Water treatment data monitoring and evaluating method based on artificial intelligence

Abstract

The invention discloses a water treatment data monitoring and evaluation method based on artificial intelligence, and relates to the technical field of water treatment data monitoring and evaluation, and the method comprises the steps: S1, building a water treatment data multistage classification model based on artificial intelligence through water treatment data; s2, based on the water treatment data, utilizing the water treatment data multi-stage classification model to obtain a monitoring result of the water treatment data; and S3, performing optimization scheduling according to the monitoring result of the water treatment data, and obtaining a water treatment data monitoring evaluation result. According to the method, accurate and efficient output of the water treatment data is improved, large-range multi-source measurement data is intelligently processed, the environmental adaptability is high, meanwhile, systematic classification of the water treatment data is achieved through the water treatment data multi-stage classification model, and the classification efficiency is remarkably improved compared with a traditional single-stage model.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of water treatment data monitoring and evaluation, and particularly relates to a water treatment data monitoring and evaluation method based on artificial intelligence. BACKGROUND

[0002] With the in-depth application of big data technology in the field of water treatment, a monitoring system covering the whole chain of water source collection, treatment process, pipe network transportation has been built at present, and the data types cover multi-source heterogeneous data such as water quality sensors, equipment operation logs, and environmental monitoring terminals. However, in actual application, the wide variety of data sources leads to significant differences in sampling frequency and accuracy standards of different equipment, such as dimensional conflicts between online monitoring instruments and laboratory test data; environmental interference factors include seasonal water quality fluctuations, transmission errors caused by pipe network aging, and human operation errors. Although the existing technology can realize the preliminary integration of data, in key scenes such as medical water and industrial ultrapure water, due to the lack of a hierarchical verification mechanism for data credibility, problems such as misjudgment of abnormal data and missing detection of key indicators often occur, leading to lagging or overreaction of water treatment process regulation.

[0003] Therefore, there is an urgent need for a water treatment data monitoring and evaluation method based on artificial intelligence to solve the technical pain points of high credibility output of data in complex scenarios. SUMMARY

[0004] The purpose of the present application is to provide a water treatment data monitoring and evaluation method based on artificial intelligence, which establishes a multi-level classification model of water treatment data, verifies and analyzes step by step, and finally outputs water treatment data and corresponding processing results.

[0005] To achieve the above purpose, the present application provides a water treatment data monitoring and evaluation method based on artificial intelligence, comprising the following steps:

[0006] S1, using water treatment data based on artificial intelligence, establishing a multi-level classification model of water treatment data;

[0007] S2, based on the water treatment data, using the multi-level classification model of water treatment data, obtaining the monitoring result of water treatment data;

[0008] S3, optimizing scheduling according to the monitoring result of the water treatment data, obtaining the monitoring and evaluation result of the water treatment data.

[0009] Optionally, using water treatment data based on artificial intelligence, establishing a multi-level classification model of water treatment data, comprising:

[0010] Using real-time collected water treatment data, obtaining environmental data of water treatment data;

[0011] The environmental data based on the water treatment data is associated screened by using the water treatment data to obtain an associated screening result of historical water treatment data.

[0012] According to the associated screening result of the historical water treatment data, a multi-level classification model of water treatment data is established based on artificial intelligence.

[0013] Optionally, the associated screening result of the historical water treatment data is obtained by using the environmental data based on the water treatment data to perform associated screening on the historical water treatment data, and the associated screening result of the historical water treatment data comprises:

[0014] According to the water treatment data, a corresponding historical water treatment data set is obtained.

[0015] The environmental data of the historical water treatment data set is associated screened by using the environmental data of the water treatment data to obtain the environmental data of the corresponding historical water treatment data set, and the environmental data of the historical water treatment data set comprises environmental data of a plurality of historical water treatment data subsets.

[0016] Whether the environmental data of the historical water treatment data subset is consistent with the environmental data of the water treatment data is sequentially judged, if yes, a core water treatment data set is constructed by using the corresponding consistent historical water treatment data subset, otherwise, an auxiliary water treatment data set is constructed by using the corresponding inconsistent historical water treatment data subset.

[0017] The core water treatment data set and the auxiliary water treatment data set are obtained as the associated screening result of the historical water treatment data.

[0018] Optionally, according to the associated screening result of the historical water treatment data, a multi-level classification model of water treatment data is established based on artificial intelligence, and the method comprises the following steps:

[0019] S1-3-1, the associated screening result of the historical water treatment data is divided and processed to obtain a division result of a core water treatment data set and a division result of an auxiliary water treatment data set, respectively, wherein the division result of the core water treatment data set comprises a core water treatment data training set and a core water treatment data verification set, and the division result of the auxiliary water treatment data set comprises an auxiliary water treatment data training set and an auxiliary water treatment data verification set.

[0020] S1-3-2, the environmental data set of the corresponding core water treatment data set is obtained by using the division result of the core water treatment data set, wherein the environmental data set of the core water treatment data set comprises a core environmental data training set and a core environmental data verification set.

[0021] S1-3-3, input the core water treatment data training set and the core environment data training set as inputs, and normal state as output, establish the number of hidden layers corresponding to the environment data category based on the core environment data training set, and train an initial core water treatment data classification model based on a convolutional neural network;

[0022] S1-3-4, input the core water treatment data validation set and the core environment data validation set into the initial core water treatment data classification model, and obtain the output result of the initial core water treatment data classification model;

[0023] S1-3-5, determine whether the output result of the initial core water treatment data classification model is normal, if yes, obtain the initial core water treatment data classification model as the core water treatment data classification model, otherwise, obtain the core water treatment data validation set and the core environment data validation set which are not normal, add the core water treatment data validation set and the core environment data validation set, and return to S1-3-3;

[0024] S1-3-6, obtain the environment data set corresponding to the auxiliary water treatment data set by using the division result of the auxiliary water treatment data set, wherein the environment data set of the auxiliary water treatment data set includes the core environment data training set and the auxiliary environment data validation set;

[0025] S1-3-7, input the auxiliary water treatment data training set and the auxiliary environment data training set as inputs, and abnormal state as output, establish the number of hidden layers corresponding to the environment data category based on the auxiliary environment data training set, and train an initial auxiliary water treatment data classification model based on a convolutional neural network;

[0026] S1-3-8, input the auxiliary water treatment data validation set and the auxiliary environment data validation set into the initial auxiliary water treatment data classification model, and obtain the output result of the initial auxiliary water treatment data classification model;

[0027] S1-3-9, determine whether the output result of the initial auxiliary water treatment data classification model is abnormal, if yes, obtain the initial auxiliary water treatment data classification model as the auxiliary water treatment data classification model, otherwise, obtain the auxiliary water treatment data validation set and the auxiliary environment data validation set which are not normal, add the auxiliary water treatment data validation set and the auxiliary environment data validation set, and return to S1-3-7;

[0028] S1-3-10, fuse the core water treatment data classification model and the auxiliary water treatment data classification model to construct a water treatment data multi-level classification model.

[0029] Optionally, based on the water treatment data, the water treatment data multi-level classification model is used to obtain a monitoring result of the water treatment data, including:

[0030] The water treatment data is input into the water treatment data multi-level classification model to obtain a multi-level classification result of the water treatment data.

[0031] The multi-level classification result of the water treatment data is used to monitor the water treatment data and the water treatment data multi-level classification model to obtain a monitoring result of the water treatment data.

[0032] Optionally, the multi-level classification result of the water treatment data is used to monitor the water treatment data and the water treatment data multi-level classification model to obtain a monitoring result of the water treatment data, including:

[0033] S2-2-1, obtaining the environment data of the water treatment data and the training times of the water treatment data multi-level classification model;

[0034] S2-2-2, using the multi-level classification result of the water treatment data and the environment data of the water treatment data for comparison processing to obtain a comparison result of the water treatment data;

[0035] S2-2-3, using the environment data of the water treatment data to obtain a data type quantity corresponding to the environment data;

[0036] S2-2-4, determining whether the training times of the water treatment data multi-level classification model are greater than the data type quantity of the environment data, if yes, obtaining overfitting of the water treatment data classification model as a detection result of the water treatment data classification model, and directly executing S2-2-7, otherwise, executing S2-2-5;

[0037] S2-2-5, using the multi-level classification result of the water treatment data to obtain a subset quantity of the core water treatment data training set and a subset quantity of the auxiliary water treatment data training set;

[0038] S2-2-6, determining whether a relative ratio of the subset quantity of the core water treatment data training set and the subset quantity of the auxiliary water treatment data training set is greater than a target ratio, if yes, obtaining overfitting of the water treatment data classification model as a detection result of the water treatment data classification model, and directly executing S2-2-7, otherwise, obtaining no overfitting of the water treatment data classification model as a detection result of the water treatment data classification model, and directly executing S2-2-7;

[0039] S2-2-7, obtaining the comparison result of the water treatment data and the detection result of the water treatment data classification model as a monitoring result of the water treatment data.

[0040] Optionally, the multi-level classification result of the water treatment data is compared with the environmental data of the water treatment data to obtain a comparison result of the water treatment data, including:

[0041] S2-2-2-1, obtaining the environmental data of the water treatment data as actual environmental data;

[0042] S2-2-2-2, based on the actual environmental data, using the historical water treatment data set and the environmental data of the historical water treatment data set to obtain a historical water treatment data set with the same actual environmental data, wherein the historical water treatment data set with the same actual environmental data includes several historical water treatment data subsets with the same real-time environmental data;

[0043] S2-2-2-3, judging whether the states corresponding to the several historical water treatment data subsets with the same real-time environmental data are all consistent, if yes, obtaining the states corresponding to the several historical water treatment data subsets with the same real-time environmental data as state information, and executing S2-2-2-4, otherwise, obtaining the states corresponding to the historical water treatment data subsets with the same real-time environmental data with a relatively high proportion as state information, and executing S2-2-2-4, wherein the state information includes a normal state or an abnormal state of the water treatment data;

[0044] S2-2-2-4, judging whether the state information is consistent with the multi-level classification result of the water treatment data, if yes, obtaining the multi-level classification result of the water treatment data as the comparison result of the water treatment data, otherwise, adding the state information corresponding water treatment data and the environmental data of the water treatment data into a corresponding training set, and returning to S1-3-3.

[0045] Optionally, the water treatment data monitoring result is optimized to obtain a water treatment data monitoring evaluation result, including:

[0046] S3-1, judging whether the comparison result of the water treatment data is a normal state, if yes, retaining the comparison result of the current water treatment data, and obtaining a scheduling result of the water treatment data as empty, directly executing S3-4, otherwise, executing S3-2;

[0047] S3-2, judging whether the water treatment data corresponding to the comparison result of the water treatment data and the environmental data of the water treatment data exist historical water treatment data or historical water treatment data environmental data, if yes, obtaining the current water treatment data and the environmental data of the current water treatment data, and obtaining a scheduling result of the water treatment data as failure, otherwise, returning to using the real-time collected water treatment data to obtain the environmental data of the water treatment data;

[0048] S3-3, performing optimization processing on the detection result of the water treatment data classification model to obtain an optimization result of the water treatment data;

[0049] S3-4, obtaining a water treatment data monitoring evaluation result by using the scheduling result of the water treatment data and the optimization result of the water treatment data.

[0050] Optionally, the optimization result of the water treatment data is obtained by performing optimization processing on the detection result of the water treatment data classification model, and the optimization result of the water treatment data includes:

[0051] S3-3-1, determining whether the detection result of the water treatment data classification model is over-fitted, if yes, performing S3-3-2, otherwise, retaining the detection result of the current water treatment data classification model, and the optimization result of the water treatment data is empty;

[0052] S3-3-2, determining whether the number of subsets of the core water treatment data training set corresponding to the multi-level classification result of the water treatment data is greater than the number of subsets of the auxiliary water treatment data training set corresponding to the multi-level classification result of the water treatment data, if yes, obtaining the auxiliary water treatment data training set corresponding to the multi-level classification result of the water treatment data as a to-be-optimized data set, and performing S3-3-3, otherwise, obtaining the core water treatment data training set corresponding to the multi-level classification result of the water treatment data as the to-be-optimized data set, and performing S3-3-3;

[0053] S3-3-3, obtaining the core water treatment data training set and the auxiliary water treatment data training set with the same subset number as the updated core water treatment data training set and the auxiliary water treatment data training set by using the to-be-optimized data set, and returning to establishing the water treatment data multi-level classification model according to the associated screening result of the historical water treatment data.

[0054] Compared with the closest prior art, the present application has the beneficial effects:

[0055] The present application mines the hierarchical relationship and correlation characteristics between data by associating and screening the water treatment data, constructs a water treatment data multi-level classification model with layer-by-layer classification ability, realizes systematic classification of water treatment data, and significantly improves the classification efficiency compared with traditional single-level models, lays a model foundation for subsequent data monitoring and analysis. In addition, the data anomaly and model reliability are monitored to form a water treatment data monitoring result, which provides a key basis for fault early warning and ensures the safe operation of the water treatment system. At the same time, the model parameters are dynamically optimized to ensure that the model maintains stable classification accuracy for a long time and improves the cross-scene classification adaptability. A "model construction-data monitoring-optimization scheduling" closed-loop system is constructed to improve the automation and intelligent level of the whole process, reduce the system operation risk and cost.

[0056] The application improves the accurate and efficient output of water treatment data, and through intelligent processing of a large range of multi-source measurement data, the environmental adaptability is relatively strong, and the logical optimization within the implementation scheme and the output of data abnormalities outside the scheme are synchronized, in the case of abnormal state and the like, while ensuring the stable operation of the optimization scheduling, the scheme implementation scene is quickly self-adjusted to adapt to the scheme implementation scene. BRIEF DESCRIPTION OF DRAWINGS

[0057] In order to more clearly illustrate the technical solutions in the specific embodiments or prior art of the present application, the drawings needed in the specific embodiments or prior art description will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present application, and those skilled in the art can also obtain other drawings according to these drawings without creative labor.

[0058] Figure 1 A flowchart of a water treatment data monitoring and evaluation method based on artificial intelligence according to an embodiment of the present application. DETAILED DESCRIPTION

[0059] In order to make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below in combination with the specific embodiments of the present application and the corresponding drawings. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0060] The terms used in the embodiment part of the present application are only used to explain the specific embodiments of the present application, and are not intended to limit the present application.

[0061] As shown in Figure 1 The embodiment of the present application provides a water treatment data monitoring and evaluation method based on artificial intelligence, which comprises:

[0062] S1, using water treatment data based on artificial intelligence, a multi-level classification model of water treatment data is established;

[0063] The embodiment uses water treatment data and corresponding environmental data to carry out correlation screening, and based on artificial intelligence, a multi-level classification model of water treatment data is constructed. This model can realize hierarchical processing from basic data cleaning to abnormal type identification, solve the feature confusion problem caused by multi-source data interference, and lay a model foundation for subsequent accurate classification. In addition, through correlation screening and the establishment of the multi-level classification model, the embodiment can effectively deal with the problem of wide water treatment data sources and many environmental interference factors.

[0064] S2, acquiring a monitoring result of the water treatment data based on the water treatment data and using the multi-level classification model of the water treatment data;

[0065] In this embodiment, the water treatment data is input into the constructed multi-level classification model to obtain a multi-level classification result containing data category attribution, hierarchical relationship and other information, and then the water treatment data itself and the model are monitored based on the multi-level classification result, forming a complete monitoring system and realizing efficient and accurate monitoring of the water treatment data, thereby providing strong support for optimization and management of the water treatment system.

[0066] S3, performing optimization scheduling according to the monitoring result of the water treatment data to obtain a monitoring evaluation result of the water treatment data.

[0067] In this embodiment, a complete monitoring evaluation process of the water treatment data is formed through judgment of the comparison result of the water treatment data, checking of the data corresponding relationship, data optimization processing and comprehensive evaluation, normal and abnormal data can be identified in time, the causes of the abnormal data are analyzed in depth and optimized, and finally an accurate monitoring evaluation is given by comprehensively considering the results of multiple aspects, thereby effectively improving the efficiency and accuracy of the water treatment data management and providing strong support for stable operation and optimization of the water treatment system.

[0068] As a possible implementation, in the above embodiment, step S1 can specifically include the following steps:

[0069] S1-1, acquiring environmental data of the water treatment data by using real-time collected water treatment data; the environmental parameters synchronized with the water treatment data are collected in real time by deploying a sensor network (such as a temperature and humidity sensor, a rainfall monitor, a pipe network pressure transmitter, etc.) in the water treatment field to form environmental data containing a time stamp, a geographical position and environmental indexes (such as temperature 25℃±2℃, humidity 60%±5%). The water treatment data refers to various monitoring and operation data generated in the whole process of water treatment, which has a wide range of coverage and a multi-source characteristic. This step realizes spatio-temporal alignment of the environmental data and the water treatment data, provides an environmental background dimension for subsequent correlation analysis, and solves the problem that environmental interference factors are difficult to quantify in traditional data processing.

[0070] S1-2, performing correlation screening of the historical water treatment data based on the environmental data of the water treatment data and using the water treatment data to obtain a correlation screening result of the historical water treatment data; in this embodiment, the environmental data of the historical water treatment data is extracted through the environmental data of the water treatment data, the environmental data is compared, and then the core and auxiliary data sets are constructed according to consistency to realize correlation screening of the historical data, which not only ensures high matching degree of data correlation characteristics under the current environment, but also covers correlation modes across environments, thereby providing correlation characteristic samples with pertinence and generalization for the model.

[0071] S1-3, the multi-level classification model of water treatment data is established based on artificial intelligence according to the association screening result of the historical water treatment data;

[0072] The embodiment utilizes the association screening result of the historical water treatment data to construct the multi-level classification model of water treatment data, solves the problem that environmental interference factors are difficult to be quantified in traditional data processing, can effectively filter misjudgment data caused by environmental interference, provides reliable support for subsequent monitoring and evaluation of water treatment data, and realizes effective connection from data processing to actual application.

[0073] As a possible implementation, in the above embodiment, step S1-2 can specifically include the following steps:

[0074] S1-2-1, according to the water treatment data, a corresponding historical water treatment data set is obtained;

[0075] The water quality indicators (such as turbidity, pH value) in the current water treatment data, the treatment process parameters (such as the amount of dosing, the filtration flow rate) and the time stamp are taken as the retrieval keywords, and precise matching is performed in the historical database. For example, when the real-time data shows that the turbidity of a certain water source is 3.2 NTU and the coagulation-sedimentation process is used, the system will automatically screen out the historical data of the same water source, the same process condition and the turbidity in the interval of 2.5-4.0 NTU in the past three years, form a candidate data set containing equipment operation log and water quality monitoring record, and provide data basis for subsequent environmental dimension screening.

[0076] S1-2-2, the environmental data of the historical water treatment data set is associated and screened by using the environmental data of the water treatment data, and the environmental data of the corresponding historical water treatment data set is obtained, wherein the environmental data of the historical water treatment data set includes the environmental data of a plurality of historical water treatment data subsets;

[0077] By extracting the environmental data in the water treatment data, the environmental data is taken as a screening condition and is matched and associated with the historical water treatment data set. Specifically, the part consistent with the current environmental data characteristics is identified from the historical data set, the historical data set is divided into a plurality of historical water treatment data subsets with similar environmental data, each subset corresponds to a specific environmental scene, so that the environmental data of the historical water treatment data set is associated and screened. The embodiment realizes the association screening of the environmental data of the historical water treatment data set by the environmental data, so that the historical data is accurately corresponding to the environmental characteristics of the current monitoring scene, avoids the data reference deviation caused by environmental difference, and provides data support with better scene adaptability for subsequent classification model training.

[0078] S1-2-3, judging whether the environment data of the historical water treatment data subset and the environment data of the water treatment data are consistent in sequence, if yes, constructing a core water treatment data set by using the corresponding consistent historical water treatment data subset, otherwise, constructing an auxiliary water treatment data set by using the corresponding inconsistent historical water treatment data subset;

[0079] In the embodiment, by comparing the consistency of the environment data of each subset and the real-time environment data, the subsets with high matching degree are classified into the core data set, the accuracy of model training under the current environment is ensured, the unmatched subsets are classified into the auxiliary data set, the generalization ability of the model across the environment is enhanced, and the accuracy of abnormal identification is improved.

[0080] S1-2-4, obtaining the core water treatment data set and the auxiliary water treatment data set as the association screening result of the historical water treatment data;

[0081] The screened core data set and auxiliary data set are output in a standardized format, so that subsequent model construction is based on more targeted data, avoiding the influence of unrelated data on the model effect, and at the same time, the changes in the environment in the actual water treatment process can be better coped with, and data classification and monitoring evaluation can be more accurately performed under different environmental scenarios.

[0082] As a possible implementation, in the above embodiment, step S1-3 can specifically include the following steps:

[0083] S1-3-1, performing division processing on the association screening result of the historical water treatment data, and respectively obtaining a division result of a core water treatment data set and a division result of an auxiliary water treatment data set, wherein the division result of the core water treatment data set includes a core water treatment data training set and a core water treatment data verification set, and the division result of the auxiliary water treatment data set includes an auxiliary water treatment data training set and an auxiliary water treatment data verification set;

[0084] The core water treatment data set is randomly divided into a core water treatment data training set and a core water treatment data verification set in a ratio of 7:3, the core water treatment data training set is used for fine training of model basic parameters, and focuses on learning data features in high matching degree environment scenarios; the core water treatment data verification set is used for verifying the classification accuracy of the model in the target scenario, and ensuring the identification accuracy of the model. The auxiliary water treatment data set is divided into an auxiliary water treatment data training set and an auxiliary water treatment data verification set in a ratio of 6:4, the auxiliary water treatment data training set is used for expanding the generalization ability of the model to the related environment scenarios, and learning data patterns under different environmental fluctuations; the auxiliary water treatment data verification set is used for verifying the adaptability of the model in the non-core scenario, and avoiding the overfitting problem caused by single data distribution.

[0085] The embodiment divides the association screening result of the historical water treatment data, facilitates the core water treatment data classification model to learn the mapping relationship of "data characteristics-environment scene" under normal working conditions, assists the water treatment data classification model to capture the feature mode of the abnormal state, forms a double-track cognitive system of "normal mode memory-abnormal mode recognition", and improves the filtering efficiency of complex environmental interference.

[0086] S1-3-2, obtaining an environment data set corresponding to the core water treatment data set by using the division result of the core water treatment data set, wherein the environment data set of the core water treatment data set includes a core environment data training set and a core environment data verification set;

[0087] From the division result of the core water treatment data set, the corresponding environment data is extracted to form an independent core environment data set. The specific division manner is: the environment data in the core water treatment data training set is extracted as the core environment data training set, which is used for model learning of the environment feature mode under normal working conditions; the environment data in the core water treatment data verification set is extracted as the core environment data verification set, which is used for verifying the classification accuracy of the model under the target environment scene. The embodiment independently extracts the environment data training set and the environment data verification set which strictly match the core water treatment data, facilitates the accurate modeling of the environment features in the water treatment process, and improves the abnormal recognition accuracy and training efficiency of the model under complex environment.

[0088] S1-3-3, taking the core water treatment data training set and the core environment data training set as inputs, taking the normal state as output, establishing the number of hidden layers corresponding to the environment data types in the core environment data training set, and training an initial core water treatment data classification model based on a convolutional neural network;

[0089] When the initial core water treatment data classification model is constructed based on the convolutional neural network, the core water treatment data training set and the core environment data training set are taken as inputs, and the normal state is taken as the output target. The number of hidden layers is dynamically determined according to the number of environment data types in the core environment data training set. The data space-time features are automatically extracted through the convolutional layer, and the environment features are combined for joint training, so as to realize accurate identification of the normal state data. The embodiment has the advantages of automatic extraction of space-time features, environment adaptive network structure, accurate modeling of normal state, strong anti-noise interference, and efficient use of computing resources.

[0090] In addition, the number of hidden layers is dynamically adjusted according to the environment data types, so that the model can adapt to the environmental complexity of different monitoring scenes. In the cross-season water source monitoring scene, the classification accuracy of the model is higher than that of the fixed structure model, especially in the extreme environment such as rainstorm and cold wave.

[0091] S1-3-4, input the core water treatment data verification set and the core environment data verification set into the initial core water treatment data classification model, obtain an output result of the initial core water treatment data classification model, and

[0092] The core water treatment data verification set and the core environment data verification set are synchronously input into the initial core water treatment data classification model, and a forward calculation is performed by means of a hierarchical classification algorithm such as a multilayer neural network, and an output result is output to verify the performance of the model. This double-dimension verification method can accurately identify the classification ability of the model under the condition of “data features + environment scenarios”, facilitate subsequent comparison of output and actual label positioning of defects of each level of model, and quantize the influence weight of environmental factors to optimize the priority of features, so that the classification accuracy of the model in the core scenario such as drinking water treatment is improved.

[0093] S1-3-5, determining whether the output result of the initial core water treatment data classification model is normal, if yes, obtaining the initial core water treatment data classification model as the core water treatment data classification model, otherwise, obtaining the core water treatment data verification set and the core environment data verification set which are not normal, adding the core water treatment data verification set and the core environment data verification set, and returning to S1-3-3;

[0094] The output result of the initial core water treatment data classification model is checked sample by sample, if the verification set samples are correctly identified as normal, the model is determined as the final model; if there is abnormal classification, the non-normal state core water treatment data and environment data verification set samples are extracted and supplemented to the database, and the model optimization process is returned to retrain parameters. This method enhances the robustness of the model in identifying abnormal data through dynamic cycle optimization, guarantees zero misjudgment in the core scenario, automatically expands the abnormal sample library, reduces the cost of manual parameter adjustment, and realizes efficient iteration and accurate monitoring of the model in the key scenario.

[0095] S1-3-6, using the division result of the auxiliary water treatment data set, obtaining an environment data set corresponding to the auxiliary water treatment data set, wherein the environment data set of the auxiliary water treatment data set includes a core environment data training set and an auxiliary environment data verification set;

[0096] From the division results of the auxiliary water treatment data set, the corresponding environmental data is extracted to construct a core environmental data training set and an auxiliary environmental data verification set respectively, wherein the core environmental data training set is used to learn the normal data characteristics in the associated environmental scene, and the auxiliary environmental data verification set is used to verify the adaptability of the model when the non-core environmental fluctuates, forming an environmental data closed loop of "training-verification". In this embodiment, by extracting the environmental data of the auxiliary data set to construct special training and verification sets, the generalization ability of the model to the associated environmental scene can be enhanced. For example, when the water source is affected by seasonal pollution, the model can learn the data pattern under similar pollution degree through the core environmental training set, and then optimize the parameters with the help of the auxiliary environmental verification set, so as to improve the abnormal recognition accuracy of the non-core scene and reduce the misjudgment rate caused by environmental parameter fluctuation, thereby providing more reliable support for water treatment data monitoring in complex environment.

[0097] S1-3-7, taking the auxiliary water treatment data training set and the auxiliary environmental data training set as inputs, and taking the abnormal state as output, the auxiliary environmental data training set corresponds to the number of hidden layers of environmental data categories, and an initial auxiliary water treatment data classification model is trained based on a convolutional neural network;

[0098] Taking the auxiliary water treatment data training set and its corresponding auxiliary environmental data training set as inputs, and taking the "abnormal state" as the output target, the number of hidden layers of the convolutional neural network (CNN) is determined according to the number of environmental data categories in the auxiliary environmental data training set, the spatial correlation pattern of data characteristics and environmental parameters is extracted through the convolution kernel, and the model is iteratively trained until the abnormal recognition accuracy of the model on the auxiliary environmental data verification set meets the standard, thereby constructing an initial auxiliary water treatment data classification model.

[0099] In this embodiment, the CNN model is constructed by combining auxiliary data and environmental characteristics, which can strengthen the model's ability to recognize abnormal data in the associated environmental scene. For example, when the water source is suddenly polluted due to surrounding construction, the model can capture the spatial feature correlation between turbidity and particle size distribution through the convolution kernel, the abnormal recognition speed is improved compared with traditional neural networks, and the detection rate of low concentration anomalies is improved. At the same time, based on the mechanism of dynamically adjusting the number of hidden layers according to the type of environmental data, the model can ensure the classification accuracy without reparameterization when crossing the environmental scene, thereby reducing the engineering deployment cost.

[0100] S1-3-8, inputting the auxiliary water treatment data verification set and the auxiliary environmental data verification set into the initial auxiliary water treatment data classification model, and obtaining the output result of the initial auxiliary water treatment data classification model,

[0101] The auxiliary water treatment data verification set and the auxiliary environment data verification set are input into the constructed initial auxiliary water treatment data classification model, the output probability of the abnormal state of each sample is calculated through forward propagation of the model, the classification result of the model in the associated environment scene is obtained, and the adaptability and accuracy of the model to the non-core scene data are evaluated. Through the independent auxiliary verification set test, the generalization ability of the model in the environment fluctuation scene can be quantified, and the overfitting problem of the core model to the associated scene is avoided.

[0102] S1-3-9, determining whether the output result of the initial auxiliary water treatment data classification model is abnormal state, if yes, obtaining the initial auxiliary water treatment data classification model as the auxiliary water treatment data classification model, otherwise, obtaining the auxiliary water treatment data verification set and the auxiliary environment data verification set which are not in the normal state, adding the auxiliary water treatment data verification set and the auxiliary environment data verification set, and returning to S1-3-7;

[0103] In this embodiment, the output result of the initial auxiliary water treatment data classification model is checked sample by sample. If all the auxiliary verification set samples are correctly recognized as abnormal state, the initial model is determined as the final auxiliary water treatment data classification model. If there is a normal state classification result, the auxiliary water treatment data verification set and the corresponding auxiliary environment data verification set sample which are judged as non-normal state are extracted, the original verification set database is re-added, and the model training link is returned to optimize the parameters. This method forces the model to calibrate the abnormal recognition boundary in the associated scene, dynamically supplements samples to enhance the generalization ability, and constructs an automatic iteration link, so that the cross-scene misjudgment rate of the model is significantly reduced, the generalization ability is improved, the adaptation period is shortened, the self-adaptive optimization to the environment fluctuation is realized, and the labor cost and emergency response time are greatly reduced.

[0104] S1-3-10, fusing the core water treatment data classification model and the auxiliary water treatment data classification model to construct a multi-level water treatment data classification model;

[0105] The core water treatment data classification model and the auxiliary water treatment data classification model are fused, a weighted decision fusion strategy is adopted, the output results of the core model and the auxiliary model in different environment scenes are given corresponding weights, and a multi-level water treatment data classification model which can comprehensively process core scene and associated scene data is constructed.

[0106] The multi-level classification model constructed by fusing the core model and the auxiliary model can realize accurate classification of water treatment data in different environmental scenarios. On the one hand, the classification accuracy and generalization ability of the model as a whole can be improved by making full use of the accurate classification ability of the core model for core scene data and the adaptability of the auxiliary model for associated scene data. On the other hand, when facing complex and variable water treatment environments, the model can be more flexible to cope with and accurately identify data characteristics in different scenarios, providing more reliable support for monitoring and evaluation of water treatment data. For example, in a multi-water source switching scenario, the correct classification rate of the model for data is improved by more than 30% compared with a single model, and the misjudgment rate of cross-scene data can be effectively reduced.

[0107] As a possible implementation, in the above embodiment, step S2 can specifically include the following steps:

[0108] S2-1, inputting the water treatment data into the water treatment data multi-level classification model to obtain a multi-level classification result of the water treatment data;

[0109] The water treatment data is input into the constructed multi-level classification model, and the model analyzes the data layer by layer according to the preset hierarchical classification rule, thereby outputting a multi-level classification result containing information such as data category attribution and hierarchical relationship. This method can realize systematic classification of water treatment data and provide a basis for subsequent fine processing, and the classification efficiency is improved compared with traditional single-level models.

[0110] S2-2, monitoring the water treatment data and the water treatment data multi-level classification model by using the multi-level classification result of the water treatment data to obtain a monitoring result of the water treatment data;

[0111] By monitoring the water treatment data by using the multi-level classification result of the water treatment data, the embodiment can accurately identify data anomalies or changes, and can timely discover water quality parameter fluctuations, equipment operation abnormalities and the like, thereby providing a key basis for fault early warning. Meanwhile, monitoring the water treatment data multi-level classification model by using the multi-level classification result of the water treatment data helps to ensure the reliability of the model, dynamically optimizes the model parameters, and enables the model to maintain stable classification accuracy in long-term use, thereby improving the cross-scenario classification accuracy of the model.

[0112] As a possible implementation, in the above embodiment, step S2-2 can specifically include the following steps:

[0113] S2-2-1, obtaining environmental data of the water treatment data and a training number of the water treatment data multi-level classification model;

[0114] The embodiment obtains the training times of the multi-level classification model of the water treatment data and the environment data of the water treatment data in the above steps, to provide basic data for subsequent overfitting detection and judgment. The training times are a key indicator for evaluating whether the model is overfitting, and the environment data is used to compare the rationality of the classification results.

[0115] S2-2-2, comparing the multi-level classification results of the water treatment data with the environment data of the water treatment data, to obtain comparison results of the water treatment data;

[0116] The embodiment can effectively improve the accuracy of the classification results and the adaptability of the model to environmental changes by comparing the multi-level classification results of the water treatment data with the environment data of the water treatment data.

[0117] S2-2-3, using the environment data of the water treatment data, to obtain the number of data types corresponding to the environment data;

[0118] The number of different feature types contained in the environment data reflects the dimension complexity of the data, and provides a reference for judging whether the training times are reasonable. For example, if the training times are much more than the feature dimension, it may lead to model fitting noise.

[0119] S2-2-4, judging whether the training times of the multi-level classification model of the water treatment data are greater than the number of data types of the environment data. If yes, obtaining that the water treatment data classification model has overfitting as the detection result of the water treatment data classification model, and directly executing S2-2-7, otherwise, executing S2-2-5;

[0120] The embodiment is based on the principle that the training times should not be significantly more than the feature dimension, to quickly identify the risk of overfitting and avoid the model being too complex on small sample high-dimensional data.

[0121] S2-2-5, using the multi-level classification results of the water treatment data, to obtain the number of subsets of the core water treatment data training set and the number of subsets of the auxiliary water treatment data training set;

[0122] The core water treatment data training set and the auxiliary water treatment data training set are split from the multi-level classification results, and the number of subsets of the two is counted, to facilitate the analysis of the subset distribution of different data sets and the evaluation of whether the model excessively relies on core data and ignores auxiliary information, to provide multi-dimensional evidence for overfitting judgment.

[0123] S2-2-6, determining whether the relative ratio of the number of subsets of the core water treatment data training set and the number of subsets of the auxiliary water treatment data training set is greater than a target ratio, if yes, obtaining overfitting of the water treatment data classification model as the detection result of the water treatment data classification model, and directly executing S2-2-7, otherwise, obtaining no overfitting of the water treatment data classification model as the detection result of the water treatment data classification model, and directly executing S2-2-7;

[0124] The ratio of the number of subsets of the core water treatment data training set and the number of subsets of the auxiliary water treatment data training set is calculated. If the ratio is greater than the target ratio (preset threshold), it is determined that the model is overfitting. Otherwise, it is determined that the model is not overfitting. In practical applications, the ratio of the number of subsets of the core water treatment data training set and the number of subsets of the auxiliary water treatment data training set or the ratio of the number of subsets of the auxiliary water treatment data training set and the number of subsets of the core water treatment data training set is greater than 1.5. Therefore, the target ratio is set to 1.5. This embodiment evaluates the rationality of model training through the balance of data subsets, avoiding the situation that the model ignores the overall distribution rule due to the high proportion of core data.

[0125] S2-2-7, obtaining the comparison result of the water treatment data and the detection result of the water treatment data classification model as the monitoring result of the water treatment data;

[0126] The comparison result of the water treatment data and the detection result of the water treatment data classification model form a comprehensive water treatment data monitoring result covering data classification state, feature abnormality and model reliability. This integration method can provide a comprehensive monitoring perspective, provide multi-dimensional data support for subsequent evaluation and decision-making, and make the monitoring conclusion more scientific. For example, in the water quality early warning scene, the comprehensive monitoring result can improve the early warning response speed and reduce the false alarm rate.

[0127] As a possible implementation, in the above embodiment, step S2-2-2 can specifically include the following steps:

[0128] S2-2-2-1, obtaining the environment data of the water treatment data as actual environment data;

[0129] This embodiment extracts the environment data corresponding to the water treatment data as the actual environment data, provides a reference for subsequent historical data matching, and ensures that the comparison process is based on the same environmental conditions.

[0130] S2-2-2-2. Based on the actual environmental data, the historical water treatment data set and the environmental data of the historical water treatment data set are used to obtain a historical water treatment data set of the same actual environmental data, wherein the historical water treatment data set of the same actual environmental data includes several historical water treatment data subsets of the same real-time environmental data, specifically:

[0131] Based on the actual environmental data, historical data with the same environmental conditions are screened out from the historical water treatment data set to form a data set containing multiple subsets. This embodiment matches historical data with environmental conditions to construct a comparable reference set, providing data support for the verification of classification results.

[0132] S2-2-2-3. Determine whether the corresponding states of several historical water treatment data subsets of the same real-time environmental data are all consistent. If so, obtain the corresponding states of several historical water treatment data subsets of the same real-time environmental data as state information, and execute S2-2-2-4. Otherwise, obtain the corresponding states of a relatively high proportion of historical water treatment data subsets of the same real-time environmental data as state information, and execute S2-2-2-4, wherein the state information includes the normal state or abnormal state of the water treatment data;

[0133] Determine whether the status of the same environmental historical data subset is consistent. If consistent, take the full status as the status information, otherwise take the high-proportion status as the status information (divided into normal / abnormal status), wherein a relatively high proportion means that in the corresponding status of several historical water treatment data subsets of the same real-time environmental data, the number of normal states or abnormal states accounts for a significantly high proportion, becoming the dominant type in the group of data. For example, when there is inconsistency in the status of the historical water treatment data subset under the same environmental conditions (such as some subsets are "normal" and some are "abnormal"), by counting the frequency of occurrence of each state, the state with the highest number of occurrences is determined as the "relatively high proportion" state. This embodiment forms an objective comparison benchmark by counting the consistency or dominant type of the historical data status, thereby avoiding the influence of single data deviation on judgment.

[0134] S2-2-2-4. Determine whether the state information is consistent with the multi-level classification result of the water treatment data. If so, obtain the multi-level classification result of the water treatment data as the comparison result of the water treatment data. Otherwise, add the water treatment data corresponding to the state information and the environmental data of the water treatment data to the corresponding training set, and return to S1-3-3.

[0135] The state information is compared with the current multi-level classification result, and the classification result is confirmed if consistent, and the corresponding data is added to the training set and the related process is restarted if inconsistent, which realizes dynamic verification of the classification result and iteration optimization of the model, improves the classification accuracy through the regular feedback of the historical environmental data, and enhances the adaptability of the model to the environmental scene.

[0136] The embodiment constructs a closed-loop process of "data comparison-result verification-model update" through the historical data matching based on the environmental data, the state statistics and the classification verification mechanism, can verify the rationality of the current classification result by using the historical environmental data, can optimize the model through the real-time feedback of the abnormal data, and effectively improves the accuracy of the water treatment data classification and the environmental adaptability of the model.

[0137] As a possible implementation, in the above embodiment, step S3 can specifically include the following steps:

[0138] S3-1, judging whether the comparison result of the water treatment data is in a normal state, if yes, retaining the comparison result of the current water treatment data, and obtaining that the scheduling result of the water treatment data is empty, directly executing S3-4, otherwise, executing S3-2;

[0139] Judging whether the comparison result of the water treatment data is in a normal state, if yes, retaining the comparison result, then the scheduling result of the water treatment data is passed, and the scheduling result of the water treatment data corresponds to an empty output, directly entering S3-4, if not, executing S3-2. The embodiment quickly identifies the normal state data, reduces unnecessary processing process, and improves the data processing efficiency; for the abnormal state data, the subsequent processing process is triggered in time, and it is ensured that the problem data is paid attention to.

[0140] S3-2, judging whether the comparison result of the water treatment data corresponds to the water treatment data and the environmental data of the water treatment data exist any non-correspondence of the historical water treatment data or the environmental data of the historical water treatment data, if yes, obtaining the current water treatment data and the environmental data of the current water treatment data, and obtaining that the scheduling result of the water treatment data fails, otherwise, returning to S1-1;

[0141] determine whether the water treatment data and the environmental data thereof that are determined to be abnormal exist a non-corresponding condition with the historical data, if yes, the scheduling result of the water treatment data fails, and the current real-time water treatment data and the associated environmental data thereof are output; if not, the environmental data of the water treatment data is acquired by using the real-time collected water treatment data, and the corresponding historical environmental data is re-acquired by using the environmental data to associate and screen the historical data set. In this embodiment, the data correspondence is checked to identify whether the data abnormality is caused by the missing or mismatching of the historical data, and a basis is provided for subsequent processing. For the non-corresponding data, the scheduling failure is explicitly determined, and the data problem can be found in time. For the corresponding data, the historical data is associated and screened to provide support for subsequent optimization processing.

[0142] S3-3, performing optimization processing by using the detection result of the water treatment data classification model to acquire an optimization result of the water treatment data;

[0143] In this embodiment, the data is optimized based on the model detection result, the data quality and usability are improved, and more accurate data support is provided for subsequent monitoring and evaluation.

[0144] S3-4, acquiring a monitoring and evaluation result of the water treatment data by using the scheduling result of the water treatment data and the optimization result of the water treatment data;

[0145] The monitoring and evaluation result of the water treatment data is obtained by comprehensively using the scheduling result and the optimization result of the water treatment data, the water treatment data is comprehensively and objectively evaluated, and a scientific decision basis is provided for the operation and management of the water treatment system.

[0146] As a possible implementation, in the above embodiment, step S3-3 can specifically include the following steps:

[0147] S3-3-1, determining whether the detection result of the water treatment data classification model exists overfitting, if yes, performing S3-3-2, otherwise, the detection result of the current water treatment data classification model is retained, and the optimization result of the water treatment data is empty;

[0148] It is determined whether the detection result of the water treatment data classification model shows overfitting, if yes, S3-3-2 is performed; otherwise, the optimization result of the water treatment data is passed, the detection result of the current water treatment data classification model is retained, and the optimization result of the water treatment data is empty. In this embodiment, the model overfitting problem is found in time to provide a trigger condition for model optimization, and the adverse effects of the overfitting model on data processing are avoided. For the case without overfitting, unnecessary optimization operation is reduced, and the processing efficiency is improved.

[0149] S3-3-2, determine whether the number of subsets of the core water treatment data training set corresponding to the multi-level classification result of the water treatment data is greater than the number of subsets of the auxiliary water treatment data training set corresponding to the multi-level classification result of the water treatment data, if yes, obtain the auxiliary water treatment data training set corresponding to the multi-level classification result of the water treatment data as the to-be-optimized data set, and perform S3-3-3, otherwise, obtain the core water treatment data training set corresponding to the multi-level classification result of the water treatment data as the to-be-optimized data set, and perform S3-3-3;

[0150] Compare the number of subsets of the core and auxiliary training sets corresponding to the multi-level classification result of the water treatment data. If the number of core subsets is greater than that of auxiliary subsets, the auxiliary training set is used as the to-be-optimized data set. Otherwise, the core training set is used as the to-be-optimized data set. This embodiment locates the problem of uneven data distribution by analyzing the difference between the number of subsets of the core and auxiliary training sets, determines the data set that needs to be optimized, and provides a target for solving the model bias caused by uneven data.

[0151] S3-3-3, using the to-be-optimized data set, obtaining the core water treatment data training set and the auxiliary water treatment data training set with the same number of subsets as the updated core water treatment data training set and the auxiliary water treatment data training set, and returning to S1-3;

[0152] Using the to-be-optimized data set, the core and auxiliary training sets with the same number of subsets are obtained as the updated training sets, and then returning to S1-3. This embodiment balances the number of subsets of the core and auxiliary training sets by adjusting the number of subsets, improves the data distribution, provides a more balanced data basis for model retraining, helps to improve the generalization ability and classification accuracy of the model, and reduces the overfitting or underfitting problems caused by uneven data.

[0153] The embodiment S3-3 forms an optimization mechanism for the overfitting problem of the water treatment data classification model. Through overfitting judgment, to-be-optimized data set determination and training set updating, the overfitting problem of the model caused by uneven data distribution can be effectively solved, and the reliability and accuracy of the model are improved. The optimization process ensures that the model can more accurately classify and detect when processing water treatment data, provides more reliable support for subsequent water treatment data monitoring and evaluation, and thus guarantees the stable operation and optimized management of the water treatment system.

[0154] Those skilled in the art will appreciate that embodiments of the application can be devised for a method, a system, or a computer program product. Accordingly, the present application can be embodied in the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present application can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROMs, optical storage devices, etc.) embodying computer readable program code.

[0155] The present application is described in reference to the flowchart illustrations and / or block diagrams of the methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams.

[0156] These computer program instructions can also be stored in a computer- readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instructions which implement the functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams.

[0157] These computer program instructions can also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams. Figure 1 one or more functions specified in the flowchart illustrations and / or block diagrams.

[0158] Finally, it should be noted that the above-mentioned embodiments are merely intended for describing the technical solutions of the present application, but not for limiting it. Although the present application is described in detail with reference to the above embodiments, those skilled in the art should understand that the specific embodiments of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and any modification or equivalent replacement without departing from the spirit and scope of the present application should be covered in the protection scope of the claims of the present application.

Claims

1. A water treatment data monitoring and evaluation method based on artificial intelligence, characterized in that: include: S1. Using water treatment data based on artificial intelligence, a multi-level classification model for water treatment data is established; S2. Obtaining monitoring results of the water treatment data using the water treatment data multi-level classification model based on the water treatment data; S3. Optimize scheduling based on the monitoring results of the water treatment data to obtain monitoring and evaluation results of the water treatment data.

2. The water treatment data monitoring and evaluation method based on artificial intelligence according to claim 1 is characterized in that: Using water treatment data based on artificial intelligence, a multi-level classification model for water treatment data is established, including: Using real-time collected water treatment data to obtain environmental data of water treatment data; Performing correlation screening using the water treatment data based on the environmental data of the water treatment data to obtain correlation screening results of historical water treatment data; A multi-level classification model for water treatment data is established based on artificial intelligence according to the associated screening results of the historical water treatment data.

3. The water treatment data monitoring and evaluation method based on artificial intelligence according to claim 2 is characterized in that: Performing correlation screening on the water treatment data based on the environmental data of the water treatment data to obtain correlation screening results of historical water treatment data includes: According to the water treatment data, obtaining a corresponding historical water treatment data set; Performing correlation screening on the historical water treatment data set using the environmental data of the water treatment data to obtain environmental data corresponding to the historical water treatment data set, wherein the environmental data of the historical water treatment data set includes environmental data of several historical water treatment data subsets; determining in sequence whether the environmental data of the historical water treatment data subset is consistent with the environmental data of the water treatment data; if so, constructing a core water treatment data set using the corresponding consistent historical water treatment data subset; otherwise, constructing an auxiliary water treatment data set using the corresponding inconsistent historical water treatment data subset; The core water treatment data set and the auxiliary water treatment data set are obtained as an association screening result of historical water treatment data.

4. The water treatment data monitoring and evaluation method based on artificial intelligence according to claim 3 is characterized in that: Based on the correlation screening results of the historical water treatment data and artificial intelligence, a multi-level classification model for water treatment data is established, including: S1-3-1. Performing division processing on the associated screening results of the historical water treatment data to obtain division results of a core water treatment data set and division results of an auxiliary water treatment data set, respectively, wherein the division results of the core water treatment data set include a core water treatment data training set and a core water treatment data validation set, and the division results of the auxiliary water treatment data set include an auxiliary water treatment data training set and an auxiliary water treatment data validation set; S1-3-2. Using the division result of the core water treatment data set, obtain an environmental data set corresponding to the core water treatment data set, wherein the environmental data set of the core water treatment data set includes a core environmental data training set and a core environmental data verification set; S1-3-3. Take the core water treatment data training set and the core environmental data training set as input, and the normal state as output. Establish the number of hidden layers in the core environmental data training set corresponding to the type of environmental data, and train an initial core water treatment data classification model based on a convolutional neural network; S1-3-4, inputting the core water treatment data verification set and the core environment data verification set into the initial core water treatment data classification model to obtain an output result of the initial core water treatment data classification model; S1-3-5. Determine whether the output results of the initial core water treatment data classification model are all in a normal state. If so, obtain the initial core water treatment data classification model as the core water treatment data classification model. Otherwise, obtain a core water treatment data verification set and a core environment data verification set that are not in a normal state, add the core water treatment data verification set and the core environment data verification set, and return to S1-3-3. S1-3-6. Using the division result of the auxiliary water treatment data set, obtain an environmental data set corresponding to the auxiliary water treatment data set, wherein the environmental data set of the auxiliary water treatment data set includes a core environmental data training set and an auxiliary environmental data verification set; S1-3-7. Using the auxiliary water treatment data training set and the auxiliary environmental data training set as inputs and the abnormal state as output, establishing a number of hidden layers for the auxiliary environmental data training set corresponding to the type of environmental data, and training an initial auxiliary water treatment data classification model based on a convolutional neural network; S1-3-8. Input the auxiliary water treatment data verification set and the auxiliary environment data verification set into the initial auxiliary water treatment data classification model to obtain an output result of the initial auxiliary water treatment data classification model; S1-3-9. Determine whether the output results of the initial auxiliary water treatment data classification model are all abnormal. If so, obtain the initial auxiliary water treatment data classification model as the auxiliary water treatment data classification model. Otherwise, obtain an auxiliary water treatment data verification set and an auxiliary environmental data verification set that are not in a normal state, add them to the auxiliary water treatment data verification set and the auxiliary environmental data verification set, and return to S1-3-7. S1-3-10. Fusion the core water treatment data classification model with the auxiliary water treatment data classification model to construct a multi-level classification model for water treatment data.

5. The water treatment data monitoring and evaluation method based on artificial intelligence according to claim 4 is characterized in that: Acquiring monitoring results of the water treatment data based on the water treatment data using the water treatment data multi-level classification model includes: Inputting the water treatment data into the water treatment data multi-level classification model to obtain a multi-level classification result of the water treatment data; The water treatment data and the water treatment data multi-level classification model are monitored and processed using the multi-level classification result of the water treatment data to obtain a monitoring result of the water treatment data.

6. The water treatment data monitoring and evaluation method based on artificial intelligence according to claim 5 is characterized in that: The water treatment data and the water treatment data multi-level classification model are monitored and processed using the multi-level classification result of the water treatment data to obtain the monitoring result of the water treatment data, including: S2-2-1. Obtaining environmental data of the water treatment data and the number of training times of the multi-level classification model of the water treatment data; S2-2-2. Compare the multi-level classification result of the water treatment data with the environmental data of the water treatment data to obtain a comparison result of the water treatment data; S2-2-3. Using the environmental data of the water treatment data, obtain the number of data types corresponding to the environmental data; S2-2-4. Determine whether the number of training times of the multi-level classification model for water treatment data is greater than the number of data types of the environmental data. If so, obtain overfitting of the water treatment data classification model as a detection result of the water treatment data classification model and directly execute S2-2-7. Otherwise, execute S2-2-5. S2-2-5. Using the multi-level classification results of the water treatment data, obtain the number of subsets corresponding to the core water treatment data training set and the number of subsets corresponding to the auxiliary water treatment data training set; S2-2-6. Determine whether the relative ratio of the number of subsets in the core water treatment data training set to the number of subsets in the auxiliary water treatment data training set is greater than a target ratio. If so, obtain the presence of overfitting in the water treatment data classification model as a detection result of the water treatment data classification model, and directly execute S2-2-7. Otherwise, obtain the absence of overfitting in the water treatment data classification model as a detection result of the water treatment data classification model, and directly execute S2-2-7. S2-2-7. Obtain the comparison result of the water treatment data and the detection result of the water treatment data classification model as the monitoring result of the water treatment data.

7. The water treatment data monitoring and evaluation method based on artificial intelligence according to claim 6 is characterized in that: Comparing the multi-level classification results of the water treatment data with the environmental data of the water treatment data to obtain the comparison results of the water treatment data includes: S2-2-2-1. Acquire environmental data of the water treatment data as actual environmental data; S2-2-2-2. Based on the actual environmental data, the historical water treatment data set and the environmental data of the historical water treatment data set are used to obtain a historical water treatment data set of the same actual environmental data, wherein the historical water treatment data set of the same actual environmental data includes several historical water treatment data subsets of the same real-time environmental data; S2-2-2-3. Determine whether the corresponding states of several historical water treatment data subsets of the same real-time environmental data are all consistent. If so, obtain the corresponding states of several historical water treatment data subsets of the same real-time environmental data as state information, and execute S2-2-2-4. Otherwise, obtain the corresponding states of a relatively high proportion of historical water treatment data subsets of the same real-time environmental data as state information, and execute S2-2-2-4, wherein the state information includes the normal state or abnormal state of the water treatment data; S2-2-2-4. Determine whether the status information is consistent with the multi-level classification result of the water treatment data. If so, obtain the multi-level classification result of the water treatment data as the comparison result of the water treatment data. Otherwise, use the water treatment data corresponding to the status information and the environmental data of the water treatment data to add them to the corresponding training set, and return to S1-3-3.

8. The water treatment data monitoring and evaluation method based on artificial intelligence according to claim 6 is characterized in that: Optimize scheduling based on the monitoring results of the water treatment data to obtain water treatment data monitoring and evaluation results, including: S3-1. Determine whether the comparison result of the water treatment data is normal. If so, retain the current comparison result of the water treatment data and obtain the scheduling result of the water treatment data. If it is empty, directly execute S3-4. Otherwise, execute S3-2. S3-2. Determine whether the comparison result of the water treatment data corresponds to the water treatment data and the environmental data of the water treatment data, and whether there is historical water treatment data or whether the environmental data of the historical water treatment data does not correspond. If so, obtain the current water treatment data and the environmental data of the current water treatment data, and obtain the scheduling result of the water treatment data. Otherwise, return to using the real-time collected water treatment data to obtain the environmental data of the water treatment data. S3-3. Optimizing the water treatment data using the detection results of the water treatment data classification model to obtain optimized results; S3-4. Obtain water treatment data monitoring and evaluation results using the water treatment data scheduling results and the water treatment data optimization results.

9. The water treatment data monitoring and evaluation method based on artificial intelligence according to claim 8 is characterized in that: Optimizing the water treatment data using the detection results of the water treatment data classification model to obtain optimized results includes: S3-3-1. Determine whether the detection result of the water treatment data classification model is overfitting. If so, execute S3-3-2. Otherwise, retain the detection result of the current water treatment data classification model and obtain the optimization result of the water treatment data as empty. S3-3-2. Determine whether the number of subsets of the core water treatment data training set corresponding to the multi-level classification result of the water treatment data is greater than the number of subsets of the auxiliary water treatment data training set corresponding to the multi-level classification result of the water treatment data. If so, obtain the auxiliary water treatment data training set corresponding to the multi-level classification result of the water treatment data as the data set to be optimized, and execute S3-3-3. Otherwise, obtain the core water treatment data training set corresponding to the multi-level classification result of the water treatment data as the data set to be optimized, and execute S3-3-3. S3-3-3. Using the data set to be optimized, obtain the core water treatment data training set and the auxiliary water treatment data training set with the same number of subsets as the updated core water treatment data training set and the auxiliary water treatment data training set, and return the associated screening results based on the historical water treatment data to establish a multi-level classification model for water treatment data.

Images