A method and system for detecting abnormal pollution discharge of enterprises

Abstract

The present invention discloses a method and system for detecting abnormal pollution discharge of enterprises, which makes full use of unlabeled data and learns an encoder based on the unlabeled data, uses the encoder to similarly encode the same type of data, and makes the encoding of different types of data as different as possible, and then uses the learned encoder to process labeled data, effectively distinguishes the same type of samples from samples of different categories, and uses a classifier to train the data processed by the encoder to improve the prediction performance of the model. In addition, when learning and training the encoder, there is no need to use labeled data, which can make full use of a large amount of unlabeled data in practice and solve the problem of time-consuming and labor-intensive labeling. For the labeled data existing in the data, the advantages of the supervised learning model are fully utilized on the basis of effectively extracting the characteristics of the same type of samples, and the performance of the model is improved. Then, the predicted results of the enterprise pollution discharge data are input by the classifier, which are compared with the current enterprise pollution discharge data, and then the abnormal drainage conditions of the enterprise are accurately judged.

Description

Technical Field

[0001] The present invention relates to the technical field of sewage detection, and in particular to a method and system for detecting abnormal sewage discharge from an enterprise. Background Art

[0002] Enterprise drainage indirectly reflects the enterprise's pollution problem. The current anomaly detection method mainly relies on sampling of environmental monitoring equipment for offline analysis and online monitoring. There are certain limitations in simply designing logic based on partial attribute analysis to judge the anomaly of enterprise drainage. Due to different data sets, anomaly detection can be roughly divided into three categories: fully supervised anomaly detection, semi-supervised anomaly detection, and unsupervised anomaly detection. Fully supervised anomaly detection is for data sets that are all labeled. In actual data sets, most data sets do not have ready-made labels, and sample labeling is a complex and tedious process. Therefore, the application scope of fully supervised anomaly detection is greatly limited. Unsupervised anomaly detection does not require data labeling, but it is inherently more difficult than supervised learning, and the results may not be accurate. In practice, data often have partial labels, making it difficult for supervised learning to process such data, and unsupervised learning processing is not effective.

[0003] It can be concluded that the prior art has the following deficiencies:

[0004] (1) At present, the inspection of enterprise drainage is mainly based on offline analysis of data sampled by environmental protection equipment. Based on the sampled data, the inspection of enterprise drainage anomalies is realized through data analysis or models. It is difficult to effectively realize anomaly detection by simply using data analysis methods. In addition, the data in reality does not contain a large amount of labeled data, making it difficult to directly analyze the abnormality of enterprise drainage.

[0005] (2) Unsupervised methods are difficult to learn and have poor results. Fully supervised methods are more effective, but in practice, data do not contain a large number of labels. Difference-based methods among semi-supervised methods are greatly affected by the model and it is difficult to select a suitable model. Graph-based methods are more complex and adding prediction samples is too complicated. Generative methods rely on a large amount of data during the training process and the output results are unpredictable.

[0006] (3) When faced with complex heterogeneous data, representation learning is used to unify data of different types, modalities, and structures, and to solve problems such as distribution heterogeneity, structural heterogeneity, and modal heterogeneity, thereby providing effective data representation for prediction algorithms. Fully supervised learning dimensionality reduction technology is difficult to apply to unlabeled data. If there is a large amount of unlabeled data in the data, the performance of semi-supervised methods will be seriously affected and it will be difficult to play its own value. Summary of the invention

[0007] The purpose of the present invention is to provide a method and system for detecting abnormal pollution discharge in enterprises, aiming to solve the technical problems mentioned in the background technology.

[0008] In order to achieve the above object, the technical solutions of the present invention are:

[0009] As one aspect of the present application, a method for detecting abnormal pollution discharge of an enterprise is provided, comprising the following steps:

[0010] S1. Obtain the historical pollution discharge data of the target enterprise, and distinguish the historical pollution discharge data into historical unlabeled data and historical labeled data;

[0011] S2, preprocessing the historical unlabeled data and constructing pollution discharge data samples based on the historical unlabeled data, inputting the pollution discharge data samples into the constructed encoder for training, so as to obtain a trained encoder;

[0012] S3, preprocessing the historical labeled data and inputting the historical labeled data into a trained encoder to obtain a labeled data set based on the historical labeled data, inputting the labeled data set and the labeled data corresponding to the labeled data set into a prediction network model for training, so as to obtain a classifier based on the prediction network model, and obtaining the prediction result of the enterprise pollution discharge data through the classifier;

[0013] S4. Obtain the current pollution discharge data of the target enterprise, compare the current pollution discharge data of the target enterprise with the predicted result of the enterprise pollution discharge data, and output the detection result.

[0014] Compared with the prior art, the present invention provides a method for detecting abnormal pollution discharge of enterprises. It makes full use of unlabeled data and learns an encoder based on the unlabeled data. The encoder is used to similarly encode the same type of data, and the encoding of different types of data is made as different as possible. The learned encoder is then used to process the labeled data, effectively distinguishing the same type of samples from samples of different categories, and using a classifier to train the data processed by the encoder to improve the prediction performance of the model. In addition, there is no need to use labeled data when learning and training the encoder, which can make full use of a large amount of unlabeled data in practice and solve the problem of time-consuming and labor-intensive labeling. For the labeled data in the data, the advantages of the supervised learning model are fully utilized on the basis of effectively extracting the characteristics of the same type of samples, improving the performance of the model, and then using the classifier to input the prediction results of the enterprise's pollution discharge data, which is compared with the current enterprise's pollution discharge data, and then accurately judge the abnormal drainage conditions of the enterprise.

[0015] Furthermore, in step S2, the preprocessing of the historical unlabeled data includes:

[0016] Eliminate data that does not conform to business common sense in historical unlabeled data; and

[0017] Fill in the missing data in historical unlabeled data.

[0018] Furthermore, in step S2, the construction of the pollution discharge data sample based on the historical unlabeled data specifically includes:

[0019] Divide the historical unlabeled data into multiple groups of data with a set batch size, each group of data has multiple samples;

[0020] Select a sample from a set of data and take the subsequence X in the sample that has the intersection of data q and subsequences Take a subsequence from the remaining samples in the current data set Subsequence X q and subsequences And subsequence The data is enhanced by Gaussian white noise with set signal-to-noise ratio, and the noise results are obtained. The subsequence X q and subsequences And subsequence Corresponding to anchor points, positive samples and negative samples respectively;

[0021] After traversing the data, the sewage discharge data samples are obtained.

[0022] In step S2, the step of inputting the sewage data sample into the constructed encoder for training to obtain a trained encoder specifically includes:

[0023] The initially constructed encoder is used to encode the sewage data sample to obtain the encoded result;

[0024] A loss function is constructed, and the initially constructed encoder is optimized and trained based on the constructed loss function and the encoding result to obtain a trained encoder.

[0025] Further, in step S3, the historical labeled data is preprocessed and input into a trained encoder to obtain a labeled data set based on the historical labeled data, the labeled data set and the labeled data corresponding to the labeled data set are input into a prediction network model for training to obtain a classifier based on the prediction network model, and the prediction result of the enterprise pollution discharge data is obtained by the classifier, specifically including:

[0026] Use the trained encoder to encode the historical labeled data to obtain a labeled data set; build a prediction network model and obtain the labeled data corresponding to the labeled data set as the data of the prediction network model, and input the labeled data set and the labeled data corresponding to the labeled data set as a training set into a classifier for training, thereby obtaining a classifier based on the prediction network model, which is used to input the prediction results of the enterprise pollution discharge data.

[0027] As a second aspect of the present application, a system for detecting abnormal pollution discharge of an enterprise is provided, comprising:

[0028] A historical data processing unit, the historical data processing unit is used to obtain the historical pollution discharge data of the target enterprise, and distinguish the historical pollution discharge data into historical unlabeled data and historical labeled data;

[0029] An encoder training unit, the encoder training unit is used to preprocess the historical unlabeled data and construct pollution discharge data samples based on the historical unlabeled data, and input the pollution discharge data samples into the constructed encoder for training to obtain a trained encoder;

[0030] A pollution discharge prediction result acquisition unit, the pollution discharge prediction result acquisition unit is used to preprocess the historical labeled data and input the historical labeled data into a trained encoder to obtain a labeled data set based on the historical labeled data, input the labeled data set and the labeled data corresponding to the labeled data set into a prediction network model for training to obtain a classifier based on the prediction network model, and obtain the prediction result of the enterprise pollution discharge data through the classifier;

[0031] The detection result output unit is used to obtain the current pollution discharge data of the target enterprise, compare the current pollution discharge data of the target enterprise with the predicted result of the enterprise's pollution discharge data, and output the detection result.

[0032] As a third aspect of the present application, a computer-readable storage medium is provided, wherein the storage medium stores a program, and the program is executed by a processor to implement the above-mentioned method for detecting abnormal pollution discharge by an enterprise.

[0033] As a fourth aspect of the present application, a computer program product is provided, including a computer program, which, when executed by a processor, implements the above-mentioned method for detecting abnormal pollution discharge by an enterprise.

[0034] For better understanding and implementation, the present invention is described in detail below with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 is a flow chart of a method for detecting abnormal pollution discharge of an enterprise in this embodiment;

[0036] Figure 2 is a flow diagram of a method for detecting abnormal pollution discharge of an enterprise in this embodiment;

[0037] Figure 3 It is a schematic diagram of data subsequence selection in a method for detecting abnormal pollution discharge of an enterprise in this embodiment;

[0038] Figure 4It is a schematic diagram of a system for detecting abnormal pollution discharge by an enterprise in this embodiment. DETAILED DESCRIPTION

[0039] In order to better illustrate the present invention, the present invention is further described in detail below with reference to the accompanying drawings.

[0040] It should be clear that the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the embodiments of the present application.

[0041] The terms used in the embodiments of the present application are only for the purpose of describing specific embodiments, and are not intended to limit the embodiments of the present application. The singular forms of "a", "said" and "the" used in the embodiments of the present application and the appended claims are also intended to include plural forms, unless the context clearly indicates other meanings. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more associated listed items.

[0042] When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present application. On the contrary, they are merely examples of devices and methods consistent with some aspects of the present application as detailed in the attached claims. In the description of the present application, it should be understood that the terms "first", "second", "third", etc. are only used to distinguish similar objects, and do not have to be used to describe a specific order or sequence, nor can they be understood as indicating or implying relative importance. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances.

[0043] In addition, in the description of this application, unless otherwise specified, "plurality" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects are in an "or" relationship.

[0044] The following is an illustrative example. Figure 1 and Figure 2 As shown, a method for detecting abnormal pollution discharge of an enterprise is provided, comprising the following steps:

[0045] Step S1, obtaining the historical pollution discharge data of the target enterprise, and distinguishing the historical pollution discharge data into historical unlabeled data and historical labeled data;

[0046] The data obtained in this embodiment is the historical drainage operation data of the target enterprise for use.

[0047] Step S2, preprocessing the historical unlabeled data and constructing pollution discharge data samples based on the historical unlabeled data, inputting the pollution discharge data samples into the constructed encoder for training to obtain a trained encoder;

[0048] Historical unlabeled data contains numerical, nominal, binary and other attributes, and often contains some "dirty data", such as error values ​​and missing values. Direct use of these data will affect the quality of the data. In order to improve the quality of the data, preprocessing operations are performed on the data. First, the data that does not conform to business common sense is eliminated, such as some data with negative numerical attributes; the missing values ​​that do not contain labeled data are analyzed, and the data is deleted or filled according to the missing data. For example, if the overall missing rate of data attributes is too large, the corresponding data is deleted, otherwise linear interpolation is used for filling.

[0049] In the present embodiment, the application uses historical unlabeled data to train a trained encoder, which involves sample construction and training speed.

[0050] Divide the historical unlabeled data into multiple groups of data with a set batch size, each group of data has multiple samples;

[0051] Combination Figure 3 As shown, select a sample from a set of data and take the subsequence X in the sample that has data intersection q and subsequences Take a subsequence from the remaining samples in the current data set Subsequence X q and subsequences And subsequence The data is enhanced by Gaussian white noise with set signal-to-noise ratio, and the noise results are obtained. The subsequence X q and subsequences And subsequence Corresponding to anchor points, positive samples and negative samples respectively;

[0052] After traversing the data, the sewage discharge data samples are obtained.

[0053] In addition, to speed up the training of the encoder, for example, if there are 1,000 samples, these samples are divided into 100 batches, and the size of each batch is 10. In the process of training the encoder, in order to speed up the training of the model, select data with a batch size of 10 from the dataset.

[0054] In addition, during the training, the pollution discharge data samples are input into the constructed encoder to obtain a trained encoder. Specifically, the pollution discharge data samples are encoded using the initial constructed encoder to obtain the encoded result; then the loss function is constructed, and the initial constructed encoder is optimized and trained based on the constructed loss function combined with the encoded result to obtain the trained encoder.

[0055] Step S3, preprocessing the historical labeled data and inputting the historical labeled data into a trained encoder to obtain a labeled data set based on the historical labeled data, inputting the labeled data set and the labeled data corresponding to the labeled data set into a prediction network model for training to obtain a classifier based on the prediction network model, and obtaining the prediction result of the enterprise pollution discharge data through the classifier;

[0056] Specifically, the trained encoder is used to encode the historical labeled data to obtain a labeled data set; a prediction network model is constructed and the labeled data corresponding to the labeled data set is obtained as the data of the prediction network model, and the labeled data set and the labeled data corresponding to the labeled data set are input into a classifier as a training set for training, thereby obtaining a classifier based on the prediction network model, which is used to input the prediction results of the enterprise pollution discharge data.

[0057] After the encoder is obtained through training and optimization based on unlabeled data, the labeled data is used to build a supervised learning model. First, the labeled data encoded by the encoder is used to obtain the encoded labeled data set, ensuring that the encoded data can effectively distinguish between samples of the same category and samples of different categories, and then the labeled data corresponding to the labeled data set is taken as the data of the prediction network model. For the construction of the prediction network model, a multi-layer perceptron (MLP) network is selected as the prediction network, and the labeled data set and the labeled data corresponding to the labeled data set are divided into a training set, and then the classifier is trained using the training set. Based on the existing classifier application, in the classifier, the number and size of hidden layers, activation function, weight optimizer, regularization term parameter, learning rate and number of iterations are used as hyperparameters, and the grid search algorithm is used to optimize the hyperparameters. After the training is completed, a classifier based on the prediction network model containing hyperparameter values ​​is obtained, and the abnormal drainage status of the enterprise is monitored according to the prediction results of the enterprise drainage data.

[0058] Step S4: obtaining the current pollution discharge data of the target enterprise, comparing the current pollution discharge data of the target enterprise with the predicted result of the enterprise pollution discharge data, and outputting the detection result.

[0059] Obtain the current pollution discharge data of the target enterprise, compare it with the predicted results of the enterprise's pollution discharge data, and while outputting the test results, push the information of the enterprise with drainage abnormalities to relevant technical personnel for secondary verification to realize the inspection of drainage abnormalities of the enterprise.

[0060] Compared with the prior art, the present invention provides a method for detecting abnormal pollution discharge of enterprises. It makes full use of unlabeled data and learns an encoder based on the unlabeled data. The encoder is used to similarly encode the same type of data, and the encoding of different types of data is made as different as possible. The learned encoder is then used to process the labeled data, effectively distinguishing the same type of samples from samples of different categories, and using a classifier to train the data processed by the encoder to improve the prediction performance of the model. In addition, there is no need to use labeled data when learning and training the encoder, which can make full use of a large amount of unlabeled data in practice and solve the problem of time-consuming and labor-intensive labeling. For the labeled data in the data, the advantages of the supervised learning model are fully utilized on the basis of effectively extracting the characteristics of the same type of samples, improving the performance of the model, and then using the classifier to input the prediction results of the enterprise's pollution discharge data, which is compared with the current enterprise's pollution discharge data, and then accurately judge the abnormal drainage conditions of the enterprise.

[0061] As two aspects of this application, Figure 4 As shown, a system for detecting abnormal pollution discharge of an enterprise is provided, comprising:

[0062] A historical data processing unit 100, which is used to obtain historical pollution discharge data of a target enterprise and distinguish the historical pollution discharge data into historical unlabeled data and historical labeled data;

[0063] An encoder training unit 200, the encoder training unit is used to pre-process the historical unlabeled data and construct pollution data samples based on the historical unlabeled data, and input the pollution data samples into the constructed encoder for training to obtain a trained encoder;

[0064] A pollution discharge prediction result acquisition unit 300, which is used to preprocess the historical labeled data and input the historical labeled data into a trained encoder to obtain a labeled data set based on the historical labeled data, input the labeled data set and the labeled data corresponding to the labeled data set into a prediction network model for training to obtain a classifier based on the prediction network model, and obtain the prediction result of the enterprise pollution discharge data through the classifier;

[0065] The detection result output unit 400 is used to obtain the current pollution discharge data of the target enterprise, compare the current pollution discharge data of the target enterprise with the predicted result of the enterprise's pollution discharge data, and output the detection result.

[0066] As a third aspect of the present application, a computer-readable storage medium is provided, wherein the storage medium stores a program, and the program is executed by a processor to implement the above-mentioned method for detecting abnormal pollution discharge by an enterprise.

[0067] As a fourth aspect of the present application, a computer program product is provided, including a computer program, which, when executed by a processor, implements the above-mentioned method for detecting abnormal pollution discharge by an enterprise.

[0068] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0069] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0070] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0071] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0072] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0073] The memory may include non-permanent storage in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0074] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0075] According to the disclosure and teaching of the above description, those skilled in the art to which the present invention belongs may also make changes and modifications to the above embodiments. Therefore, the present invention is not limited to the specific embodiments disclosed and described above, and some modifications and changes to the present invention should also fall within the scope of protection of the claims of the present invention. In addition, although some specific terms are used in this specification, these terms are only for the convenience of description and do not constitute any limitation to the present invention.

Claims

1. A method for detecting abnormal pollution discharge in an enterprise, characterized in that: The steps include: S1. Obtain the historical pollution discharge data of the target enterprise, and distinguish the historical pollution discharge data into historical unlabeled data and historical labeled data; S2, preprocessing the historical unlabeled data and constructing pollution discharge data samples based on the historical unlabeled data, inputting the pollution discharge data samples into the constructed encoder for training, so as to obtain a trained encoder, wherein the encoder performs similar encoding on the same type of data and makes the encoding of different types of data as different as possible; S3, preprocessing the historical labeled data and inputting the historical labeled data into a trained encoder to obtain a labeled data set based on the historical labeled data, inputting the labeled data set and the labeled data corresponding to the labeled data set into a prediction network model for training, so as to obtain a classifier based on the prediction network model, and obtaining the prediction result of the enterprise pollution discharge data through the classifier; S4. Obtain the current pollution discharge data of the target enterprise, compare the current pollution discharge data of the target enterprise with the predicted result of the enterprise pollution discharge data, and output the detection result; Among them, in step S2, the construction of the pollution data sample based on the historical unlabeled data specifically includes: dividing the historical unlabeled data into multiple groups of data according to the set batch size, each group of data has multiple samples; selecting a sample from a group of data, and taking the subsequence X with data intersection in the sample q and subsequences Take a subsequence from the remaining samples in the current data set Subsequence X q and subsequences And subsequence The data is enhanced by Gaussian white noise with set signal-to-noise ratio, and the noise results are obtained. The subsequence X q and subsequences And subsequence They correspond to anchor points, positive samples and negative samples respectively; after traversing the data, the pollution data samples are obtained.

2. The method for detecting abnormal pollution discharge of enterprises according to claim 1, characterized in that: In step S2, the preprocessing of the historical unlabeled data includes: Eliminate data that does not conform to business common sense from historical unlabeled data; and Fill in the missing data in historical unlabeled data.

3. The method for detecting abnormal pollution discharge of enterprises according to claim 1, characterized in that: In step S2, the step of inputting the sewage data sample into the constructed encoder for training to obtain a trained encoder specifically includes: The initially constructed encoder is used to encode the sewage data sample to obtain the encoded result; A loss function is constructed, and the initially constructed encoder is optimized and trained based on the constructed loss function and the encoding result to obtain a trained encoder.

4. The method for detecting abnormal pollution discharge of enterprises according to claim 3, characterized in that: In step S3, the historical labeled data is preprocessed and the historical labeled data is input into a trained encoder to obtain a labeled data set based on the historical labeled data, the labeled data set and the labeled data corresponding to the labeled data set are input into a prediction network model for training to obtain a classifier based on the prediction network model, and the prediction result of the enterprise pollution discharge data is obtained through the classifier, specifically including: encoding the historical labeled data using the trained encoder to obtain a labeled data set; A prediction network model is constructed and the labeled data corresponding to the labeled data set is obtained as the data of the prediction network model. The labeled data set and the labeled data corresponding to the labeled data set are input into a classifier as a training set for training, thereby obtaining a classifier based on the prediction network model. The classifier is used to input the prediction results of the enterprise pollution discharge data.

5. An enterprise sewage discharge abnormality detection system, characterized in that: include: A historical data processing unit, the historical data processing unit is used to obtain the historical pollution discharge data of the target enterprise, and distinguish the historical pollution discharge data into historical unlabeled data and historical labeled data; An encoder training unit, the encoder training unit is used to preprocess the historical unlabeled data and construct pollution data samples based on the historical unlabeled data, input the pollution data samples into the constructed encoder for training, so as to obtain a trained encoder, the encoder performs similar encoding on the same type of data and makes the encoding of different types of data as different as possible; A pollution discharge prediction result acquisition unit, the pollution discharge prediction result acquisition unit is used to preprocess the historical labeled data and input the historical labeled data into a trained encoder to obtain a labeled data set based on the historical labeled data, input the labeled data set and the labeled data corresponding to the labeled data set into a prediction network model for training to obtain a classifier based on the prediction network model, and obtain the prediction result of the enterprise pollution discharge data through the classifier; A detection result output unit, the detection result output unit is used to obtain the current pollution discharge data of the target enterprise, compare the current pollution discharge data of the target enterprise with the predicted result of the enterprise pollution discharge data, and output the detection result; The method of constructing a pollution data sample based on screening historical unlabeled data specifically includes: dividing the historical unlabeled data into multiple groups of data according to a set batch size, each group of data has multiple samples; selecting a sample from a group of data, and taking a subsequence X with data intersection in the sample q and subsequences Take a subsequence from the remaining samples in the current data set Subsequence X q and subsequences And subsequence The data is enhanced by Gaussian white noise with set signal-to-noise ratio, and the noise results are obtained. The subsequence X q and subsequences And subsequence They correspond to anchor points, positive samples and negative samples respectively; after traversing the data, the pollution data samples are obtained.

6. A computer-readable storage medium, characterized in that: The storage medium stores a program, and the program is executed by a processor to implement the enterprise pollution discharge abnormality detection method according to any one of claims 1 to 4.

7. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method for detecting abnormal pollution discharge by an enterprise as claimed in any one of claims 1 to 4 is implemented.

Images