-clustering clusters and the multiple water quality pollution sub-clustering clusters, identify and label multiple mutation clustering clusters, wherein the anomaly detection mechanism includes an anomaly detection algorithm; density index evaluation is performed on the multiple mutation clustering clusters, and multiple key mutation clustering clusters whose density index is greater than or equal to a density index threshold are screened; environmental attribute authentication is performed on the multiple key mutation clustering clusters, and they are merged with the M standard environmental attributes to generate M calibration standard environmental attributes.

[0057] Furthermore, the calibration model building module 5 is also used to execute the following method:

[0058] Based on the historical monitoring data and the basic environmental calibration data, standardization processing and division are performed to obtain first training data and second training data; the machine learning model is initialized, and an optimization function with the goal of minimizing measurement errors is constructed; in a multi-dimensional hyperparameter space, based on the acquisition function and the optimization function, parameter search and optimization strategies are performed on the machine learning model according to the first training data, hyperparameter combinations are determined, and an initial dynamic basic calibration model is constructed; based on the second training data, the initial dynamic basic calibration model is deeply trained to construct a dynamic basic calibration model.

[0059] Furthermore, the calibration model building module 5 is also used to execute the following method:

[0060] The acquisition function is as follows: EI(x) = (μ(x) - f(x) best ))*Φ(Z)+σ(x)*ф(Z); where EI(x) represents the expected improvement, i.e., the improvement relative to the current optimal value f(x) at the hyperparameter combination x. best ) is the expected improvement in the objective function value; x represents the current hyperparameter combination to be evaluated; μ(x) is the mean value of the objective function predicted by the Gaussian process model at the hyperparameter combination x; f(xbest ) is the optimal value of the current objective function; Φ(Z) is the cumulative distribution function (CDF) of the standard normal distribution; Z is a standardized variable used to measure the difference between the predicted value and the current optimal value, σ(x) is the square root of the variance of the Gaussian process model at the hyperparameter combination x; ф(Z) is the probability density function (PDF) of the standard normal distribution.

[0061] Furthermore, the real-time calibration module 7 is also used to execute the following method:

[0062] Acquire an iteration cycle, collect the basic environmental calibration data and water pollution calibration data based on the iteration cycle, and synchronize them to a calibration database; based on the calibration database and in combination with a historical monitoring data update mechanism, adaptively optimize the dynamic basic calibration model.

[0063] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. The processes depicted in the accompanying drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0064] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0065] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A dynamic calibration method for an environmental monitoring instrument, characterized in that: The method comprises: Determine a target industrial zone, and retrieve and construct an environmental pollution database based on the target industrial zone, wherein the environmental pollution database includes an environmental basic attribute database and a water quality attribute database; Acquire M standard environmental attributes based on the basic environmental attribute library and the water quality attribute library, and configure a standard environment based on the M standard environmental attributes; Under the standard environment, water quality environment monitoring is performed using standard environmental monitoring instruments to obtain basic environmental calibration data and water quality pollution calibration data; Conduct real-time water quality environmental monitoring through target environmental monitoring instruments to obtain target basic environmental data and target water quality pollution data; Introducing a Bayesian optimization mechanism, constructing a dynamic basic calibration model based on the basic environmental calibration data, and constructing a dynamic pollution calibration model based on the water pollution calibration data; Inputting the target basic environmental data into the dynamic basic calibration model to obtain a target basic calibration coefficient, and inputting the target water quality pollution data into the dynamic pollution calibration model to obtain a target pollution calibration coefficient; Based on the target basic calibration coefficient and the target pollution calibration coefficient, the target basic environmental data and the target water quality pollution data are calibrated in real time to obtain calibrated environmental data and calibrated water quality pollution data; Based on the basic environmental attribute library and the water quality attribute library, M standard environmental attributes are obtained, including: Extracting historical monitoring data from the environmental pollution database, wherein the historical monitoring data includes at least I basic environmental attributes and J water quality pollution attributes; Performing environmental attribute clustering on the historical monitoring data to obtain I basic environmental attribute clustering clusters and J water quality pollution attribute clustering clusters, wherein the basic environmental attribute clustering clusters correspond one-to-one to the basic environmental attributes, and the water quality pollution attribute clustering clusters correspond one-to-one to the water quality pollution attributes; Performing sub-cluster analysis based on the I basic environmental attribute clusters and the J water quality pollution attribute clusters to obtain I basic sub-cluster sets and J water quality pollution sub-cluster sets, wherein the basic sub-cluster set includes multiple basic sub-cluster clusters, and the water quality pollution sub-cluster set includes multiple water quality pollution sub-cluster clusters; Priority extraction is performed on the I basic sub-cluster sets and the J water pollution sub-cluster sets to construct the M standard environmental attributes, where I+J=M.

2. A method for dynamic calibration of an environmental monitoring instrument as claimed in claim 1, characterized in that: Priority extraction is performed on the I basic sub-cluster sets and the J water pollution sub-cluster sets, including: Performing aggregation statistics on the I basic sub-cluster sets and the J water quality pollution sub-cluster sets to generate I basic quantity distribution matrices and J water quality pollution quantity distribution matrices; Obtaining a priority determination threshold, performing sub-cluster quantity threshold screening based on the I basic quantity distribution matrices, the J water quality pollution quantity distribution matrices and the priority determination threshold, and identifying N preferred sub-cluster clusters, wherein N≥I+J; The N preferred sub-clusters are fused with the same attributes to obtain the M standard environment attributes.

3. A dynamic calibration method for an environmental monitoring instrument as claimed in claim 1, characterized in that: Also includes: Introducing an anomaly detection mechanism, performing dynamic mutation feature analysis on the multiple basic sub-clustering clusters and the multiple water pollution sub-clustering clusters, identifying and marking multiple mutation clustering clusters, wherein the anomaly detection mechanism includes an anomaly detection algorithm; Performing density index evaluation on the multiple mutation clusters, and screening multiple key mutation clusters whose density index is greater than or equal to a density index threshold; The multiple key mutation clusters are authenticated for environmental attributes and fused with the M standard environmental attributes to generate M calibration standard environmental attributes.

4. A method for dynamic calibration of an environmental monitoring instrument as claimed in claim 1, characterized in that: A Bayesian optimization mechanism is introduced to construct a dynamic basic calibration model based on the basic environmental calibration data, including: Performing standardization processing and division based on the historical monitoring data and the basic environment calibration data to obtain first training data and second training data; Initialize the machine learning model and construct an optimization function to minimize the measurement error; In a multidimensional hyperparameter space, based on the acquisition function and the optimization function, a parameter search and optimization strategy is performed on the machine learning model according to the first training data to determine a hyperparameter combination and construct an initial dynamic basic calibration model; The initial dynamic basic calibration model is deeply trained based on the second training data to construct a dynamic basic calibration model.

5. A method for dynamic calibration of an environmental monitoring instrument as claimed in claim 4, characterized in that: The acquisition function is as follows: EI(x)=(μ(x)-f(x best ))*Φ(Z)+σ(x)*ф(Z); Among them, EI(x) represents the expected improvement, that is, at the hyperparameter combination x, relative to the current optimal value f(x best ) is the expected improvement in the objective function value; x represents the current hyperparameter combination to be evaluated; μ(x) is the mean value of the objective function predicted by the Gaussian process model at the hyperparameter combination x; f(x best ) is the optimal value of the current objective function; Φ(Z) is the cumulative distribution function (CDF) of the standard normal distribution; Z is a standardized variable used to measure the difference between the predicted value and the current optimal value. σ(x) is the square root of the variance of the Gaussian process model at the hyperparameter combination x; ф(Z) is the probability density function (PDF) of the standard normal distribution.

6. A method for dynamic calibration of an environmental monitoring instrument as claimed in claim 1, characterized in that: Also includes: Acquire an iteration cycle, collect the basic environment calibration data and water pollution calibration data based on the iteration cycle, and synchronize them to a calibration database; Based on the calibration database and in combination with the historical monitoring data updating mechanism, the dynamic basic calibration model is adaptively optimized.

7. A dynamic calibration system for environmental monitoring instruments, characterized in that: The steps for implementing the method for dynamic calibration of an environmental monitoring instrument as claimed in any one of claims 1 to 6 include: Database construction module: determine the target industrial zone, retrieve and construct an environmental pollution database based on the target industrial zone, and the environmental pollution database includes an environmental basic attribute database and a water quality attribute database; Standard environment configuration module: obtaining M standard environment attributes based on the basic environment attribute library and the water quality attribute library, and configuring a standard environment based on the M standard environment attributes; Water quality environment monitoring module: Under the standard environment, water quality environment monitoring is performed through standard environment monitoring instruments to obtain basic environment calibration data and water quality pollution calibration data; Real-time water quality environment monitoring module: Real-time water quality environment monitoring is carried out through target environment monitoring instruments to obtain target basic environment data and target water quality pollution data; Calibration model construction module: introduces Bayesian optimization mechanism, constructs a dynamic basic calibration model based on the basic environmental calibration data, and constructs a dynamic pollution calibration model based on the water quality pollution calibration data; Calibration coefficient acquisition module: input the target basic environmental data into the dynamic basic calibration model to obtain the target basic calibration coefficient, input the target water quality pollution data into the dynamic pollution calibration model to obtain the target pollution calibration coefficient; Real-time calibration module: based on the target basic calibration coefficient and the target pollution calibration coefficient, the target basic environmental data and the target water quality pollution data are calibrated in real time to obtain calibrated environmental data and calibrated water quality pollution data.

Images