Method and device for predicting daily fluid production capacity of oil well
By extracting the characteristic data of the water injection sequence using an autoencoder, the problem of limited data processing capacity and interference in the prediction of daily production of oil wells is solved, and the accurate prediction of daily production of oil wells is achieved.
Patent Information
- Application Number
- CN202310573863.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2023-05-19
- Publication Date
- 2024-11-22
AI Technical Summary
In the prior art, when predicting the daily fluid production of oil wells, neural network models are difficult to accurately capture useful features due to limited data processing capabilities and interfering with data, resulting in inaccurate prediction results.
The feature data of the water injection sequence is extracted using an autoencoder and input it into a pre-trained neural network model to improve feature capture capabilities and prediction accuracy.
By extracting characteristic data of the water injection sequence, neural network models can more easily capture useful features, thereby achieving accurate prediction of daily fluid production in the well.
Smart Images

Figure CN119005374A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of data processing, and in particular to a method and device for predicting the daily liquid production of an oil well. Background Art
[0002] In the early stage of oil field exploitation, the original formation pressure of the oil layer in the oil field can drive crude oil and natural gas, which in turn makes crude oil and natural gas spray to the surface pipeline through the oil well pipe string, thereby realizing the exploitation of crude oil and natural gas. However, after a period of production, due to the continuous decrease in the pressure inside the formation, the pressure inside the formation is not enough to drive crude oil and natural gas, so that the oil well no longer sprays oil and cannot continue to be exploited. In order to realize the continued exploitation of the oil well, some water injection wells are drilled at the edge of the oil field or the low part of the oil layer or the position between the oil wells, and water is injected into these water injection wells, so that the oil well continues to produce oil. It can be seen that the liquid production of the oil well is related to the water injection amount of the water injection well. Therefore, in order to know the liquid production of the oil well in advance, the liquid production of the oil well can be predicted based on the water injection amount of the water injection well, for example, the daily liquid production of each oil well on a certain day can be predicted.
[0003] In the related art, when it is necessary to predict the daily liquid production of each oil well on a certain day, it is necessary to first obtain the daily water injection volume of each water injection well for a preset number of consecutive days before the day, and then form these data into a data sequence according to a preset rule. After that, the data sequence is input into the pre-trained neural network model to obtain the daily liquid production of each oil well on a certain day.
[0004] In the above process, the data processing capability of the neural network model is limited, and there is certain interference data in the data sequence input to the neural network model, which makes it difficult for the neural network model to capture useful features in the data sequence, and thus cannot accurately predict the daily liquid production of the oil well. Summary of the invention
[0005] In view of this, the present application provides a method and device for predicting the daily liquid production of an oil well, and the present application can accurately predict the liquid production of an oil well.
[0006] In order to achieve the above objectives, this application mainly provides the following technical solutions:
[0007] In a first aspect, the present application provides a method for predicting daily fluid production of an oil well, the method comprising:
[0008] Get the daily water injection volume of each water injection well for N consecutive days before the prediction date;
[0009] Generate a water injection volume sequence based on the daily water injection volume of each water injection well N consecutive days before the prediction date;
[0010] Determining water injection amount characteristic data according to the water injection amount sequence and the pre-trained autoencoder;
[0011] The water injection volume characteristic data is input into a pre-trained neural network model to obtain a prediction sequence, wherein the prediction sequence includes the daily liquid production of each oil well on the prediction day.
[0012] In a second aspect, the present application provides a device for predicting daily liquid production of an oil well, the device comprising:
[0013] An acquisition unit is used to obtain the daily water injection volume of each water injection well for N consecutive days before the prediction date;
[0014] A generating unit, used for generating a water injection volume sequence according to the daily water injection volume of each water injection well N consecutive days before the prediction date;
[0015] A determination unit, used to determine water injection amount characteristic data according to the water injection amount sequence and a pre-trained autoencoder;
[0016] The input unit is used to input the water injection volume characteristic data into a pre-trained neural network model to obtain a prediction sequence, wherein the prediction sequence includes the daily liquid production of each oil well on the prediction day.
[0017] In a third aspect, the present application also provides an electronic device, comprising at least one processor, and at least one memory and a bus connected to the processor; wherein the processor and the memory communicate with each other via the bus; and the processor is used to call program instructions in the memory to execute the method for predicting daily liquid production of an oil well according to the first aspect above.
[0018] In a fourth aspect, the present application provides a storage medium for storing a computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute the method for predicting the daily liquid production of an oil well as described in the first aspect.
[0019] By means of the above technical scheme, the present application provides a method and device for predicting the daily liquid production of an oil well, obtaining the daily water injection volume of each water injection well N consecutive days before the prediction date; generating a water injection volume sequence according to the daily water injection volume of each water injection well N consecutive days before the prediction date; determining the water injection volume characteristic data according to the water injection volume sequence and a pre-trained autoencoder; inputting the water injection volume characteristic data into a pre-trained neural network model to obtain a prediction sequence, and the prediction sequence includes the daily liquid production of each oil well on the prediction date. In the present application, since the autoencoder is first used to extract the characteristic data of the water injection volume sequence, after the characteristic data is input into the neural network model, the useful features in the characteristic data can be more easily captured, thereby realizing accurate prediction of the daily liquid production of the oil well.
[0020] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0022] Figure 1 A schematic diagram of a process for predicting daily liquid production of an oil well disclosed in the present application;
[0023] Figure 2 A flowchart of a method for obtaining an autoencoder disclosed in this application;
[0024] Figure 3 A schematic diagram of a process for obtaining a neural network model disclosed in this application;
[0025] Figure 4 A schematic diagram of another method for predicting daily liquid production of an oil well disclosed in the present application;
[0026] Figure 5 This is a schematic diagram of the structure of a device for predicting daily liquid production of an oil well disclosed in the present application;
[0027] Figure 6 This is a schematic diagram of the structure of another device for predicting daily liquid production of an oil well disclosed in the present application;
[0028] Figure 7 A block diagram of a device disclosed in this application. DETAILED DESCRIPTION
[0029] The exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided in order to enable a more thorough understanding of the present application and to fully convey the scope of the present application to those skilled in the art.
[0030] In the early stage of oil field exploitation, the original formation pressure of the oil layer in the oil field can drive crude oil and natural gas, so that crude oil and natural gas can be sprayed into the surface pipeline through the oil well pipe string, thereby realizing the exploitation of crude oil and natural gas. However, after a period of production, due to the continuous decrease in the pressure inside the formation, the pressure inside the formation is not enough to drive crude oil and natural gas, so that the oil well no longer sprays oil and cannot continue to be exploited. In order to realize the continued exploitation of the oil well, some water injection wells are drilled at the edge of the oil field or the low part of the oil layer or the position between the oil wells, and water is injected into these water injection wells to enable the oil well to continue to produce oil. It can be seen that the liquid production of the oil well is related to the water injection amount of the water injection well. Therefore, in order to know the liquid production of the oil well in advance, the liquid production of the oil well can be predicted based on the water injection amount of the water injection well, for example, the daily liquid production of each oil well on a certain day can be predicted.
[0031] In the related art, when it is necessary to predict the daily liquid production of each oil well on a certain day, it is necessary to first obtain the daily water injection volume of each water injection well for a preset number of consecutive days before the day, and then form these data into a data sequence according to a preset rule. After that, the data sequence is input into the pre-trained neural network model to obtain the daily liquid production of each oil well on a certain day.
[0032] In the above process, the data processing capability of the neural network model is limited, and there is certain interference data in the data sequence input to the neural network model, which makes it difficult for the neural network model to capture useful features in the data sequence, and thus cannot accurately predict the daily liquid production of the oil well.
[0033] In order to solve the above problems, the embodiment of the present application provides a method for predicting the daily liquid production of an oil well, the execution subject of which is an electronic device, such as a smart computer, and the method can achieve accurate prediction of the daily liquid production of an oil well. The specific execution steps are as follows: Figure 1 As shown, including:
[0034] Step 101, obtaining the daily water injection volume of each water injection well for N consecutive days before the prediction date.
[0035] The daily water injection volume is the total water injection volume of the water injection well in one day. For example, the daily water injection volume of a water injection well is 1 ton. The water injection wells can be all the water injection wells in the oil field or only the water injection wells adjacent to the oil wells.
[0036] In a specific implementation of this step, when it is necessary to predict the daily liquid production of an oil well on a prediction day, it is necessary to first obtain the daily water injection volume of each water injection well for N consecutive days before the prediction day.
[0037] For example, let N be 10. When it is necessary to predict the daily liquid production of the oil well on February 30, 2023, it is necessary to first obtain the daily water injection volume of each water injection well from February 20, 2023 to February 29, 2023.
[0038] Step 102, generating a water injection volume sequence according to the daily water injection volume of each water injection well N consecutive days before the prediction date.
[0039] The water injection volume sequence includes water injection quantum sequences corresponding to N consecutive days, and each water injection quantum sequence includes the daily water injection volume of each injection well.
[0040] In a specific implementation of this step, for each of the N consecutive days before the forecast date, the water injection quantum sequence of the day can be formed according to the daily water injection volume of each water injection well on the day in a preset order. The water injection volume sequence is formed according to the water injection quantum sequence of each of the N consecutive days before the forecast date in a chronological order.
[0041] Step 103, determining the water injection amount characteristic data according to the water injection amount sequence and the pre-trained autoencoder.
[0042] In the actual processing process, in order to speed up the data processing process of the autoencoder, the water injection amount sequence can be normalized first, and then the normalized water injection amount sequence can be input into the pre-trained autoencoder, so that the pre-trained autoencoder outputs the water injection amount characteristic data. Of course, the water injection amount sequence can also be directly input into the pre-trained autoencoder to obtain the water injection amount characteristic data.
[0043] In addition, if during the training process of the autoencoder, the input data of the autoencoder is a water injection sequence, then during use, the input data of the autoencoder is also a water injection sequence. If during the training process of the autoencoder, the input data of the autoencoder is a normalized water injection sequence, then during use, the input data of the autoencoder is also a normalized water injection sequence.
[0044] Step 104, inputting the water injection volume characteristic data into the pre-trained neural network model to obtain a prediction sequence.
[0045] The prediction sequence includes the daily liquid production of each oil well on the prediction day. The pre-trained neural network model is used to predict the daily liquid production of each oil well on the prediction day.
[0046] The embodiment of the present application can obtain the daily water injection volume of each water injection well N consecutive days before the prediction date; generate a water injection volume sequence based on the daily water injection volume of each water injection well N consecutive days before the prediction date; determine the water injection volume feature data based on the water injection volume sequence and the pre-trained autoencoder; input the water injection volume feature data into the pre-trained neural network model to obtain a prediction sequence, and the prediction sequence includes the daily liquid production of each oil well on the prediction day. In the present application, since the autoencoder is first used to extract feature data from the water injection volume sequence, after the feature data is input into the neural network model, it is easier to capture useful features in the feature data, thereby achieving accurate prediction of the daily liquid production of the oil well.
[0047] Furthermore, before predicting the daily fluid production of an oil well, it is necessary to first obtain a pre-trained autoencoder, and then obtain the injection volume characteristic data based on the pre-trained autoencoder. Therefore, the embodiment of the present application also provides a method for obtaining a pre-trained autoencoder, such as Figure 2 The specific steps include:
[0048] Step 201: Obtain a first sample sequence in a training sample set.
[0049] The first sample sequence is generated according to the daily water injection volume of each water injection well for any N consecutive days in the historical period. The first sample sequence can be a sequence composed of the daily water injection volume of each water injection well for any N consecutive days in the historical period, or a sequence obtained by normalizing the sequence.
[0050] In a specific implementation of this step, the training sample set includes multiple sample sequences. In each process of training the autoencoder, a sample sequence is randomly selected from the training sample set and used as the first sample sequence.
[0051] Step 202: input the first sample sequence into the autoencoder to obtain first sample feature data.
[0052] In a specific implementation of this step, the first sample sequence is input into the autoencoder so that the autoencoder re-encodes the first sample sequence to obtain first sample feature data.
[0053] Step 203, input the first sample feature data into the decoder to obtain the first sample prediction sequence. The decoder is used to decode the feature data corresponding to the input data to obtain the input data again. The autoencoder is used to re-encode the input data to obtain the feature data corresponding to the input data. Both the autoencoder and the decoder are two-layer deep LSTM (Long Short-Term Memory) models. The activation function in the LSTM layer uses the tanh function, and the activation function for the gate mechanism uses the sigmoid function. The first neuron in each layer is initialized to the state Both are set to 0. The number of hidden layer units in the first LSTM network (layer1) of the encoder and the second LSTM network (layer4) of the decoder are set to 32 dimensions, and the number of hidden layer units in the second LSTM network (layer2) of the encoder and the first network (layer3) of the decoder are set to 16 dimensions.
[0054] In a specific implementation of this step, the first sample feature data is input into a decoder, so that the decoder decodes the first sample feature data to obtain a first sample prediction sequence.
[0055] Step 204: Based on the first sample sequence and the first sample prediction sequence, the autoencoder and the decoder are trained and adjusted to obtain a pre-trained autoencoder when the training condition meets the first preset condition.
[0056] Among them, the first preset condition may be that the number of training times reaches a preset number, or that the loss value of the first sample sequence and the first sample prediction sequence is less than a preset value, or other conditions, which are not limited here.
[0057] In a specific implementation of this step, the loss values of the first sample sequence and the first sample prediction sequence are calculated according to the loss function. Afterwards, the autoencoder and decoder are adjusted according to the loss value and the time-based backpropagation algorithm, and the weights in the model are updated using the gradient descent optimization method, thereby completing a training of the autoencoder and decoder. Afterwards, it is determined whether the training condition meets the first preset condition. If not, other sample sequences are selected from the training sample set, and the autoencoder and decoder obtained from the last training are trained and adjusted using the other samples until the training condition meets the first preset condition, thereby obtaining a pre-trained autoencoder and decoder.
[0058] Among them, the above loss function is Among them, x t is the data corresponding to the t-th day in the first sample sequence, xt′ is the data corresponding to the t-th day in the first sample prediction sequence, and N is the number of days.
[0059] Furthermore, before predicting the daily liquid production of the oil well, it is necessary to first obtain a pre-trained neural network model, and then obtain prediction data based on the pre-trained neural network model. Therefore, the embodiment of the present application also provides a method for obtaining a pre-trained network neural model. Figure 3 The specific steps include:
[0060] Step 301: Obtain a second sample sequence and its corresponding third sample sequence from a training sample set.
[0061] Among them, the second sample sequence is generated based on the daily water injection volume of each water injection well in any N consecutive days in the historical period, and the third sample sequence is composed of the daily liquid production of each oil well on the next day of any N consecutive days. For example, the second sample sequence is generated based on the daily water injection volume of each water injection well from February 20, 2022 to February 29, 2022. The third sample sequence corresponding to the second sample sequence is a sequence composed of the daily liquid production of each oil well on February 30, 2022.
[0062] In a specific implementation of this step, there are multiple sample sequence groups in the training sample set, and each sample sequence group includes a second sample sequence and a third sample sequence. Thus, when training the neural network model, a sample sequence group is randomly selected from the training sample set to obtain the second sample sequence and the third sample sequence.
[0063] In addition, this step provides a method for generating a training sample set, which is used to generate a training sample set. The specific steps include: obtaining the daily injection volume of each injection well in the historical period; screening and normalizing the daily injection volume data in the historical period to obtain processed data, and sorting the processed data in chronological order to obtain a data sequence; using a sliding time window, with a step size as a preset value, to divide the data sequence into time, to obtain multiple sample sequences, and to form a training sample set.
[0064] Among them, the length of the sliding time window is N.
[0065] In the implementation, the daily water injection volume of each water injection well in the historical period is used to form a data set, and the null values, abnormal values and duplicate values in the data set are removed, and the data in the data set are normalized to obtain a data set composed of normalized data. After that, the daily water injection volume data is composed of subsequences, and the daily subsequences are sorted in chronological order to obtain a data sequence. Using a sliding time window, the data sequence is divided into time with the step size as a preset value to obtain multiple sample sequences to form a training sample set.
[0066] Specifically, the above data series are divided into time series, and the sliding window method is used to cut the data, and the time window length is set to 1 (usually 32 or 64 is used in actual use), and the sliding step is 1. Assume that the data set is divided into M time series, each time series includes the water injection volume data of 1 day, then the divided data set is expressed as:
[0067] X 1 =(x 1 ,x 2 ,...,x l )
[0068] X 2 =(x 2 ,x 3 ,...,x l+1 )
[0069] X t =(x t ,x t+1 ,...,x l+t-1 )
[0070] …
[0071] X M =(x M ,x M+1 ,...,x T )
[0072] 75% of the above sequences are used as sample sequences in the training sample set, and 25% of the above sequences are used as sequences in the test set. The test set is used to test whether the output data of the pre-trained autoencoder and pre-trained neural network model are accurate.
[0073] Step 302: input the second sample sequence into a pre-trained autoencoder to obtain second sample feature data.
[0074] In a specific implementation of this step, the pre-trained autoencoder is based on Figure 2 Therefore, after obtaining the pre-trained autoencoder, the second sample sequence is input into the pre-trained autoencoder, and the second sample feature data is output.
[0075] Step 303: input the second sample feature data into the neural network model to obtain a second sample prediction sequence.
[0076] The second sample prediction sequence consists of the daily liquid production of each oil well on the prediction day.
[0077] Step 304: Based on the third sample sequence and the second sample prediction sequence, the neural network model is trained and adjusted to obtain a pre-trained neural network model when the training condition meets the second preset condition.
[0078] Among them, the second preset condition can be that the number of training times reaches a preset number, or that the loss value of the third sample sequence and the second sample prediction sequence is less than a preset value, or other conditions, which are not limited here.
[0079] In a specific implementation of this step, the loss value between the third sample sequence and the second sample prediction sequence is calculated according to the loss function, and the neural network model is trained and adjusted based on the loss value. Afterwards, it is determined whether the training condition meets the second preset condition. If not, the neural network model adjusted after the last training is further trained and adjusted according to other sample sequences in the training sample set. If yes, a pre-trained neural network model is obtained.
[0080] Furthermore, the present application provides a specific method for predicting the daily fluid production of an oil well, the method being as follows: Figure 4 As shown, the specific steps include:
[0081] Step 401, obtaining the daily water injection volume of each water injection well for N consecutive days before the prediction date.
[0082] Step 402, for each of the N consecutive days before the forecast date, the daily water injection volume of all water injection wells on that day is sorted according to a preset order to obtain the water injection quantum sequence of that day.
[0083] For example, the N consecutive days before the forecast date are from February 20, 2023 to February 29, 2023. The oil field includes the daily water injection volume A01 of injection well 01, the daily water injection volume A02 of injection well 02, the daily water injection volume A03 of injection well 03, and the daily water injection volume A04 of injection well 04. The preset order is A01, A02, A04, and A03. For each day from February 20, 2023 to February 29, 2023, the daily water injection volume A01 of injection well 01, the daily water injection volume A02 of injection well 02, the daily water injection volume A03 of injection well 03, and the daily water injection volume A04 of injection well 04 are obtained, and they are sorted according to the set order to obtain the water injection quantum sequence (A01, A02, A04, A03) of the day.
[0084] Step 403, combining the water injection quantum sequences corresponding to each day in chronological order to obtain a water injection volume sequence.
[0085] In the implementation, the water injection quantum sequence of each day in the N consecutive days before the forecast date is combined in chronological order to obtain the water injection volume sequence (X1, X2, X3...X N). Among them, X1 is the water injection quantum sequence of the first day among the N consecutive days before the prediction date, X2 is the water injection quantum sequence of the second day among the N consecutive days before the prediction date, and X3 is the water injection quantum sequence of the third day among the N consecutive days before the prediction date. N To predict the water injection quantum sequence of the Nth day among the N consecutive days before the day.
[0086] Step 404, normalizing the water injection volume sequence to obtain a normalized sequence.
[0087] Normalization refers to converting the data range to the range of 0 to 1. The purpose of normalization is to control the numerical range of the input vector, which cannot be too large or too small. Because the running of complex programs is already very time-consuming, if the value is too large, the running speed will be slower.
[0088] In a specific implementation of this step, the maximum value and the minimum value in the water injection volume sequence are determined, and each data in the water injection volume sequence is normalized according to the maximum value, the minimum value and the normalization formula to obtain a normalized sequence.
[0089] The normalization formula is: Among them, x max is the maximum value, x min is the minimum value, x is the data to be normalized in the injection volume sequence, x std is the normalized data.
[0090] Step 405: input the normalized sequence into a pre-trained autoencoder to obtain water injection volume feature data.
[0091] In a specific implementation of this step, the normalized sequence is input into a pre-trained autoencoder, so that the autoencoder re-encodes the normalized sequence to obtain water injection volume characteristic data.
[0092] Step 406, inputting the water injection volume characteristic data into the pre-trained neural network model to obtain a prediction sequence.
[0093] The prediction sequence includes the daily liquid production of each oil well on the prediction day.
[0094] The embodiment of the present application can obtain the daily water injection volume of each water injection well N consecutive days before the prediction date; generate a water injection volume sequence based on the daily water injection volume of each water injection well N consecutive days before the prediction date; determine the water injection volume feature data based on the water injection volume sequence and the pre-trained autoencoder; input the water injection volume feature data into the pre-trained neural network model to obtain a prediction sequence, and the prediction sequence includes the daily liquid production of each oil well on the prediction day. In the present application, since the autoencoder is first used to extract feature data from the water injection volume sequence, after the feature data is input into the neural network model, it is easier to capture useful features in the feature data, thereby achieving accurate prediction of the daily liquid production of the oil well.
[0095] Furthermore, as a response to the above Figure 1-4 The implementation of the method embodiment shown in the figure, the embodiment of the present application provides a device for predicting the daily liquid production of an oil well, which can accurately predict the daily liquid production of an oil well. The embodiment of the device corresponds to the aforementioned method embodiment. For ease of reading, this embodiment will no longer repeat the details of the aforementioned method embodiment one by one, but it should be clear that the device in this embodiment can correspond to all the contents of the aforementioned method embodiment. Specifically, Figure 5 As shown, the device comprises:
[0096] An acquisition unit 501 is used to acquire the daily water injection volume of each water injection well for N consecutive days before the prediction date;
[0097] A generating unit 502 is used to generate a water injection volume sequence according to the daily water injection volume of each water injection well N consecutive days before the prediction date;
[0098] A determination unit 503, configured to determine water injection amount characteristic data according to the water injection amount sequence and a pre-trained autoencoder;
[0099] The input unit 504 is used to input the water injection volume characteristic data into a pre-trained neural network model to obtain a prediction sequence, wherein the prediction sequence includes the daily liquid production of each oil well on the prediction day.
[0100] Optional, such as Figure 6 As shown, the generating unit 502 includes:
[0101] The sorting module 5021 is used to sort the daily water injection volume of all water injection wells on each day in the N days before the prediction date according to a preset order to obtain the water injection quantum sequence of the day;
[0102] The combination module 5022 is used to combine the water injection quantum sequences corresponding to each day in chronological order to obtain a water injection volume sequence.
[0103] Optional, such as Figure 6As shown, the determining unit 503 includes:
[0104] A normalization module 5031 is used to perform normalization processing on the water injection volume sequence to obtain a normalized sequence;
[0105] The first input module 5031 is used to input the normalized sequence into a pre-trained autoencoder to obtain water injection volume characteristic data.
[0106] Optional, such as Figure 6 As shown, the apparatus further includes a first training unit 505, comprising:
[0107] The first acquisition module 5051 is used to acquire a first sample sequence in the training sample set, where the first sample sequence is generated based on the daily water injection volume of each water injection well for any N consecutive days in a historical period;
[0108] A second input module 5052 is used to input the first sample sequence into the autoencoder to obtain first sample feature data;
[0109] A third input module 5053 is used to input the first sample feature data into a decoder to obtain a first sample prediction sequence;
[0110] The first training module 5054 is used to perform training adjustment on the autoencoder and the decoder based on the first sample sequence and the first sample prediction sequence, so as to obtain a pre-trained autoencoder when the training condition meets a first preset condition.
[0111] Optional, such as Figure 6 As shown, the apparatus further includes a second training unit 506, comprising:
[0112] The second acquisition module 5061 is used to acquire a second sample sequence and a corresponding third sample sequence from the training sample set, wherein the second sample sequence is generated based on the daily water injection volume of each water injection well for any N consecutive days in the historical period, and the third sample sequence is composed of the daily liquid production of each oil well on the next day after any N consecutive days;
[0113] A fourth input module 5062, configured to input the second sample sequence into a pre-trained autoencoder to obtain second sample feature data;
[0114] A fifth input module 5063, configured to input the second sample feature data into a neural network model to obtain a second sample prediction sequence, wherein the second sample prediction sequence consists of a predicted daily liquid production of each oil well;
[0115] The second training module 5064 is used to perform training adjustment on the neural network model based on the third sample sequence and the second sample prediction sequence, so as to obtain a pre-trained neural network model when the training condition meets the second preset condition.
[0116] Optional, such as Figure 6 As shown, the device further includes a generating unit 507, including:
[0117] The third acquisition module 5071 is used to obtain the daily water injection volume of each water injection well in the historical period;
[0118] The processing module 5072 is used to filter and normalize the daily water injection volume data in the historical period to obtain processed data, and sort the processed data in chronological order to obtain a data sequence;
[0119] The division module 5073 is used to use a sliding time window and a step size as a preset value to divide the data sequence into time to obtain multiple sample sequences to form the training sample set, wherein the length of the sliding time window is N.
[0120] Furthermore, an embodiment of the present application also provides an electronic device, the electronic device comprising at least one processor, and at least one memory and a bus connected to the processor; wherein the processor and the memory communicate with each other through the bus; the processor is used to call program instructions in the memory to execute the above Figure 1-4 The method for predicting the daily liquid production of an oil well as described in .
[0121] Furthermore, the embodiment of the present application also provides a storage medium, the storage medium is used to store a computer program, wherein when the computer program is running, the device where the storage medium is located is controlled to execute the above Figure 1-4 A method for predicting daily liquid production of oil wells.
[0122] Figure 7 70 is a block diagram of a device 70 provided in an embodiment of the present application. The device 70 includes at least one processor 701, and at least one memory 702 and a bus 703 connected to the processor 701; wherein the processor 701 and the memory 702 communicate with each other through the bus 703. The processor 701 is used to call the program instructions in the memory 702 to execute the above-mentioned method for predicting the daily liquid production of an oil well. The device in this article can be a server (for example: a local server or a cloud server), a smart phone, a tablet computer, a PDA, a portable computer, or a fixed terminal such as a desktop computer.
[0123] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.
[0124] It is understandable that the related features in the above methods and devices can be referenced to each other. In addition, the "first", "second" and the like in the above embodiments are used to distinguish the embodiments, but do not represent the advantages and disadvantages of the embodiments.
[0125] 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.
[0126] The algorithm and display provided herein are not inherently related to any particular computer, virtual system or other device. Various general purpose systems can also be used together with the teachings herein. According to the above description, it is obvious that the structure required for constructing such systems. In addition, the application is not directed to any specific programming language either. It should be understood that various programming languages can be utilized to realize the content of the application described herein, and the description of the specific language above is for the purpose of disclosing the best mode of implementation of the application.
[0127] In addition, the memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.
[0128] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0129] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0130] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.
[0131] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.
[0132] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.
[0133] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.
[0134] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.
[0135] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.
[0136] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.
Claims
1. A method for predicting daily liquid production of an oil well, characterized in that: The method comprises: Get the daily water injection volume of each water injection well for N consecutive days before the prediction date; Generate a water injection volume sequence based on the daily water injection volume of each water injection well N consecutive days before the prediction date; Determining water injection amount characteristic data according to the water injection amount sequence and the pre-trained autoencoder; The water injection volume characteristic data is input into a pre-trained neural network model to obtain a prediction sequence, wherein the prediction sequence includes the daily liquid production of each oil well on the prediction day.
2. The method according to claim 1, characterized in that The method generates a water injection volume sequence according to the daily water injection volume of each water injection well for N consecutive days before the prediction date, including: For each of the N consecutive days before the prediction date, the daily water injection volume of all water injection wells on that day is sorted according to a preset order to obtain a water injection quantum sequence for that day; The water injection quantum sequences corresponding to each day are combined in chronological order to obtain the water injection volume sequence.
3. The method according to claim 1, characterized in that Determining the water injection amount characteristic data according to the water injection amount sequence and the pre-trained autoencoder includes: Normalizing the water injection volume sequence to obtain a normalized sequence; The normalized sequence is input into a pre-trained autoencoder to obtain water injection volume feature data.
4. The method according to claim 1, characterized in that: The training process of the autoencoder is: Obtaining a first sample sequence in a training sample set, where the first sample sequence is generated based on the daily water injection volume of each water injection well for any N consecutive days in a historical period; Inputting the first sample sequence into the autoencoder to obtain first sample feature data; Inputting the first sample feature data into a decoder to obtain a first sample prediction sequence; Based on the first sample sequence and the first sample prediction sequence, the autoencoder and the decoder are trained and adjusted to obtain a pre-trained autoencoder when a training condition meets a first preset condition.
5. The method according to claim 4, characterized in that The training process of the neural network model is: In the training sample set, a second sample sequence and a corresponding third sample sequence are obtained, wherein the second sample sequence is generated according to the daily water injection volume of each water injection well for any N consecutive days in a historical period, and the third sample sequence is composed of the daily liquid production of each oil well on the next day after any N consecutive days; Inputting the second sample sequence into a pre-trained autoencoder to obtain second sample feature data; Inputting the second sample feature data into a neural network model to obtain a second sample prediction sequence, wherein the second sample prediction sequence consists of a predicted daily liquid production of each oil well; Based on the third sample sequence and the second sample prediction sequence, the neural network model is trained and adjusted to obtain a pre-trained neural network model when the training condition meets the second preset condition.
6. The method according to claim 4 or 5, characterized in that: The generation process of the training sample set is as follows: Get the daily water injection volume of each injection well during the historical period; Screening and normalizing the daily water injection volume data in the historical period to obtain processed data, and sorting the processed data in chronological order to obtain a data sequence; The data sequence is divided into time segments using a sliding time window with a step length as a preset value to obtain multiple sample sequences to form the training sample set, wherein the length of the sliding time window is N.
7. A device for predicting daily liquid production of an oil well, characterized in that: The device comprises: An acquisition unit is used to obtain the daily water injection volume of each water injection well for N consecutive days before the prediction date; A generating unit, used for generating a water injection volume sequence according to the daily water injection volume of each water injection well for N consecutive days before the prediction date; A determination unit, used to determine water injection amount characteristic data according to the water injection amount sequence and a pre-trained autoencoder; The input unit is used to input the water injection volume characteristic data into a pre-trained neural network model to obtain a prediction sequence, wherein the prediction sequence includes the daily liquid production of each oil well on the prediction day.
8. The device according to claim 7, characterized in that The generating unit comprises: A sorting module is used to sort the daily water injection volume of all water injection wells on each day of the N days before the prediction date according to a preset order to obtain a water injection quantum sequence for that day; The combination module is used to combine the water injection quantum sequences corresponding to each day in chronological order to obtain the water injection volume sequence.
9. An electronic device, characterized in that: The electronic device includes at least one processor, and at least one memory and a bus connected to the processor; wherein the processor and the memory communicate with each other through the bus; the processor is used to call program instructions in the memory to execute the method for predicting the daily liquid production of an oil well as described in any one of claims 1-6.
10. A readable storage medium, characterized in that: The storage medium is used to store a computer program, wherein when the computer program is running, it controls the device where the storage medium is located to execute the method for predicting the daily liquid production of an oil well according to any one of claims 1 to 6.
