Industrial sewage illegal discharge monitoring system and method

Abstract

The present application provides a system and method for monitoring the illegal discharge of industrial sewage. The system includes at least a processing module, an electricity data collection module, a water data collection module and an alarm module, wherein the processing module is used to determine whether the electricity consumption of a target plant area is abnormal based on a first electricity data sequence; if so, determine whether there is illegal discharge of sewage in the target plant area according to the water flow monitoring data and drainage images of the drainage monitoring points corresponding to the target plant area. If so, perform an alarm action to improve the accuracy of monitoring the illegal discharge of sewage and clearly locate the object that actually discharges sewage.

Description

Technical Field

[0001] The present application relates to the field of urban construction technology, and in particular to a system and method for monitoring the illegal discharge of industrial sewage. Background Art

[0002] The discharge of industrial wastewater has always been an important factor affecting river water quality. Existing wastewater discharge monitoring technologies mostly rely on the placement of monitoring equipment in rivers, outlets, and drainage networks, and use the collected water quality, water volume, water level and other parameters to determine the discharge of wastewater. However, due to the large number of factories and complex drainage networks in the park, traditional methods often cannot trace the source, are not accurate, and have poor interpretability. Summary of the invention

[0003] The purpose of the embodiments of the present application is to provide a system and method for monitoring the illegal discharge of industrial sewage, so as to improve the accuracy of monitoring the illegal discharge of sewage and clearly locate the objects that actually discharge sewage.

[0004] In a first aspect, the present application provides an industrial sewage illegal discharge monitoring system, the system comprises at least a processing module, an electricity data collection module, a water data collection module and an alarm module, wherein the electricity data collection module is used to collect a first electricity data sequence of a target plant area within a first preset time period, and send it to the processing module; the water data collection module is used to collect water flow monitoring data and drainage images of drainage monitoring points, and send them to the processing module; the processing module is used to determine whether the electricity consumption of the target plant area is abnormal based on the first electricity data sequence; if so, determine whether there is sewage illegal discharge in the target plant area according to the water flow monitoring data and drainage images of the drainage monitoring points corresponding to the target plant area, and if so, execute the alarm. An alarm action is taken; wherein the processing module determines whether the power consumption of the target plant is abnormal in the following manner: a first power consumption data sequence within a first preset time period of the target plant is input into a power consumption data prediction model, so as to predict a second power consumption data sequence generated by the target plant within a second preset time period after the first preset time period through the power consumption data prediction model, wherein the first preset time period is greater than the second preset time period; the first power consumption data sequence and the second power consumption data sequence are spliced ​​to obtain a target power consumption data sequence; the target power consumption data sequence is input into a power consumption data recognition model to obtain a power consumption data recognition result output by the power consumption data recognition model, and the power consumption data recognition result is used to indicate whether the power consumption of the target plant is abnormal.

[0005] In an optional embodiment, the water data acquisition module includes a satellite, the drainage image includes a multispectral image and a full-color image of the drainage monitoring point acquired by the satellite, and the processing module obtains the sewage illegal discharge detection result output by the sewage illegal discharge detection model in the following manner:

[0006] The multispectral image and the panchromatic image are fused to generate a fused image; the fused image is input into the sewage illegal discharge detection model for target detection to determine the sewage illegal discharge detection result.

[0007] In an optional embodiment, the processing module generates a fused image in the following manner:

[0008] The multispectral image and the panchromatic image are respectively converted from the RGB color space to the IHS color space to obtain the hue component, saturation component and brightness component of the multispectral image and the panchromatic image respectively; the brightness component of the multispectral image and the brightness component of the panchromatic image are weightedly summed to obtain a fused brightness component; based on the fused brightness component and the hue component and saturation component of the multispectral image, a fused image in the IHS color space is constructed; the fused image in the IHS color space is inversely transformed into a fused image in the RGB color space.

[0009] In an optional implementation, the sewage illegal discharge detection model is generated based on YOLO model training, and the processing module is specifically used for:

[0010] The fused image is input into the sewage illegal discharge detection model for target detection to obtain the first number of detection frames of preset sizes output by the sewage illegal discharge detection model and the confidence and IOU value corresponding to each detection frame; multiple detection frames are screened based on the IOU value and the confidence to generate a second number of screened detection frames; it is determined whether there is a detection frame with an intersection area greater than a preset area in the positional relationship between the second number of detection frames on the fused image; if so, it is determined that sewage illegal discharge detection exists.

[0011] In an optional implementation, the electricity consumption data prediction model is trained and generated in the following manner:

[0012] Collect historical data sequences of a third preset time period generated by electricity consumption in multiple factory areas under normal and illegal sewage discharge conditions; divide each historical electricity consumption data into a first historical data sequence of a first preset time period and a second historical data sequence of a second preset time period; input the first historical data sequence of the first preset time period into the LSTM model in sequence, so that it outputs the second historical data sequence of the second preset time period; evaluate the prediction ability of the model, and adjust the model parameters to generate a trained electricity consumption data prediction model.

[0013] In an optional implementation, the electricity consumption data recognition model is trained and generated in the following manner:

[0014] Label the collected historical data sequences of multiple third preset time periods, including whether the electricity consumption is abnormal and whether it is a high-incidence time; input the labeled third preset time periods into the LSTM model in sequence, so that it outputs the corresponding label values; calculate the difference between the label value and the predicted value corresponding to each third preset time period according to the preset loss function, and adjust the model parameters to generate a trained electricity consumption data recognition model.

[0015] In an optional implementation, the preset loss function is:

[0016]

[0017]

[0018] Among them, ω1 and ω2 are weight coefficients, y i1 is the abnormal power consumption label value corresponding to the historical data sequence i, p i1 is the abnormal power consumption prediction value of historical data sequence i output by the model, y i2 is the prone time label value corresponding to the historical data sequence i, p i2 It is the predicted value of the prone time of the historical data series i output by the model.

[0019] In a second aspect, the present application provides a method for monitoring the illegal discharge of industrial sewage, the method comprising:

[0020] Determine whether the target plant's electricity consumption is abnormal based on the first electricity consumption data sequence; if so, determine whether there is any illegal sewage discharge in the target plant based on the water flow monitoring data and drainage image corresponding to the target plant, and if so, execute alarm measures; wherein, determine whether the target plant's electricity consumption is abnormal in the following manner: input a first electricity consumption data sequence within a first preset time period of the target plant into an electricity consumption data prediction model, so as to predict a second electricity consumption data sequence generated by the target plant within a second preset time period after the first preset time period through the electricity consumption data prediction model, wherein the first preset time period is greater than the second preset time period; splice the first electricity consumption data sequence and the second electricity consumption data sequence to obtain a target electricity consumption data sequence; input the target electricity consumption data sequence into an electricity consumption data recognition model to obtain an electricity consumption data recognition result output by the electricity consumption data recognition model, and the electricity consumption data recognition result is used to indicate whether the target plant's electricity consumption is abnormal.

[0021] In an optional embodiment, the water flow monitoring data includes one or more of a water flow velocity value, a water level height value, a water flow rate value, and a water quality parameter content value, and whether there is illegal sewage discharge in the target plant area is determined by the following methods:

[0022] For each item of water flow monitoring data, determine whether the real-time difference between the current value and the average value of the water flow monitoring data is greater than the preset difference, and the average value is the average value of the water flow monitoring data within the third preset time period; if so, input the drainage image into the sewage illegal discharge detection model to obtain the sewage illegal discharge detection result output by the sewage illegal discharge detection model, and the sewage illegal discharge detection result is used to indicate whether there is illegal sewage discharge in the target plant area.

[0023] The present application provides a system and method for monitoring the illegal discharge of industrial sewage, the system at least includes a processing module, an electricity data collection module, a water data collection module and an alarm module, wherein the electricity data collection module is used to collect a first electricity data sequence of a target plant area within a first preset time period, and send it to the processing module; the water data collection module is used to collect water flow monitoring data and drainage images of drainage monitoring points, and send them to the processing module; the processing module is used to determine whether the electricity consumption of the target plant area is abnormal based on the first electricity data sequence; if so, determine whether there is illegal discharge of sewage in the target plant area according to the water flow monitoring data and drainage images of the drainage monitoring points corresponding to the target plant area, and if so, execute an alarm Action; wherein, the processing module determines whether the target plant area's electricity consumption is abnormal in the following manner: inputting the first electricity consumption data sequence within the first preset time length of the target plant area into the electricity consumption data prediction model, so as to predict the second electricity consumption data sequence generated by the target plant area within the second preset time length after the first preset time length through the electricity consumption data prediction model, wherein the first preset time length is greater than the second preset time length; splicing the first electricity consumption data sequence and the second electricity consumption data sequence to obtain the target electricity consumption data sequence; inputting the target electricity consumption data sequence into the electricity consumption data recognition model to obtain the electricity consumption data recognition result output by the electricity consumption data recognition model, and the electricity consumption data recognition result is used to indicate whether the electricity consumption of the target plant area is abnormal. Combining the electricity consumption data of the plant area with the detection parameters and images of the drainage monitoring points, it is possible to jointly determine whether there is a phenomenon of illegal sewage discharge in the factory, accurately locate the object that actually discharges sewage, and use the AI ​​model to recognize the drainage image with high accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments of the present application will be briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present application and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without paying creative work.

[0025] Figure 1 A schematic diagram of the structure of an industrial sewage illegal discharge monitoring system provided in an embodiment of the present application;

[0026] Figure 2A flowchart of the first step of determining whether the power consumption of a target plant area is abnormal is provided for the embodiment of the present application;

[0027] Figure 3 A flow chart of a method for monitoring the illegal discharge of industrial wastewater provided in an embodiment of the present application. DETAILED DESCRIPTION

[0028] The technical solutions in the embodiments of the present application will be described below in conjunction with the drawings in the embodiments of the present application.

[0029] Figure 1 The present invention provides a schematic diagram of a system for monitoring the illegal discharge of industrial sewage. The present invention provides a system for monitoring the illegal discharge of industrial sewage, the system comprising a processing module, an electricity data collection module, a water data collection module and an alarm module.

[0030] Among them, the electricity consumption data collection module is used to collect the first electricity consumption data sequence of the target factory area within a first preset time period and send it to the processing module.

[0031] Here, for the power distribution rooms of different factories and units in the industrial park, corresponding power consumption data collection modules can be arranged respectively. For large factories with different process flows and different workshops, power consumption data collection modules can also be arranged for the power distribution rooms of different workshops. The power consumption data collection module here can be an electric energy meter. The power consumption data sequence includes power consumption and timestamp, etc.

[0032] The water data acquisition module is used to collect water flow monitoring data and drainage images from drainage monitoring points and send them to the processing module.

[0033] For industrial parks and their nearby rivers, multiple drainage monitoring points can be reasonably set up, such as river channels, outlets, drainage network nodes, etc. The water data acquisition module here can include flow rate sensors, liquid level sensors, water quality sensors, etc. Water quality sensors can be used to collect the pH, nitrogen, phosphorus, sulfide, chloride and organic compound content in water, as well as the content of four heavy metals: mercury, cadmium, lead and chromium.

[0034] The water data acquisition module also includes satellites or drones, etc., for collecting drainage images. The drainage images here are specifically multispectral images and panchromatic images of drainage monitoring points collected by multispectral sensors carried by satellites or drones.

[0035] The processing module is used to determine whether the target plant area has abnormal electricity consumption based on the first electricity consumption data sequence. If so, determine whether the target plant area has sewage discharge secretly based on the water flow monitoring data and drainage image of the drainage monitoring point corresponding to the target plant area. If so, perform an alarm action.

[0036] Here, the processing module first determines whether there is an abnormal increase in electricity consumption through the electricity consumption data corresponding to each factory area. If the electricity consumption of a factory area or workshop increases, the water flow monitoring data and drainage images corresponding to the factory area or workshop are obtained.

[0037] Water flow monitoring data may include water flow velocity values, water level height values, water flow rate values, water quality parameter content values, and the like.

[0038] The processing module can determine, for each item of water flow monitoring data, whether the real-time difference between the current value and the average value of the water flow monitoring data is greater than the preset difference, and the average value is the average value of the water flow monitoring data within the third preset time period.

[0039] For example, the water flow velocity value and the water flow rate value can be monitored, and the real-time difference between the average value in the last minute and the current value can be calculated. The real-time difference is compared with the preset difference. If it is less than the preset difference, the next step of judgment is not entered. The preset difference here can be obtained by pre-calibration, and different preset differences can be set in different time periods. For example, the preset difference can be reduced during the night time to improve the sensitivity of the detection.

[0040] It is also possible to add corresponding water quality parameter content judgment rules based on the material content of the wastewater generated by the factory.

[0041] If so, the drainage image is input into the sewage illegal discharge detection model to obtain the sewage illegal discharge detection result output by the sewage illegal discharge detection model. The sewage illegal discharge detection result is used to indicate whether there is sewage illegal discharge in the target factory area.

[0042] Here, the AI ​​model is finally used to identify the multispectral image and full-color image formed by the reflection of the water body in multiple bands, and further determine whether there is any illegal discharge of sewage.

[0043] Specifically, the multispectral image and the panchromatic image may be fused to generate a fused image; the fused image is input into a sewage illegal discharge detection model for target detection to determine a sewage illegal discharge detection result.

[0044] The alarm module can promptly report to supervisors when there is illegal sewage discharge so that the corresponding factory area can be dealt with. It can also cut off power to the corresponding factory area to avoid the expansion of sewage discharge and reduce pollution to the environment.

[0045] The present application provides an industrial sewage illegal discharge monitoring system, which combines the power consumption data of the factory area with the detection parameters and images of the drainage monitoring points to jointly determine whether there is sewage illegal discharge in the factory. It can accurately locate the objects that actually discharge sewage, and use the AI ​​model to identify the drainage images of the drainage monitoring points with high accuracy.

[0046] In one embodiment of the present application, in order to improve the accuracy of identifying abnormalities in electricity consumption data, it is possible to consider identifying the electricity consumption data sequences of each plant area through an AI model. The selection of the length of the electricity consumption data sequence is an important aspect. In order to improve the recognition accuracy of the electricity consumption data recognition model, it is necessary to increase the length of the electricity consumption data sequence. However, the electricity consumption data sequence is real-time. If the collection length of the electricity consumption data sequence is simply increased, it will not only affect the timeliness of monitoring, but also fail to improve the model recognition effect. Therefore, the present application also proposes a power consumption data prediction model, which predicts the power consumption data for a period of time in the future based on the power consumption data of the plant area in the current period of time, splices it into complete power consumption data, and then uses the power consumption data recognition model to identify abnormalities.

[0047] Specifically, Figure 2 The processing module can determine whether the target plant area's electricity consumption is abnormal by:

[0048] S100. Input a first electricity consumption data sequence within a first preset time period of a target plant into an electricity consumption data prediction model, so as to predict a second electricity consumption data sequence generated by the target plant within a second preset time period after the first preset time period through the electricity consumption data prediction model, wherein the first preset time period is greater than the second preset time period.

[0049] S110: Concatenate the first power usage data sequence and the second power usage data sequence to obtain a target power usage data sequence.

[0050] In a specific embodiment, the first preset time length may be 10 minutes, and the second preset time length may be 3 minutes or 5 minutes. That is, the power consumption data prediction model is used to predict the second power consumption data sequence of the next 3 or 5 minutes based on the first power consumption data sequence of the target plant area in the previous 10 minutes.

[0051] S120, inputting the target power consumption data sequence into the power consumption data recognition model to obtain the power consumption data recognition result output by the power consumption data recognition model, wherein the power consumption data recognition result is used to indicate whether the power consumption of the target plant area is abnormal.

[0052] In this way, by collecting the actual data generated in the factory and the predicted results of future electricity consumption, the electricity consumption can be jointly identified, which not only improves the effectiveness of model identification, but also improves the timeliness of the system in monitoring illegal sewage discharge.

[0053] Here, the training process of the power consumption data prediction model is described by taking the second preset time length of 5 minutes as an example. Specifically, the power consumption data prediction model can be trained and generated in the following manner:

[0054] Collect historical data series of the third preset time period generated by electricity consumption of multiple plant areas under normal and illegal sewage discharge conditions.

[0055] The third preset time length here can be a 15-minute continuous historical electricity consumption data sequence, and the step length can be half a minute or 10 seconds. Here, the 15-minute historical electricity consumption data sequence can include data collected under normal electricity consumption in different periods of the factory area, data collected from the Nth minute of illegal sewage discharge, and data collected from the Nth minute of illegal sewage discharge.

[0056] And preprocess the collected historical data series, including removing outliers, removing duplicate data, filling missing values, etc. The data can also be normalized to improve the convergence speed of the model.

[0057] The sample set is divided into training set and test set according to the ratio.

[0058] Each historical electricity consumption data is divided into a first historical data sequence of a first preset time length and a second historical data sequence of a second preset time length.

[0059] The first historical data sequence of the first preset time length is sequentially input into the LSTM model so that it outputs the second historical data sequence of the second preset time length.

[0060] The prediction ability of the model is evaluated and the model parameters are adjusted to generate a trained electricity consumption data prediction model.

[0061] In a specific embodiment, an LSTM (Long Short-Term Memory Network) model suitable for processing time series data can be selected as the initial model. The training is performed using samples in the training set. The test set is used to evaluate the model performance, and indicators such as mean square error and root mean square error can be used.

[0062] By predicting the electricity consumption data sequence that may be generated in the factory in the future, the predicted electricity consumption data sequence and the actually collected electricity consumption data sequence are spliced ​​together to identify anomalies in electricity consumption data, which can improve the accuracy of anomalies in electricity consumption data and the timeliness of detection of illegal discharge of sewage.

[0063] Furthermore, the electricity consumption data recognition model can be trained and generated in the following ways:

[0064] In the 15-minute historical electricity consumption data sequence, a one-step historical electricity consumption data sequence can also be extracted, and the remaining part is predicted and generated using an existing electricity consumption data prediction model. Here, the electricity consumption data prediction model trained and generated in the aforementioned exemplary embodiment can be used, or other models with the same function available on the Internet can be used.

[0065] Labels are added to the collected historical data sequences of the third preset time period, where the labels include whether the power consumption is abnormal and whether it is a high-incidence time.

[0066] The same sample data as the electricity consumption data prediction model can be used here. The labels here may include a first label and a second label. The first label is used to indicate whether there is an abnormal increase in electricity consumption in the factory due to illegal sewage discharge. The second label is used to indicate whether the time corresponding to the electricity consumption data sequence is a time when illegal sewage discharge is prone to occur, such as abnormal working hours such as night or early morning.

[0067] The marked third preset time length is sequentially input into the LSTM model so that it outputs the corresponding label value.

[0068] Similarly, the LSTM model can be used for training. Since this involves a multi-classification problem, in order to further improve the accuracy of the model, the cross entropy function can be used to optimize the model.

[0069] The difference between the label value and the predicted value corresponding to each third preset time length is calculated according to the preset loss function, and the model parameters are adjusted to generate a trained electricity consumption data recognition model.

[0070] The preset loss function that can be used here can be:

[0071]

[0072] Among them, ω1 and ω2 are weight coefficients, y i1 is the abnormal power consumption label value corresponding to the historical data sequence i, p i1 is the abnormal power consumption prediction value of historical data sequence i output by the model, y i2 is the prone time label value corresponding to the historical data sequence i, p i2 is the predicted value of the prone time of historical data series i output by the model, and N is the number of samples.

[0073] In one embodiment of the present application, the drainage image captures multispectral images and full-color images of the drainage monitoring points. A sewage illegal discharge detection model is provided here to identify the spectrum with large fluctuations on the drainage image caused by illegal sewage discharge.

[0074] Specifically, the sewage illegal discharge detection model can be trained and generated in the following way.

[0075] 1. Sample collection and pretreatment

[0076] The 4m multispectral images and 1m panchromatic images of the water surface collected by satellite-borne sensors in different weather conditions and light intensities during the illegal discharge of sewage. The multispectral images here contain multiple bands (blue, green, red, and near-infrared), each of which covers a different spectral range.

[0077] The multispectral image and the panchromatic image are fused to generate a fused image. Specifically, the fused image can be generated in the following ways:

[0078] The multispectral image and the panchromatic image are respectively converted from the RGB color space to the IHS color space to obtain the hue component, saturation component and brightness component of the multispectral image and the panchromatic image respectively.

[0079] Using the HIS color space conversion method, the calculation formula for the intensity component I of each image can be:

[0080] I = (R + G + B) / 3;

[0081] The calculation formula of the hue component H can be:

[0082]

[0083] The calculation formula of the brightness component S can be:

[0084]

[0085] Among them, R, G, and B are the corresponding three-channel pixel values.

[0086] The brightness component of the multispectral image and the brightness component of the panchromatic image are weighted summed to obtain the fused brightness component.

[0087] The brightness component is reconstructed by the brightness component of the multispectral image and the brightness component of the panchromatic image, which can enhance the quality of the final fused image. The weight coefficient of the multispectral image can be greater than the weight coefficient of the panchromatic image. Other fusion algorithms can also be used to reconstruct the brightness component, such as through deep learning models. It should be noted that before reconstruction, the panchromatic image can be downsampled to the same resolution as the multispectral image, and then the fused brightness component is calculated.

[0088] Based on the fusion of the brightness component and the hue component and saturation component of the multispectral image, a fused image in the IHS color space is constructed.

[0089] The fused image in the IHS color space is inversely transformed into a fused image in the RGB color space.

[0090] The fused image in the sample is annotated. Here, the identification frame and label of the preset size can be used for annotation. The identification frame is used to indicate the location of the large ripples in the image caused by the illegal discharge of sewage, and the label is used to indicate whether there is illegal discharge of sewage. The size of the identification frame can be minimized to ensure the accuracy of recognition. For example, the fused image of the sample can usually be cropped to a size of 640×640 or 448×448. Here, the size of the identification frame can be selected to be 100×100, 80×60 or even smaller.

[0091] The labeled samples are divided into test set and training set.

[0092] 2. Model Selection and Training

[0093] In a specific embodiment, a YOLO model may be selected to train a sewage illegal discharge detection model, and specifically, the YOLO model may be a YOLOv7 model.

[0094] Taking YOLOv7 as an example, the model specifically includes three parts: Backbone, Neck and Head. In order to make the model better identify the spectral characteristics in water bodies, the backbone network of the model can be improved, using ResNet101 or ResNeXt as the network structure.

[0095] And after the Neck network part processes the downsampling module, a certain number of feature recalibration modules (such as Global Context Block, GCB) are added before the activation function layer.

[0096] The model is trained using the training set, and the cross entropy loss function and Adam optimizer are used to adjust and optimize the model. After the training is completed, the model output results are post-processed, and non-maximum suppression (NMS) is calculated to filter out repeated or low-confidence recognition boxes, and each recognition result is cropped to a specified size. Finally, the model is verified and tested to evaluate whether the model performance, such as accuracy and recall, meets the requirements.

[0097] In the actual deployment process, since the actual recognition frame of the model is cropped during the post-processing process, the following steps can be used to judge the sewage discharge identification results based on the recognition frame:

[0098] The fused image is input into the sewage illegal discharge detection model for target detection to obtain a first number of detection frames of preset sizes output by the sewage illegal discharge detection model and the confidence and IOU value corresponding to each detection frame. The detection frame of preset size here is the result of cropping after post-processing.

[0099] The plurality of detection boxes are screened based on the IOU value and the confidence level to generate a second number of screened detection boxes.

[0100] First, the detection boxes are screened by IOU value and confidence level to select the detection boxes with higher probability.

[0101] The second number of detection frames are cropped, and after the cropping, the multiple detection frames share an IOU value and a confidence value.

[0102] Determine whether there are any detection frames whose intersection area is greater than a preset area in the positional relationship between the second number of detection frames on the fused image. If so, determine that there is a case of illegal sewage discharge detection.

[0103] Here it is necessary to calculate the intersection area between the cropped detection frames with different IOU values ​​or confidence values. If there are two cropped detection frames with an area larger than the preset area, it can be determined that a hit has been detected and there is sewage discharge.

[0104] In this way, by cropping the detection frame and then making an intersection judgment, even smaller ripples can be effectively identified, ensuring the accuracy of sewage discharge detection.

[0105] In one embodiment of the present application, Figure 3 As shown, a method for monitoring the illegal discharge of industrial sewage is also provided, the method comprising:

[0106] S1. Determine whether the power consumption of the target plant area is abnormal according to the first power consumption data sequence;

[0107] S2. If yes, for each item of water flow monitoring data, determine whether the real-time difference between the current value and the average value of the water flow monitoring data is greater than a preset difference, and the average value is the average value of the water flow monitoring data within a third preset time period;

[0108] S3. If yes, the drainage image is input into the sewage illegal discharge detection model to obtain the sewage illegal discharge detection result output by the sewage illegal discharge detection model, and the sewage illegal discharge detection result is used to indicate whether there is sewage illegal discharge in the target factory area;

[0109] S4. Determine whether there is illegal sewage discharge in the target factory area according to the illegal sewage discharge detection result output by the illegal sewage discharge detection model;

[0110] S5. If yes, execute alarm measures.

[0111] Among them, whether the electricity consumption of the target plant is abnormal is determined in the following manner: a first electricity consumption data sequence within a first preset time period of the target plant is input into an electricity consumption data prediction model, so as to predict a second electricity consumption data sequence generated by the target plant within a second preset time period after the first preset time period through the electricity consumption data prediction model, wherein the first preset time period is greater than the second preset time period; the first electricity consumption data sequence and the second electricity consumption data sequence are spliced ​​to obtain a target electricity consumption data sequence; the target electricity consumption data sequence is input into an electricity consumption data recognition model to obtain an electricity consumption data recognition result output by the electricity consumption data recognition model, and the electricity consumption data recognition result is used to indicate whether the electricity consumption of the target plant is abnormal.

[0112] In the embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, and the indirect coupling or communication connection of the devices or units can be electrical, mechanical or other forms.

[0113] In addition, the units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0114] Furthermore, the functional modules in the various embodiments of the present application may be integrated together to form an independent part, or each module may exist separately, or two or more modules may be integrated to form an independent part.

[0115] It should be noted that if the function is implemented in the form of a software function module and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can essentially be embodied in the form of a software product, or the part that contributes to the prior art or the part of the technical solution. The computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM) random access memory (RAM), disk or optical disk, and other media that can store program codes.

[0116] In this document, relational terms such as first and second, etc. are used merely to distinguish one entity or operation from another entity or operation, but do not necessarily require or imply any such actual relationship or order between these entities or operations.

[0117] The above description is only an embodiment of the present application and is not intended to limit the protection scope of the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. An industrial sewage illegal discharge monitoring system, characterized in that: The system at least includes a processing module, an electricity data collection module, a water data collection module and an alarm module, wherein: The power consumption data collection module is used to collect a first power consumption data sequence of the target plant area within a first preset time period and send it to the processing module; The water data acquisition module is used to collect water flow monitoring data and drainage images of drainage monitoring points and send them to the processing module; The processing module is used to determine whether the power consumption of the target plant area is abnormal based on the first power consumption data sequence; If yes, then determine whether there is illegal sewage discharge in the target plant area according to the water flow monitoring data and drainage images of the drainage monitoring points corresponding to the target plant area. If yes, perform an alarm action; The processing module determines whether the target plant area's electricity consumption is abnormal in the following manner: Inputting a first power consumption data sequence within a first preset time period of the target plant area into a power consumption data prediction model, so as to predict a second power consumption data sequence generated by the target plant area within a second preset time period after the first preset time period by the power consumption data prediction model, wherein the first preset time period is greater than the second preset time period; splicing the first power consumption data sequence and the second power consumption data sequence to obtain a target power consumption data sequence; The target power consumption data sequence is input into the power consumption data recognition model to obtain the power consumption data recognition result output by the power consumption data recognition model, and the power consumption data recognition result is used to indicate whether the power consumption of the target plant area is abnormal.

2. The system according to claim 1, characterized in that The water flow monitoring data includes one or more of the water flow velocity value, the water level height value, the water flow rate value, and the water quality parameter content value. The processing module determines whether there is illegal sewage discharge in the target plant area by the following methods: For each item of water flow monitoring data, determine whether a real-time difference between a current value and an average value of the water flow monitoring data is greater than a preset difference, wherein the average value is an average value of the water flow monitoring data within a third preset time period; If so, the drainage image is input into the sewage illegal discharge detection model to obtain the sewage illegal discharge detection result output by the sewage illegal discharge detection model, and the sewage illegal discharge detection result is used to indicate whether there is sewage illegal discharge in the target factory area.

3. The system according to claim 2, characterized in that The water data acquisition module includes a satellite, the drainage image includes a multispectral image and a panchromatic image of the drainage monitoring point acquired by the satellite, and the processing module obtains the sewage illegal discharge detection result output by the sewage illegal discharge detection model in the following manner: fusing the multispectral image and the panchromatic image to generate a fused image; The fused image is input into the sewage illegal discharge detection model for target detection to determine the sewage illegal discharge detection result.

4. The system according to claim 3, characterized in that The processing module generates a fused image in the following manner: Respectively converting the multispectral image and the panchromatic image from the RGB color space to the IHS color space to obtain a hue component, a saturation component, and a brightness component of the multispectral image and the panchromatic image; Performing a weighted summation on the brightness component of the multispectral image and the brightness component of the panchromatic image to obtain a fused brightness component; Based on the fused brightness component and the hue component and saturation component of the multispectral image, a fused image in an IHS color space is formed; The fused image in the IHS color space is inversely transformed into a fused image in the RGB color space.

5. The system according to claim 3, characterized in that The sewage illegal discharge detection model is generated based on YOLO model training, and the processing module is specifically used for: Inputting the fused image into the sewage illegal discharge detection model for target detection to obtain a first number of detection frames of preset sizes output by the sewage illegal discharge detection model and a confidence level and an IOU value corresponding to each detection frame; Filtering a plurality of detection boxes based on the IOU value and the confidence level to generate a second number of filtered detection boxes; Determine whether there are any detection frames whose intersection area is greater than a preset area in the positional relationship between the second number of detection frames on the fused image; If so, it is determined that there is illegal discharge of sewage.

6. The system according to claim 1, characterized in that The electricity consumption data prediction model is trained and generated in the following manner: Collect historical data series of a third preset time period generated by electricity consumption in multiple plant areas under normal and illegal sewage discharge conditions; Dividing each historical electricity consumption data into a first historical data sequence of a first preset time length and a second historical data sequence of a second preset time length; Sequentially inputting a first historical data sequence of a first preset time length into the LSTM model so that it outputs a second historical data sequence of a second preset time length; The prediction ability of the model is evaluated and the model parameters are adjusted to generate a trained electricity consumption data prediction model.

7. The system according to claim 6, characterized in that The electricity consumption data recognition model is trained and generated in the following manner: Labeling the collected historical data sequences of the third preset time period with labels, wherein the labels include whether the power consumption is abnormal and whether it is a prone time; The marked third preset duration is sequentially input into the LSTM model so that it outputs the corresponding label value; The difference between the label value and the predicted value corresponding to each third preset time length is calculated according to the preset loss function, and the model parameters are adjusted to generate a trained electricity consumption data recognition model.

8. The system according to claim 7, characterized in that The preset loss function is: Among them, ω1 and ω2 are weight coefficients, y i1 is the abnormal power consumption label value corresponding to the historical data sequence i, p i1 is the abnormal power consumption prediction value of historical data sequence i output by the model, y i2 is the prone time label value corresponding to the historical data sequence i, p i2 It is the predicted value of the prone time of the historical data series i output by the model.

9. A method for monitoring the illegal discharge of industrial sewage, characterized in that: The method comprises: Determining whether the power consumption of the target plant area is abnormal according to the first power consumption data sequence; If yes, then determine whether there is illegal sewage discharge in the target plant area according to the water flow monitoring data and drainage images corresponding to the target plant area. If yes, execute alarm measures; Among them, whether the power consumption of the target plant is abnormal is determined by the following methods: Inputting a first power consumption data sequence within a first preset time period of the target plant area into a power consumption data prediction model, so as to predict a second power consumption data sequence generated by the target plant area within a second preset time period after the first preset time period by the power consumption data prediction model, wherein the first preset time period is greater than the second preset time period; splicing the first power consumption data sequence and the second power consumption data sequence to obtain a target power consumption data sequence; The target power consumption data sequence is input into the power consumption data recognition model to obtain the power consumption data recognition result output by the power consumption data recognition model, and the power consumption data recognition result is used to indicate whether the power consumption of the target plant area is abnormal.

10. The method according to claim 9, characterized in that The water flow monitoring data includes one or more of the water flow velocity value, water level height value, water flow rate value, and water quality parameter content value, and is used to determine whether there is illegal sewage discharge in the target plant area by the following methods: For each item of water flow monitoring data, determine whether a real-time difference between a current value and an average value of the water flow monitoring data is greater than a preset difference, wherein the average value is an average value of the water flow monitoring data within a third preset time period; If so, the drainage image is input into the sewage illegal discharge detection model to obtain the sewage illegal discharge detection result output by the sewage illegal discharge detection model, and the sewage illegal discharge detection result is used to indicate whether there is sewage illegal discharge in the target factory area.

Images