Outlet water quality parameter prediction method and device based on full-connection neural network model

Abstract

The invention provides an effluent quality parameter prediction method and device based on a full-connection neural network model, relates to the technical field of water quality prediction, is applied to a sewage treatment plant, and comprises the following steps: obtaining real-time monitoring data of the sewage treatment plant; inputting the real-time monitoring data into a pre-constructed effluent quality parameter prediction model, and outputting an effluent quality parameter at a preset time point; wherein the pre-constructed effluent quality parameter prediction model is obtained by inputting a target monitoring data sequence into a pre-constructed full-connection neural network model in a manner of overlapping time windows for training; the target monitoring data sequence is obtained by cleaning the monitoring data sequence through a preset data cleaning strategy; the monitoring data sequence comprises a plurality of historical monitoring data acquired at a preset time interval. In the mode, when the effluent quality is predicted, limited data resources are effectively utilized, and time sequence change characteristics in the sewage treatment process are captured and analyzed.

Description

Technical Field

[0001] The present invention relates to the technical field of water quality prediction, and in particular to a method and device for predicting effluent water quality parameters based on a fully connected neural network model. Background Art

[0002] The efficient and stable operation of sewage treatment plants requires real-time knowledge of the sewage treatment process status, and obtaining the future effluent quality in advance can help operation and maintenance personnel take countermeasures in advance, so it is of great significance. The prediction of effluent quality is often based on mechanism models, but the mechanism model is difficult to adapt to the increasingly complex sewage treatment process, and its prediction accuracy is relatively low. The rapid development of big data technology has provided a new way to use daily monitoring data to predict the effluent quality of sewage treatment plants. Many studies have explored the application of advanced models such as long short-term memory networks and recurrent neural networks to predict water quality parameters. However, these deep learning models are usually accompanied by high data volumes and computing resource requirements, which may exceed the carrying capacity of many sewage treatment plants under existing conditions in reality. In contrast, although the traditional fully connected neural network has a relatively mild demand for computing resources, it is insufficient in dealing with changes in sewage parameters with time series characteristics.

[0003] Therefore, it is particularly urgent and important to develop a prediction method that can effectively utilize limited data resources and properly capture and analyze the time series change characteristics in the sewage treatment process. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a method and device for predicting effluent water quality parameters based on a fully connected neural network model, so as to effectively utilize limited data resources in prediction and capture and analyze the time series change characteristics in the sewage treatment process.

[0005] In a first aspect, an embodiment of the present invention provides a method for predicting effluent water quality parameters based on a fully connected neural network model, which is applied to a sewage treatment plant, and the method includes: obtaining real-time monitoring data of the sewage treatment plant; inputting the real-time monitoring data into a pre-constructed effluent water quality parameter prediction model, and outputting the effluent water quality parameters at a preset time point; wherein the pre-constructed effluent water quality parameter prediction model is obtained by inputting a target monitoring data sequence into a pre-constructed fully connected neural network model through overlapping time windows for training; the target monitoring data sequence is obtained by cleaning the monitoring data sequence through a pre-set data cleaning strategy; the monitoring data sequence includes multiple historical monitoring data obtained at preset time intervals.

[0006] In a preferred embodiment of the present invention, the above-mentioned cleaning process of the monitoring data sequence is carried out through a pre-set data cleaning strategy, including: eliminating historical monitoring data in the monitoring data sequence that does not meet the preset standard deviation range to obtain multiple target historical monitoring data; based on the distance between adjacent target historical monitoring data, linearly filling missing values ​​and then normalizing to obtain the target monitoring data sequence.

[0007] In a preferred embodiment of the present invention, the above-mentioned fully connected neural network model is characterized by four layers; wherein, the first layer includes 256 neural network nodes, the second layer includes 128 neural network nodes, the third layer includes 64 neural network nodes, and the fourth layer includes 32 neural network nodes; the first layer and the second layer both discard 30% of the neural network nodes, the third layer discards 20% of the neural network nodes, and the fourth layer does not discard any neural network nodes.

[0008] In a preferred embodiment of the present invention, the data type of the input layer of the above-mentioned input fully connected neural network model includes a vector formed by a sliding time window; the output of the output layer of the fully connected neural network model includes a predicted value corresponding to a preset time point; and the loss function of the fully connected neural network model is the average loss of the predicted value corresponding to each preset time point.

[0009] In a preferred embodiment of the present invention, the training method of the above-mentioned effluent water quality parameter prediction model is: inputting the target monitoring data sequence into the fully connected neural network model through overlapping time windows; at the same time, inputting the water inlet monitoring data and process parameters at x moments to complete the training in the effluent water quality parameter prediction model; wherein x is the size of the overlapping time window.

[0010] In a preferred embodiment of the present invention, the above-mentioned inputting of the target monitoring data sequence into the fully connected neural network model by means of overlapping time windows includes: using overlapping time windows to input the water inlet monitoring data and process parameters in the target monitoring data sequence into the fully connected neural network model; the overlap amount of the overlapping time windows is y; wherein y<x.

[0011] In a preferred embodiment of the present invention, the above-mentioned historical monitoring data include: inlet chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, suspended solids, dissolved oxygen in the biochemical pool, suspended solids in the biochemical pool, aeration volume in the biochemical pool, transmembrane pressure difference, and transmembrane flow rate; inlet monitoring data include: inlet chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, and suspended solids; process parameters include: biochemical pool dissolved oxygen, biochemical pool suspended solids, biochemical pool aeration volume, transmembrane pressure difference, and transmembrane flow rate; effluent water quality parameters include: effluent chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, and suspended solids.

[0012] In the second aspect, an embodiment of the present invention also provides an effluent water quality parameter prediction device based on a fully connected neural network model, which is applied to a sewage treatment plant, comprising: a real-time monitoring data acquisition module, used to obtain real-time monitoring data of the sewage treatment plant; an effluent water quality parameter output module, used to input the real-time monitoring data into a pre-constructed effluent water quality parameter prediction model, and output the effluent water quality parameters at a preset time point; wherein the pre-constructed effluent water quality parameter prediction model is obtained by inputting the target monitoring data sequence into a pre-constructed fully connected neural network model through overlapping time windows for training; the target monitoring data sequence is obtained by cleaning the monitoring data sequence through a pre-set data cleaning strategy; the monitoring data sequence includes multiple historical monitoring data acquired at preset time intervals.

[0013] In a third aspect, an embodiment of the present invention further provides an electronic device, comprising a processor and a memory, wherein the memory stores computer executable instructions that can be executed by the processor, and the processor executes the computer executable instructions to implement the method for predicting effluent water quality parameters based on a fully connected neural network model of the first aspect mentioned above.

[0014] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the method for predicting effluent water quality parameters based on a fully connected neural network model of the first aspect.

[0015] The embodiments of the present invention bring the following beneficial effects:

[0016] The embodiment of the present invention provides a method and device for predicting effluent water quality parameters based on a fully connected neural network model, which obtains real-time monitoring data from a sewage treatment plant, inputs the real-time monitoring data into a pre-constructed effluent water quality parameter prediction model, and outputs effluent water quality parameters at a preset time point. The pre-constructed effluent water quality parameter prediction model is obtained by inputting a target monitoring data sequence into a pre-constructed fully connected neural network model through overlapping time windows for training; the target monitoring data sequence is obtained by cleaning the monitoring data sequence through a pre-set data cleaning strategy; the monitoring data sequence includes multiple historical monitoring data obtained at preset time intervals. In this method, when predicting effluent water quality, limited data resources are effectively utilized, and the time series change characteristics during sewage treatment are captured and analyzed.

[0017] Other features and advantages of the present disclosure will be set forth in the following description, or some features and advantages may be inferred or unambiguously determined from the description, or may be learned by implementing the above-mentioned technology of the present disclosure.

[0018] In order to make the above-mentioned objectives, features and advantages of the present disclosure more obvious and easy to understand, preferred embodiments are specifically cited below and described in detail with reference to the attached drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the specific implementation methods of the present invention or the technical solutions in the prior art, the drawings required for use in the specific implementation methods or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some implementation methods of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0020] Figure 1 A flow chart of a method for predicting effluent water quality parameters based on a fully connected neural network model provided in an embodiment of the present invention;

[0021] Figure 2 A flowchart of another method for predicting effluent water quality parameters based on a fully connected neural network model provided by an embodiment of the present invention;

[0022] Figure 3 A schematic diagram of the structure of an outlet water quality parameter prediction device based on a fully connected neural network model provided in an embodiment of the present invention;

[0023] Figure 4 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] The efficient and stable operation of sewage treatment plants requires real-time knowledge of the sewage treatment process status, and obtaining the future effluent quality in advance can help operation and maintenance personnel take countermeasures in advance, so it is of great significance. The prediction of effluent quality is often based on mechanism models, but the mechanism model is difficult to adapt to the increasingly complex sewage treatment process, and its prediction accuracy is relatively low. The rapid development of big data technology has provided a new way to use daily monitoring data to predict the effluent quality of sewage treatment plants. Many studies have explored the application of advanced models such as long short-term memory networks and recurrent neural networks to predict water quality parameters. However, these deep learning models are usually accompanied by high data volumes and computing resource requirements, which may exceed the carrying capacity of many sewage treatment plants under existing conditions in reality. In contrast, although the traditional fully connected neural network has a relatively mild demand for computing resources, it is insufficient in dealing with changes in sewage parameters with time series characteristics.

[0026] Therefore, it is particularly urgent and important to develop a prediction method that can effectively utilize limited data resources and properly capture and analyze the time series change characteristics in the sewage treatment process.

[0027] Based on this, an embodiment of the present invention provides a method and device for predicting effluent water quality parameters based on a fully connected neural network model. The method and device can obtain real-time monitoring data of a sewage treatment plant, input the real-time monitoring data into a pre-constructed effluent water quality parameter prediction model, and output the effluent water quality parameters at a preset time point. Among them, the pre-constructed effluent water quality parameter prediction model is obtained by inputting the target monitoring data sequence into a pre-constructed fully connected neural network model through overlapping time windows for training; the target monitoring data sequence is obtained by cleaning the monitoring data sequence through a pre-set data cleaning strategy; the monitoring data sequence includes multiple historical monitoring data obtained at preset time intervals. In this way, when predicting effluent water quality, limited data resources are effectively utilized, and the time series change characteristics during sewage treatment are captured and analyzed. To facilitate the understanding of this embodiment, a method for predicting effluent water quality parameters based on a fully connected neural network model disclosed in an embodiment of the present invention is first introduced in detail.

[0028] Example 1

[0029] The embodiment of the present invention provides a method for predicting effluent water quality parameters based on a fully connected neural network model, which is applied to a sewage treatment plant. Figure 1 The present invention provides a flowchart of a method for predicting effluent water quality parameters based on a fully connected neural network model. Figure 1 As shown, the method for predicting effluent water quality parameters based on the fully connected neural network model may include the following steps:

[0030] Step S101, obtaining real-time monitoring data of the sewage treatment plant.

[0031] Among them, real-time monitoring data may include influent chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, suspended solids, biochemical pool dissolved oxygen, biochemical pool suspended solids, biochemical pool aeration volume, transmembrane pressure difference, and transmembrane flow.

[0032] Step S102, input the real-time monitoring data into a pre-built effluent water quality parameter prediction model, and output the effluent water quality parameters at a preset time point.

[0033] The effluent water quality parameters include: effluent chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, and suspended solids. It should be noted that the output effluent water quality parameter can be at least one of effluent chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, and suspended solids.

[0034] Among them, the pre-constructed effluent water quality parameter prediction model is obtained by inputting the target monitoring data sequence into the pre-constructed fully connected neural network model through overlapping time windows for training; the target monitoring data sequence is obtained after cleaning the monitoring data sequence through a pre-set data cleaning strategy; the monitoring data sequence includes multiple historical monitoring data obtained at preset time intervals.

[0035] Exemplarily, the preset time interval may be 1 hour, and the plurality of historical monitoring data obtained may be obtained within a period of time, and the period of time may be within a specified time period, such as between date A and date B.

[0036] Among them, historical monitoring data include: inlet chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, suspended solids, biochemical pool dissolved oxygen, biochemical pool suspended solids, biochemical pool aeration volume, transmembrane pressure difference, and transmembrane flow.

[0037] Specifically, the fully connected neural network model features four layers.

[0038] Among them, the first layer includes 256 neural network nodes, the second layer includes 128 neural network nodes, the third layer includes 64 neural network nodes, and the fourth layer includes 32 neural network nodes.

[0039] Among them, since not discarding some nodes will cause the model's fitting rate for the training data to reach 100%, but the fitting effect for the new data is very poor, 30% of the neural network nodes are discarded in the first and second layers, 20% of the neural network nodes are discarded in the third layer, and no neural network nodes are discarded in the fourth layer for overfitting.

[0040] Among them, the learning rate of the fully connected neural network model during the training process can be 0.0003, the number of times it completely traverses the data set, that is, the number of target monitoring data sequences, can be 500, the number of samples used to calculate gradients and update weights in one iteration can be 16, and 20% of the data can be divided from the target monitoring data sequence as a validation set.

[0041] Specifically, the data type of the input layer of the fully connected neural network model includes a vector formed by a sliding time window; the output of the output layer of the fully connected neural network model includes a predicted value corresponding to a preset time point; and the loss function of the fully connected neural network model is the average loss of the predicted value corresponding to each preset time point.

[0042] Specifically, the training method of the effluent water quality parameter prediction model is:

[0043] The target monitoring data sequence is input into the fully connected neural network model through overlapping time windows; at the same time, the inlet monitoring data and process parameters at x moments are input to complete the training of the effluent water quality parameter prediction model.

[0044] Where x is the size of the overlapping time window, for example, x moments can be 24 hours.

[0045] Among them, the inlet monitoring data include: inlet chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, and suspended solids.

[0046] Among them, the process parameters include: dissolved oxygen in the biochemical pool, suspended solids in the biochemical pool, aeration volume in the biochemical pool, transmembrane pressure difference, and transmembrane flow rate.

[0047] Specifically, inputting the target monitoring data sequence into the fully connected neural network model through overlapping time windows may include: using overlapping time windows to input the water inlet monitoring data and process parameters in the target monitoring data sequence into the fully connected neural network model.

[0048] The overlap amount of the overlapping time windows is y, wherein y<x. Since the sample size of the embodiment of the present application is relatively small, the overlap amount may be 23 hours.

[0049] Therefore, when the data volume and computing power requirements are low, the characteristics of the time series in the sewage treatment process and the effluent quality at multiple time points can be taken into account.

[0050] The effluent water quality parameter prediction method based on the fully connected neural network model provided by the embodiment of the present invention can obtain the real-time monitoring data of the sewage treatment plant, input the real-time monitoring data into the pre-constructed effluent water quality parameter prediction model, and output the effluent water quality parameters at a preset time point. Among them, the pre-constructed effluent water quality parameter prediction model is obtained by inputting the target monitoring data sequence into the pre-constructed fully connected neural network model through overlapping time windows for training; the target monitoring data sequence is obtained by cleaning the monitoring data sequence through a pre-set data cleaning strategy; the monitoring data sequence includes multiple historical monitoring data obtained at preset time intervals. In this method, when predicting the effluent water quality, limited data resources are effectively utilized, and the time series change characteristics in the sewage treatment process are captured and analyzed.

[0051] Example 2

[0052] The embodiment of the present invention also provides another method for predicting effluent water quality parameters based on a fully connected neural network model; this method is implemented on the basis of the method in the above embodiment; this method focuses on describing the specific implementation method of cleaning the monitoring data sequence through a pre-set data cleaning strategy.

[0053] Figure 2 A flowchart of another method for predicting effluent water quality parameters based on a fully connected neural network model provided by an embodiment of the present invention is shown in FIG. Figure 2 As shown, the cleaning process of the monitoring data sequence by using the pre-set data cleaning strategy may include the following steps:

[0054] Step S201, eliminating historical monitoring data that does not meet a preset standard deviation range in a monitoring data sequence to obtain a plurality of target historical monitoring data.

[0055] Specifically, historical monitoring data that does not meet three times the standard deviation in the monitoring data sequence can be eliminated to obtain multiple target historical monitoring data.

[0056] Among them, the standard deviation range is the range of plus or minus 3 times the average value of the historical monitoring data. The calculation of the standard deviation is the conventional standard deviation calculation formula. After calculating the standard deviation of each historical monitoring data, the historical monitoring data outside the range of plus or minus 3 times the standard deviation of the average value is eliminated.

[0057] The calculation formula of standard deviation can be:

[0058] Where s represents the standard deviation, N represents the number of time points, and x i represents the historical monitoring data at the i-th time point in the monitoring data sequence, Represents the average value of historical monitoring data at all time points. Represents the square of the difference between the historical monitoring data at each time point and the average value.

[0059] Specifically, in, represents the average value of historical monitoring data at all time points, N represents the number of time points, x i Represents the historical monitoring data at the i-th time point in the monitoring data sequence.

[0060] Step S202, based on the distance between adjacent target historical monitoring data, linearly fill in missing values ​​and then perform normalization processing to obtain a target monitoring data sequence.

[0061] Specifically, since historical monitoring data that does not meet the preset standard deviation range is eliminated from the monitoring data sequence, missing values ​​will appear, so the missing values ​​need to be supplemented.

[0062] Among them, based on the distance between adjacent target historical monitoring data, missing values ​​can be filled in according to a preset linear missing value filling rule to obtain an initial monitoring data sequence, and then the initial monitoring data sequence is normalized to obtain a target monitoring data sequence.

[0063] Among them, the linear filling missing value rule can be:

[0064] If the distance characterization target historical monitoring data has non-missing values ​​only on the left side, the values ​​on the left side are used to fill the missing values; if the distance characterization target historical monitoring data has non-missing values ​​only on the right side, the values ​​on the right side are used to fill the missing values; if the distance characterization target historical monitoring data has non-missing values ​​on both sides, linear interpolation is performed based on the non-missing values ​​on the left and right sides to obtain estimated values ​​of the missing values, so as to fill the missing values ​​with the estimated values.

[0065] Specifically, the estimated value is calculated using the following formula:

[0066] Among them, y i Represents the estimated value of the missing value; t0 represents the position of the non-missing value on the left, and its corresponding value is x0; t1 represents the position of the non-missing value on the right, and its corresponding value is x1.

[0067] Among them, for normalization processing, the maximum and minimum values ​​of the values ​​in the initial monitoring data sequence can be determined, and then the normalization processing is performed based on the maximum, minimum and original values ​​of the data in the initial monitoring data sequence through the following formula:

[0068] The method for predicting effluent water quality parameters based on a fully connected neural network model provided in an embodiment of the present invention can obtain multiple target historical monitoring data by eliminating historical monitoring data that does not meet a preset standard deviation range in a monitoring data sequence, and then linearly fill in missing values ​​based on the distance between adjacent target historical monitoring data and perform normalization to obtain a target monitoring data sequence, thereby completing data cleaning so that the fully connected neural network model can be trained through the target monitoring data sequence.

[0069] Example 3

[0070] Corresponding to the above method embodiment, the embodiment of the present invention provides a device for predicting effluent water quality parameters based on a fully connected neural network model. Figure 3 A schematic diagram of the structure of a device for predicting effluent water quality parameters based on a fully connected neural network model provided by an embodiment of the present invention is shown in FIG. Figure 3 As shown, the effluent water quality parameter prediction device based on the fully connected neural network model may include:

[0071] The real-time monitoring data acquisition module 301 is used to acquire the real-time monitoring data of the sewage treatment plant.

[0072] The outlet water quality parameter output module 302 is used to input the real-time monitoring data into a pre-built outlet water quality parameter prediction model, and output the outlet water quality parameters at a preset time point.

[0073] Among them, the pre-constructed effluent water quality parameter prediction model is obtained by inputting the target monitoring data sequence into the pre-constructed fully connected neural network model through overlapping time windows for training; the target monitoring data sequence is obtained after cleaning the monitoring data sequence through a pre-set data cleaning strategy; the monitoring data sequence includes multiple historical monitoring data obtained at preset time intervals.

[0074] The effluent water quality parameter prediction device based on the fully connected neural network model provided by the embodiment of the present invention can obtain the real-time monitoring data of the sewage treatment plant, input the real-time monitoring data into the pre-constructed effluent water quality parameter prediction model, and output the effluent water quality parameters at a preset time point. Among them, the pre-constructed effluent water quality parameter prediction model is obtained by inputting the target monitoring data sequence into the pre-constructed fully connected neural network model through overlapping time windows for training; the target monitoring data sequence is obtained by cleaning the monitoring data sequence through a pre-set data cleaning strategy; the monitoring data sequence includes multiple historical monitoring data obtained at preset time intervals. In this way, when predicting the effluent water quality, limited data resources are effectively utilized, and the time series change characteristics in the sewage treatment process are captured and analyzed.

[0075] In some embodiments, the outlet water quality parameter output module is also used to eliminate historical monitoring data in the monitoring data sequence that does not meet the preset standard deviation range to obtain multiple target historical monitoring data; based on the distance between adjacent target historical monitoring data, the missing values ​​are linearly filled and normalized to obtain the target monitoring data sequence.

[0076] In some embodiments, the fully connected neural network model is characterized by four layers; wherein the first layer includes 256 neural network nodes, the second layer includes 128 neural network nodes, the third layer includes 64 neural network nodes, and the fourth layer includes 32 neural network nodes; the first and second layers both discard 30% of the neural network nodes, the third layer discards 20% of the neural network nodes, and the fourth layer does not discard any neural network nodes.

[0077] In some embodiments, the data type of the input layer of the fully connected neural network model includes a vector formed by a sliding time window; the output of the output layer of the fully connected neural network model includes a predicted value corresponding to a preset time point; and the loss function of the fully connected neural network model is the average loss of the predicted value corresponding to each preset time point.

[0078] In some embodiments, the effluent water quality parameter output module is also used to input the target monitoring data sequence into the fully connected neural network model through overlapping time windows; at the same time, the water inlet monitoring data and process parameters at x moments are input to complete the training in the effluent water quality parameter prediction model; wherein x is the size of the overlapping time window.

[0079] In some embodiments, the outlet water quality parameter output module is also used to input the inlet monitoring data and process parameters in the target monitoring data sequence into the fully connected neural network model using overlapping time windows; the overlap amount of the overlapping time windows is y; wherein y<x.

[0080] In some embodiments, historical monitoring data include: inlet chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, suspended solids, dissolved oxygen in biochemical pool, suspended solids in biochemical pool, aeration volume in biochemical pool, transmembrane pressure difference, and transmembrane flow rate; inlet monitoring data include: inlet chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, and suspended solids; process parameters include: dissolved oxygen in biochemical pool, suspended solids in biochemical pool, aeration volume in biochemical pool, transmembrane pressure difference, and transmembrane flow rate; effluent water quality parameters include: effluent chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, and suspended solids.

[0081] The device provided in the embodiment of the present invention has the same implementation principle and technical effects as those of the aforementioned method embodiment. For the sake of brief description, for matters not mentioned in the device embodiment, reference may be made to the corresponding contents in the aforementioned method embodiment.

[0082] Example 4

[0083] The embodiment of the present invention also provides an electronic device for running the above-mentioned method for predicting effluent water quality parameters based on the fully connected neural network model; see Figure 4 A structural schematic diagram of an electronic device is shown, which includes a memory 400 and a processor 401, wherein the memory 400 is used to store one or more computer instructions, and the one or more computer instructions are executed by the processor 401 to implement the above-mentioned method for predicting effluent water quality parameters based on a fully connected neural network model.

[0084] Further, Figure 4 The electronic device shown further includes a bus 402 and a communication interface 403 , and the processor 401 , the communication interface 403 and the memory 400 are connected via the bus 402 .

[0085] The memory 400 may include a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk storage. The communication connection between the system network element and at least one other network element is realized through at least one communication interface 403 (which may be wired or wireless), and the Internet, wide area network, local area network, metropolitan area network, etc. may be used. The bus 402 may be an ISA bus, a PCI bus, or an EISA bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 Only one bidirectional arrow is used in the diagram, but this does not mean that there is only one bus or only one type of bus.

[0086] The processor 401 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 401. The above processor 401 can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The methods, steps and logic block diagrams disclosed in the embodiments of the present invention can be implemented or executed. The general processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiment of the present invention can be directly embodied as a hardware decoding processor for execution, or a combination of hardware and software modules in the decoding processor for execution. The software module may be located in a storage medium mature in the art, such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 400, and the processor 401 reads the information in the memory 400 and completes the steps of the method of the above embodiment in combination with its hardware.

[0087] An embodiment of the present invention also provides a computer-readable storage medium, which stores computer-executable instructions. When the computer-executable instructions are called and executed by a processor, the computer-executable instructions prompt the processor to implement the above-mentioned method for predicting effluent water quality parameters based on a fully connected neural network model. The specific implementation can be found in the method embodiment, which will not be repeated here.

[0088] The computer program product for the method for predicting effluent water quality parameters based on a fully connected neural network model provided in an embodiment of the present invention includes a computer-readable storage medium storing a non-volatile program code executable by a processor. The instructions included in the program code can be used to execute the method in the previous method embodiment. The specific implementation can be found in the method embodiment, which will not be repeated here.

[0089] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0090] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. The device embodiments described above are only schematic. For example, the division of 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 devices or units can be electrical, mechanical or other forms.

[0091] 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.

[0092] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0093] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for enabling 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 methods described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0094] Finally, it should be noted that the above-described embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above-mentioned embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above-mentioned embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A method for predicting effluent water quality parameters based on a fully connected neural network model, characterized in that: Applied to a sewage treatment plant, the method comprises: Obtaining real-time monitoring data of the sewage treatment plant; Input the real-time monitoring data into a pre-built effluent water quality parameter prediction model to output the effluent water quality parameters at a preset time point; Among them, the pre-constructed effluent water quality parameter prediction model is obtained by inputting the target monitoring data sequence into a pre-constructed fully connected neural network model through overlapping time windows for training; the target monitoring data sequence is obtained after cleaning the monitoring data sequence through a pre-set data cleaning strategy; the monitoring data sequence includes multiple historical monitoring data obtained at preset time intervals.

2. The method according to claim 1, characterized in that The cleaning process of the monitoring data sequence by using a preset data cleaning strategy includes: Eliminate historical monitoring data that does not meet a preset standard deviation range in the monitoring data sequence to obtain multiple target historical monitoring data; Based on the distance between adjacent target historical monitoring data, missing values ​​are linearly filled and then normalized to obtain the target monitoring data sequence.

3. The method according to claim 1, characterized in that The fully connected neural network model features four layers; Among them, the first layer includes 256 neural network nodes, the second layer includes 128 neural network nodes, the third layer includes 64 neural network nodes, and the fourth layer includes 32 neural network nodes; the first layer and the second layer both discard 30% of the neural network nodes, the third layer discards 20% of the neural network nodes, and the fourth layer does not discard any neural network nodes.

4. The method according to claim 3, characterized in that The data type of the input layer of the fully connected neural network model includes a vector formed by a sliding time window; the output of the output layer of the fully connected neural network model includes the predicted value corresponding to the preset time point; the loss function of the fully connected neural network model is the average loss of the predicted value corresponding to each preset time point.

5. The method according to claim 1, characterized in that The training method of the effluent water quality parameter prediction model is: Inputting the target monitoring data sequence into the fully connected neural network model in the form of overlapping time windows; At the same time, input water inlet monitoring data and process parameters at x moments to complete the training of the effluent water quality parameter prediction model; Wherein, x is the size of the overlapping time window.

6. The method according to claim 5, characterized in that The step of inputting the target monitoring data sequence into the fully connected neural network model in an overlapping time window manner includes: Inputting the water inlet monitoring data and the process parameters in the target monitoring data sequence into the fully connected neural network model using overlapping time windows; The overlap amount of the overlapping time windows is y, wherein y<x.

7. The method according to any one of claims 1 to 6, characterized in that: Historical monitoring data include: influent chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, suspended solids, biochemical pool dissolved oxygen, biochemical pool suspended solids, biochemical pool aeration volume, transmembrane pressure difference, transmembrane flow; Influent monitoring data include: influent chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, and suspended solids; Process parameters include: dissolved oxygen in the biochemical pool, suspended solids in the biochemical pool, aeration volume in the biochemical pool, transmembrane pressure difference, and transmembrane flow rate; The effluent water quality parameters include: effluent chemical oxygen demand, ammonia nitrogen, total nitrogen, total phosphorus, and suspended solids.

8. A device for predicting effluent water quality parameters based on a fully connected neural network model, characterized in that: Applied to a sewage treatment plant, the device comprises: A real-time monitoring data acquisition module, used to acquire real-time monitoring data of the sewage treatment plant; An outlet water quality parameter output module is used to input the real-time monitoring data into a pre-built outlet water quality parameter prediction model and output the outlet water quality parameters at a preset time point; Among them, the pre-constructed effluent water quality parameter prediction model is obtained by inputting the target monitoring data sequence into a pre-constructed fully connected neural network model through overlapping time windows for training; the target monitoring data sequence is obtained after cleaning the monitoring data sequence through a pre-set data cleaning strategy; the monitoring data sequence includes multiple historical monitoring data obtained at preset time intervals.

9. An electronic device, characterized in that: It includes a processor and a memory, wherein the memory stores computer executable instructions that can be executed by the processor, and the processor executes the computer executable instructions to implement the method for predicting effluent water quality parameters based on a fully connected neural network model as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions. When the computer-executable instructions are called and executed by the processor, the computer-executable instructions prompt the processor to implement the method for predicting effluent water quality parameters based on a fully connected neural network model as described in any one of claims 1 to 7.

Images