CN118032062A - SAGD downhole temperature and pressure monitoring system and method based on artificial intelligence - Google Patents

Abstract

The application relates to the technical field of intelligent monitoring, and particularly discloses an artificial intelligence-based SAGD (steam assisted gravity drainage) underground temperature and pressure monitoring system and method. Therefore, automatic monitoring of underground temperature and pressure can be realized, manual intervention is reduced, monitoring efficiency is improved, early warning can be timely sent out when abnormal conditions are found, workers are helped to quickly respond, and the possibility of accidents is reduced.

Description

SAGD downhole temperature and pressure monitoring system and method based on artificial intelligence

Technical Field

The application relates to the technical field of intelligent monitoring, in particular to an artificial intelligence-based SAGD (steam assisted gravity drainage) downhole temperature and pressure monitoring system and method.

Background

SAGD (Steam-ASSISTED GRAVITY DRAINAGE) is an effective thickened oil recovery technique, and the mechanism is that Steam is injected into a Steam injection well, steam injected into the ground forms a Steam cavity around the Steam injection well, surrounding crude oil is heated to improve the fluidity of the oil, a part of the Steam is condensed into water, and the water and the hot oil flow downwards under the action of gravity and flow into a production well to be recovered.

In the SAGD technical field, the downhole temperature and pressure has an important influence on oil extraction, and reasonable control and monitoring of the downhole temperature and pressure are key to ensuring oil extraction efficiency and production stability. However, the conventional downhole temperature and pressure monitoring method mainly relies on sensors to collect data and identify abnormal conditions through manual analysis and judgment. The method requires professionals to analyze and read the data, and the efficiency and timeliness of data processing are low, so that the timely regulation and control and the capability of coping with emergency are affected. Especially when the monitoring points are more, the manual analysis has certain limitation, and the omission and the erroneous judgment are easy to occur.

Accordingly, an artificial intelligence based SAGD downhole temperature and pressure monitoring system and method is desired.

Disclosure of Invention

The present application has been made to solve the above-mentioned technical problems. The embodiment of the application provides an artificial intelligence based SAGD downhole temperature and pressure monitoring system and method, which monitor and analyze downhole temperature and pressure by adopting an artificial intelligence technology based on deep learning, and mine a time sequence association change rule between the downhole temperature and pressure so as to intelligently identify abnormal downhole temperature and pressure conditions. Therefore, automatic monitoring of underground temperature and pressure can be realized, manual intervention is reduced, monitoring efficiency is improved, early warning can be timely sent out when abnormal conditions are found, workers are helped to quickly respond, and the possibility of accidents is reduced.

Accordingly, in accordance with one aspect of the present application, there is provided an artificial intelligence based SAGD downhole temperature and pressure monitoring method comprising: acquiring a time sequence of downhole temperature values and a time sequence of downhole pressure values acquired by a temperature sensor and a pressure sensor; performing data preprocessing on the time sequence of the downhole temperature value and the time sequence of the downhole pressure value to obtain a sequence of downhole temperature-downhole pressure time sequence correlation vectors; performing multi-scale feature extraction on the sequence of downhole temperature-downhole pressure time sequence correlation vectors to obtain a sequence of downhole temperature-downhole pressure time sequence correlation feature vectors; inputting the sequence of downhole temperature-downhole pressure time sequence correlation feature vectors into a feature salizer based on an autocorrelation attention network to obtain a downhole temperature-downhole pressure time sequence feature vector; determining a monitoring result based on the downhole temperature-downhole pressure significant timing feature vector; wherein inputting the sequence of downhole temperature-downhole pressure timing related feature vectors into an autocorrelation-attention-network-based feature salizer to obtain a downhole temperature-downhole pressure saliency timing feature vector, comprising: processing the sequence of downhole temperature-downhole pressure timing correlation feature vectors in an autocorrelation attention formula to obtain the downhole temperature-downhole pressure significance timing feature vector; wherein the autocorrelation attention formula is: ; wherein/> For/>Score of attention, v-For/>Score of attention, v-A/>, in a sequence of time-series-correlated eigenvectors for the downhole temperature-downhole pressureCharacteristic vector related to downhole temperature and downhole pressure time sequence,/>For/>Characteristic vector related to downhole temperature and downhole pressure time sequence,/>Is a weight coefficient matrix,/>Is a weight coefficient vector,/>Is bias vector,/>Representing hyperbolic tangent function processing,/>Expressed as/>Exponential function processing for the base,/>For/>Attention weighting coefficient,/>For/>Attention weighting coefficient,/>For the length of the sequence of downhole temperature-downhole pressure timing correlation feature vectors,/>Representing a cascade function,/>Is the downhole temperature-downhole pressure significant timing feature vector.

In the above artificial intelligence-based SAGD downhole temperature and pressure monitoring method, performing data preprocessing on the time sequence of downhole temperature values and the time sequence of downhole pressure values to obtain a sequence of downhole temperature-downhole pressure time sequence correlation vectors, including: data normalization is carried out on the time sequence of the downhole temperature value and the time sequence of the downhole pressure value to obtain a downhole temperature time sequence input vector and a downhole pressure time sequence input vector; and performing association coding and data structure adjustment on the downhole temperature time sequence input vector and the downhole pressure time sequence input vector to obtain a sequence of the downhole temperature-downhole pressure time sequence association vector.

In the above-mentioned SAGD downhole temperature and pressure monitoring method based on artificial intelligence, the data normalization is performed on the time series of downhole temperature values and the time series of downhole pressure values to obtain a downhole temperature time sequence input vector and a downhole pressure time sequence input vector, including: and arranging the time sequence of the downhole temperature value and the time sequence of the downhole pressure value into the downhole temperature time sequence input vector and the downhole pressure time sequence input vector according to a time dimension respectively.

In the above-mentioned artificial intelligence-based SAGD downhole temperature and pressure monitoring method, performing association coding and data structure adjustment on the downhole temperature timing input vector and the downhole pressure timing input vector to obtain a sequence of the downhole temperature-downhole pressure timing association vector, including: calculating a sample covariance correlation matrix of the downhole temperature time sequence input vector relative to the downhole pressure time sequence input vector to obtain a downhole temperature-downhole pressure time sequence correlation matrix; and carrying out matrix splitting on the downhole temperature-downhole pressure time sequence correlation matrix to obtain a sequence of the downhole temperature-downhole pressure time sequence correlation vector.

In the above artificial intelligence-based SAGD downhole temperature and pressure monitoring method, calculating a sample covariance correlation matrix of the downhole temperature timing input vector relative to the downhole pressure timing input vector to obtain a downhole temperature-downhole pressure timing correlation matrix, comprising: calculating a sample covariance correlation matrix of the downhole temperature timing input vector relative to the downhole pressure timing input vector with a sample covariance correlation formula to obtain the downhole temperature-downhole pressure timing correlation matrix; the sample covariance correlation formula is as follows: ; wherein/> Inputting a vector for the downhole temperature timing,/>Transposed vector of input vector for downhole temperature timing,/>A vector is input for the downhole pressure timing,Transposed vector of the downhole pressure timing input vector,/>And (3) the downhole temperature-downhole pressure time sequence correlation matrix is obtained.

In the above artificial intelligence-based SAGD downhole temperature and pressure monitoring method, performing multi-scale feature extraction on the sequence of downhole temperature-downhole pressure time sequence correlation vectors to obtain a sequence of downhole temperature-downhole pressure time sequence correlation feature vectors, including: and the sequence of downhole temperature-downhole pressure time sequence correlation vectors is obtained through a time sequence correlation mode feature extractor based on a multi-scale neighborhood feature extraction network.

In the above artificial intelligence-based SAGD downhole temperature and pressure monitoring method, determining a monitoring result based on the downhole temperature-downhole pressure significant time sequence feature vector comprises: performing feature distribution optimization on the downhole temperature-downhole pressure significant time sequence feature vector to obtain an optimized downhole temperature-downhole pressure significant time sequence feature vector; and inputting the optimized downhole temperature-downhole pressure significant time sequence feature vector into a classifier-based monitoring result generator to obtain the monitoring result, wherein the monitoring result is used for indicating whether an abnormality exists.

In the above artificial intelligence-based SAGD downhole temperature and pressure monitoring method, inputting the optimized downhole temperature-downhole pressure significant timing feature vector into a classifier-based monitoring result generator to obtain the monitoring result, where the monitoring result is used to indicate whether an abnormality exists, and the method includes: performing full-connection coding on the optimized downhole temperature-downhole pressure significant time sequence feature vector by using a full-connection layer of the classifier to obtain a full-connection coding feature vector; inputting the fully-connected coding feature vector into a Softmax classification function of the classifier to obtain probability values of the optimized downhole temperature-downhole pressure significant time sequence feature vector belonging to various classification labels, wherein the classification labels comprise normal and abnormal; and determining the classification label corresponding to the maximum probability value as the classification result.

The artificial intelligence-based SAGD downhole temperature and pressure monitoring method further comprises the following steps: training the time sequence association mode feature extractor based on the multi-scale neighborhood feature extraction network, the feature saliency device based on the autocorrelation attention network and the monitoring result generator based on the classifier; the training step comprises the steps of obtaining training data, wherein the training data comprises the following steps: a time series of training downhole temperature values and a time series of training downhole pressure values; performing data preprocessing on the time sequence of the training downhole temperature value and the time sequence of the training downhole pressure value to obtain a sequence of training downhole temperature-downhole pressure time sequence correlation vectors; the sequence of the training downhole temperature-downhole pressure time sequence correlation vector is processed through a time sequence correlation mode feature extractor based on a multi-scale neighborhood feature extraction network to obtain the sequence of the training downhole temperature-downhole pressure time sequence correlation feature vector; inputting the sequence of the training downhole temperature-downhole pressure time sequence correlation feature vectors into a feature salizer based on an autocorrelation attention network to obtain training downhole temperature-downhole pressure time sequence feature vectors; performing feature distribution optimization on the training downhole temperature-downhole pressure significant time sequence feature vector to obtain a training optimized downhole temperature-downhole pressure significant time sequence feature vector; inputting the training optimized downhole temperature-downhole pressure significant time sequence feature vector into a classifier-based monitoring result generator to obtain a classification loss function value; training the time sequence association mode feature extractor based on the multi-scale neighborhood feature extraction network, the feature saliency device based on the autocorrelation attention network and the monitoring result generator based on the classifier by using the classification loss function value, wherein when the time sequence feature vector with the temperature-pressure saliency in the well after training optimization is subjected to classification iteration by the monitoring result generator based on the classifier, the time sequence feature vector with the temperature-pressure saliency in the well after training optimization is optimized.

In accordance with another aspect of the present application, there is provided an artificial intelligence based SAGD downhole temperature and pressure monitoring system comprising: the temperature and pressure data acquisition module is used for acquiring a time sequence of downhole temperature values and a time sequence of downhole pressure values acquired by the temperature sensor and the pressure sensor; the data preprocessing module is used for preprocessing data of the time sequence of the downhole temperature value and the time sequence of the downhole pressure value to obtain a sequence of downhole temperature-downhole pressure time sequence correlation vectors; the multi-scale feature extraction module is used for carrying out multi-scale feature extraction on the sequence of the downhole temperature-downhole pressure time sequence correlation vectors so as to obtain the sequence of the downhole temperature-downhole pressure time sequence correlation feature vectors; the characteristic salifying module is used for inputting the sequence of the downhole temperature-downhole pressure time sequence correlation characteristic vectors into a characteristic salizer based on an autocorrelation attention network so as to obtain a downhole temperature-downhole pressure time sequence characteristic vector; the monitoring result generation module is used for determining a monitoring result based on the downhole temperature-downhole pressure significant time sequence feature vector; wherein, the characteristic salifying module is used for: processing the sequence of downhole temperature-downhole pressure timing correlation feature vectors in an autocorrelation attention formula to obtain the downhole temperature-downhole pressure significance timing feature vector; wherein the autocorrelation attention formula is: ; wherein/> For/>Score of attention, v-For/>Score of attention, v-A/>, in a sequence of time-series-correlated eigenvectors for the downhole temperature-downhole pressureCharacteristic vector related to downhole temperature and downhole pressure time sequence,/>For/>Characteristic vector related to downhole temperature and downhole pressure time sequence,/>Is a weight coefficient matrix,/>Is a weight coefficient vector,/>Is bias vector,/>Representing hyperbolic tangent function processing,/>Expressed as/>Exponential function processing for the base,/>For/>Attention weighting coefficient,/>For/>Attention weighting coefficient,/>For the length of the sequence of downhole temperature-downhole pressure timing correlation feature vectors,/>Representing a cascade function,/>Is the downhole temperature-downhole pressure significant timing feature vector.

Compared with the prior art, the artificial intelligence-based SAGD downhole temperature and pressure monitoring system and method provided by the application monitor and analyze the downhole temperature and pressure by adopting an artificial intelligence technology based on deep learning, and mine a time sequence association change rule between the downhole temperature and pressure, so that abnormal downhole temperature and pressure conditions are intelligently identified. Therefore, automatic monitoring of underground temperature and pressure can be realized, manual intervention is reduced, monitoring efficiency is improved, early warning can be timely sent out when abnormal conditions are found, workers are helped to quickly respond, and the possibility of accidents is reduced.

Drawings

The above and other objects, features and advantages of the present application will become more apparent by describing embodiments of the present application in more detail with reference to the attached drawings. The accompanying drawings are included to provide a further understanding of embodiments of the application and are incorporated in and constitute a part of this specification, illustrate the application and together with the embodiments of the application, and not constitute a limitation to the application. In the drawings, like reference numerals generally refer to like parts or steps.

FIG. 1 is a flow chart of an artificial intelligence based SAGD downhole temperature and pressure monitoring method according to an embodiment of the present application.

FIG. 2 is a schematic diagram of an architecture of an artificial intelligence based SAGD downhole temperature and pressure monitoring method according to an embodiment of the present application.

FIG. 3 is a flow chart of a sequence of pre-processing data of the time series of downhole temperature values and the time series of downhole pressure values to obtain downhole temperature-downhole pressure timing correlation vectors in an artificial intelligence based SAGD downhole temperature and pressure monitoring method according to an embodiment of the present application.

FIG. 4 is a flow chart of a sequence of correlating encoding and data structure adjustment of the downhole temperature timing input vector and the downhole pressure timing input vector to obtain the downhole temperature-downhole pressure timing correlation vector in an artificial intelligence based SAGD downhole temperature and pressure monitoring method according to an embodiment of the present application.

FIG. 5 is a flow chart of determining a monitoring result based on the downhole temperature-downhole pressure salient timing feature vector in an artificial intelligence-based SAGD downhole temperature-pressure monitoring method according to an embodiment of the present application.

FIG. 6 is a flow chart of inputting the optimized downhole temperature-downhole pressure salient timing feature vector into a classifier-based monitoring result generator to obtain the monitoring result in an artificial intelligence-based SAGD downhole temperature-pressure monitoring method according to an embodiment of the present application.

FIG. 7 is a block diagram of an artificial intelligence based SAGD downhole temperature and pressure monitoring system according to an embodiment of the present application.

Detailed Description

For an understanding of embodiments of the present invention, specific embodiments of the invention will be described in more detail below with reference to the drawings. While specific embodiments of the invention are shown in the drawings, it should be understood that the invention may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

FIG. 1 is a flow chart of an artificial intelligence based SAGD downhole temperature and pressure monitoring method according to an embodiment of the present application. FIG. 2 is a schematic diagram of an architecture of an artificial intelligence based SAGD downhole temperature and pressure monitoring method according to an embodiment of the present application. As shown in fig. 1 and 2, the artificial intelligence-based SAGD downhole temperature and pressure monitoring method according to an embodiment of the present application includes the steps of: s110, acquiring a time sequence of downhole temperature values and a time sequence of downhole pressure values acquired by a temperature sensor and a pressure sensor; s120, carrying out data preprocessing on the time sequence of the downhole temperature value and the time sequence of the downhole pressure value to obtain a sequence of downhole temperature-downhole pressure time sequence correlation vectors; s130, performing multi-scale feature extraction on the sequence of the downhole temperature-downhole pressure time sequence correlation vectors to obtain a sequence of downhole temperature-downhole pressure time sequence correlation feature vectors; s140, inputting the sequence of downhole temperature-downhole pressure time sequence correlation feature vectors into a feature salizer based on an autocorrelation attention network to obtain downhole temperature-downhole pressure time sequence feature vectors; and S150, determining a monitoring result based on the downhole temperature-downhole pressure significant time sequence feature vector.

As described in the background art above, in the SAGD oil recovery process, the underground reservoir is heated by injecting high-temperature steam, so that the viscosity of crude oil is lowered and the fluidity is increased. That is, the downhole temperature directly affects the viscosity of the crude oil. If the temperature is too low, fluidity of crude oil is lowered, thereby affecting oil recovery efficiency, resulting in a decrease in yield. However, while an increase in temperature may increase the flowability of the crude oil, an excessively high temperature may cause volatilization of light components in the crude oil, which may change the physical properties of the crude oil, affecting the production and processing processes. Meanwhile, in the SAGD oil extraction process, the intensity of the steam thermal displacement effect, the stability and the permeability of an oil reservoir can be influenced by the underground pressure, and the oil extraction efficiency and the oil extraction yield are further influenced. In addition, excessive temperatures and pressures may cause damage to downhole equipment and tubing, while causing changes in the reservoir rock structure, resulting in reservoir damage, decreasing reservoir capacity and long-term production efficiency. Thus, real-time monitoring of downhole temperature and pressure is critical to ensuring production efficiency and yield during SAGD production.

However, the conventional downhole temperature and pressure monitoring method mainly relies on sensors to collect data and identify abnormal conditions through manual analysis and judgment. The method requires professionals to analyze and read the data, and the efficiency and timeliness of data processing are low, so that the timely regulation and control and the capability of coping with emergency are affected. Especially when the monitoring points are more, the manual analysis has certain limitation, and the omission and the erroneous judgment are easy to occur.

Aiming at the technical problems, the technical concept of the application is to monitor and analyze the underground temperature and pressure by adopting an artificial intelligence technology based on deep learning, and excavate the time sequence association change rule between the underground temperature and pressure, thereby intelligently identifying abnormal underground temperature and pressure conditions. Therefore, automatic monitoring of underground temperature and pressure can be realized, manual intervention is reduced, monitoring efficiency is improved, early warning can be timely sent out when abnormal conditions are found, workers are helped to quickly respond, and the possibility of accidents is reduced.

In the above-mentioned artificial intelligence-based SAGD downhole temperature and pressure monitoring method, the step S110 obtains a time sequence of downhole temperature values and a time sequence of downhole pressure values collected by a temperature sensor and a pressure sensor. It should be understood that temperature and pressure are the two most basic and most important monitoring parameters downhole. By acquiring the time sequence data of the temperature sensor and the pressure sensor, the underground dynamic change condition can be comprehensively known, and then the dynamic change rule of the temperature and the pressure, such as the change rate, the periodic change and the like of the temperature and the pressure, is excavated by utilizing an artificial intelligence technology, so that the abnormal condition is timely discovered and early warning treatment is carried out. That is, the time sequence of acquiring the downhole temperature value and the time sequence of acquiring the downhole pressure value are the basis for constructing the downhole temperature and pressure monitoring scheme, which is helpful for comprehensively and accurately monitoring the downhole situation, thereby realizing the intelligent abnormality detection and early warning functions.

In the above SAGD downhole temperature and pressure monitoring method based on artificial intelligence, in step S120, the time series of downhole temperature values and the time series of downhole pressure values are subjected to data preprocessing to obtain a sequence of downhole temperature-downhole pressure time sequence correlation vectors. Specifically, fig. 3 is a flowchart of a sequence of performing data preprocessing on the time series of downhole temperature values and the time series of downhole pressure values to obtain downhole temperature-downhole pressure timing correlation vectors in an artificial intelligence-based SAGD downhole temperature and pressure monitoring method according to an embodiment of the present application. As shown in fig. 3, the step S120 includes: s121, data normalization is carried out on the time sequence of the downhole temperature value and the time sequence of the downhole pressure value to obtain a downhole temperature time sequence input vector and a downhole pressure time sequence input vector; s122, performing association coding and data structure adjustment on the downhole temperature time sequence input vector and the downhole pressure time sequence input vector to obtain a sequence of the downhole temperature-downhole pressure time sequence association vector.

Specifically, in step S121, the time series of downhole temperature values and the time series of downhole pressure values are data normalized to obtain a downhole temperature time series input vector and a downhole pressure time series input vector. In a specific example of the present application, the data normalization is performed on the time series of downhole temperature values and the time series of downhole pressure values to obtain a downhole temperature time series input vector and a downhole pressure time series input vector, and the time series of downhole temperature values and the time series of downhole pressure values are respectively arranged into the downhole temperature time series input vector and the downhole pressure time series input vector according to a time dimension. It should be appreciated that considering that the downhole temperature value and the downhole pressure value are time-varying data, both have time-sequential dynamic variation characteristics in the time dimension. Therefore, in order to preserve the time sequence distribution characteristics of the data and facilitate the subsequent analysis of the time sequence dynamic change conditions of the underground temperature and the underground pressure, the time sequence of the underground temperature value and the time sequence of the underground pressure value are further arranged according to the time dimension respectively, so that the time sequence change information of the underground temperature value and the underground pressure value in the time dimension is integrated, the time sequence association change rule between the underground temperature and the underground pressure is learned, and the intelligent identification of the underground abnormal temperature and pressure condition is realized.

Specifically, in step S122, the downhole temperature time sequence input vector and the downhole pressure time sequence input vector are subjected to association coding and data structure adjustment to obtain the sequence of the downhole temperature-downhole pressure time sequence association vector. More specifically, FIG. 4 is a flow chart of a sequence of correlating encoding and data structure adjustments of the downhole temperature timing input vector and the downhole pressure timing input vector to obtain the downhole temperature-downhole pressure timing correlation vector in an artificial intelligence based SAGD downhole temperature pressure monitoring method according to an embodiment of the present application. As shown in fig. 4, the step S122 includes: s1221, calculating a sample covariance correlation matrix of the downhole temperature time sequence input vector relative to the downhole pressure time sequence input vector to obtain a downhole temperature-downhole pressure time sequence correlation matrix; s1222, performing matrix splitting on the downhole temperature-downhole pressure time sequence correlation matrix to obtain a sequence of downhole temperature-downhole pressure time sequence correlation vectors.

Specifically, in step S1221, a sample covariance correlation matrix of the downhole temperature timing input vector relative to the downhole pressure timing input vector is calculated to obtain a downhole temperature-downhole pressure timing correlation matrix. It should be appreciated that it is contemplated that the downhole temperatures and pressures are generally interrelated during SAGD oil recovery. The pressure is affected by the change in temperature and the change in pressure is reflected in temperature. Thus, to capture the timing correlation law between downhole temperature and downhole pressure, a sample covariance correlation matrix of the downhole temperature timing input vector relative to the downhole pressure timing input vector is further calculated. It should be understood that covariance is a statistic that measures the correlation between two random variables and is used to reflect the linear relationship and trend of variation between the variables. Specifically, if the eigenvalue of the sample covariance correlation matrix is a positive value, the eigenvalue and the eigenvalue of the sample covariance correlation matrix are positively correlated; if negative, it indicates that the two are inversely related; if 0, the two are independent. According to the technical scheme, the linear correlation between the underground temperature and the underground pressure can be revealed by calculating the sample covariance correlation matrix of the underground temperature time sequence input vector relative to the underground pressure time sequence input vector, and the information such as the correlation direction, the correlation strength, the correlation time delay and the like between the underground temperature and the underground pressure can be obtained, so that the time sequence correlation rule between the underground temperature and the underground pressure is captured, and a data basis is provided for further underground temperature and pressure state analysis.

In a specific example of the present application, the step S1221 includes: calculating a sample covariance correlation matrix of the downhole temperature timing input vector relative to the downhole pressure timing input vector with a sample covariance correlation formula to obtain the downhole temperature-downhole pressure timing correlation matrix; the sample covariance correlation formula is as follows: ; wherein/> Inputting a vector for the downhole temperature timing,/>Transposed vector of input vector for downhole temperature timing,/>Inputting a vector for the downhole pressure timing,/>Transposed vector of the downhole pressure timing input vector,/>And (3) the downhole temperature-downhole pressure time sequence correlation matrix is obtained.

Specifically, in step S1222, the downhole temperature-downhole pressure time sequence correlation matrix is matrix-split to obtain a sequence of the downhole temperature-downhole pressure time sequence correlation vectors. It should be appreciated that by splitting the downhole temperature-downhole pressure timing correlation matrix into a vector sequence to split covariance correlation features of downhole temperature and downhole pressure in the global time domain into correlation features in the local time domain, timing correlations between downhole temperature and downhole pressure can be more clearly demonstrated, helping to capture key patterns and trends in the local time series data more carefully.

In the above SAGD downhole temperature and pressure monitoring method based on artificial intelligence, in step S130, the sequence of downhole temperature-downhole pressure time sequence correlation vectors is subjected to multi-scale feature extraction to obtain the sequence of downhole temperature-downhole pressure time sequence correlation feature vectors. In a specific example of the present application, the multi-scale feature extraction is performed on the sequence of downhole temperature-downhole pressure time sequence correlation vectors to obtain a sequence of downhole temperature-downhole pressure time sequence correlation feature vectors, and the sequence of downhole temperature-downhole pressure time sequence correlation vectors is obtained by using a time sequence correlation mode feature extractor based on a multi-scale neighborhood feature extraction network. It should be appreciated that it is contemplated that there may be correlation features on different time scales in the sequence data of the downhole temperature-downhole pressure timing correlation vector. Thus, to capture timing correlation patterns of downhole temperature and pressure on different time scales, the sequence of downhole temperature-downhole pressure timing correlation vectors is further processed using a multi-scale neighborhood feature extraction network-based timing correlation pattern feature extractor. Those skilled in the art will appreciate that a multi-scale neighborhood feature extraction network is a feature extraction method used in deep learning, and is mainly used for capturing feature information under different scales so as to enhance the perceptibility of a model to input data of different scales. In the technical scheme of the application, the time sequence association mode feature extractor based on the multi-scale neighborhood feature extraction network can capture the local time sequence association relation of temperature and pressure on different time scales by applying a plurality of convolution cores of different scales to carry out sliding convolution operation on the downhole temperature-downhole pressure time sequence association vector, and find out the mode and rule hidden in the data, thereby enhancing the characterization capability of the time sequence association feature between the downhole temperature and pressure.

In the above-mentioned artificial intelligence-based SAGD downhole temperature and pressure monitoring method, in step S140, the sequence of the downhole temperature-downhole pressure time sequence correlation feature vectors is input into a feature salizer based on an autocorrelation attention network to obtain a downhole temperature-downhole pressure saliency time sequence feature vector. It should be appreciated that an autocorrelation Attention network, also known as Self-Attention mechanism (Self-Attention), is a method of implementing information autocorrelation in a neural network. In conventional neural networks, information is transferred layer by layer from an input layer, and each neuron can only receive information from the upper layer. The self-attention mechanism breaks this limitation, allowing each neuron to receive information from all levels simultaneously, thereby capturing the inherent links of the input data more effectively. In the technical scheme of the application, the characteristic salizer based on the autocorrelation attention network can carry out saliency weighting on each downhole temperature-downhole pressure time sequence correlation characteristic vector by calculating the correlation among sequences of the downhole temperature-downhole pressure time sequence correlation characteristic vectors, so that the network focuses on important time sequence correlation characteristics more and reduces interference of irrelevant information, thereby better understanding inherent correlation characteristics between downhole temperature and downhole pressure, improving characterization capability of the characteristics and providing more useful information for subsequent downhole temperature-pressure anomaly analysis.

In a specific example of the present application, the step S140 includes: processing the sequence of downhole temperature-downhole pressure timing correlation feature vectors in an autocorrelation attention formula to obtain the downhole temperature-downhole pressure significance timing feature vector; wherein the autocorrelation attention formula is: ; wherein, For/>Score of attention, v-For/>Score of attention, v-A/>, in a sequence of time-series-correlated eigenvectors for the downhole temperature-downhole pressureCharacteristic vector related to downhole temperature and downhole pressure time sequence,/>For/>Characteristic vector related to downhole temperature and downhole pressure time sequence,/>Is a weight coefficient matrix,/>Is a weight coefficient vector,/>As a result of the offset vector,Representing hyperbolic tangent function processing,/>Expressed as/>Exponential function processing for the base,/>For/>Attention weighting coefficient,/>For/>Attention weighting coefficient,/>For the length of the sequence of downhole temperature-downhole pressure timing correlation feature vectors,/>Representing a cascade function,/>Is the downhole temperature-downhole pressure significant timing feature vector.

In the above SAGD downhole temperature and pressure monitoring method based on artificial intelligence, in step S150, a monitoring result is determined based on the downhole temperature-downhole pressure significant time sequence feature vector. Specifically, fig. 5 is a flowchart of determining a monitoring result based on the downhole temperature-downhole pressure significant time sequence feature vector in the artificial intelligence-based SAGD downhole temperature-pressure monitoring method according to an embodiment of the present application. As shown in fig. 5, the step S150 includes: s151, performing feature distribution optimization on the downhole temperature-downhole pressure significant time sequence feature vector to obtain an optimized downhole temperature-downhole pressure significant time sequence feature vector; and S152, inputting the optimized downhole temperature-downhole pressure significant time sequence feature vector into a monitoring result generator based on a classifier to obtain the monitoring result, wherein the monitoring result is used for indicating whether an abnormality exists.

Specifically, in step S151, the feature distribution optimization is performed on the downhole temperature-downhole pressure significant timing feature vector to obtain an optimized downhole temperature-downhole pressure significant timing feature vector. In the technical solution of the present application, each downhole temperature-downhole pressure time series correlation feature vector in the sequence of downhole temperature-downhole pressure time series correlation feature vectors expresses a multi-scale local time series neighborhood high-order correlation feature of downhole temperature and downhole pressure in a local time series determined by matrix splitting, so that after the sequence of downhole temperature-downhole pressure time series correlation feature vectors is input into a feature salizer based on an autocorrelation attention network, the downhole temperature-downhole pressure salient time series feature vectors can perform local time series distribution self-reinforcement based on an overall high-order time series correlation feature distribution in each local time series, but this also makes the downhole temperature-downhole pressure salient time series feature vectors deviate from a local time series correlation feature distribution in a global time series while promoting the feature representation salience of time series correlation features in some local time series, thereby, it is desirable to optimize the downhole temperature-downhole time series salient feature vectors by further fusing the downhole temperature-downhole pressure salient time series feature vectors and the sequence of downhole temperature-downhole pressure correlation feature vectors.

Meanwhile, considering different feature representation modes based on local time domain autocorrelation attention enhancement of the sequence of downhole temperature-downhole pressure significant time sequence feature vectors and the sequence of downhole temperature-downhole pressure time sequence correlation feature vectors, it is desirable to promote the mapping effect of the sequences of downhole temperature-downhole pressure significant time sequence feature vectors and downhole temperature-downhole pressure time sequence correlation feature vectors to the fused feature distribution domain respectively.

Therefore, in the technical scheme of the application, the sequence of the downhole temperature-downhole pressure significant time sequence feature vector and the downhole temperature-downhole pressure time sequence correlation feature vector is optimally fused by the following optimal fusion formula so as to obtain an optimized downhole temperature-downhole pressure significant time sequence feature vector; wherein, the optimized fusion formula is: ; wherein/> Is the significant timing eigenvector of the downhole temperature-downhole pressure,/>Is a second feature vector obtained by cascading the sequence of downhole temperature-downhole pressure time sequence associated feature vectors, and the downhole temperature-downhole pressure significant time sequence feature vector/>And a second eigenvector/>Having the same length,/>And/>Significant timing feature vector/>, respectively, for the downhole temperature-downhole pressureMean and standard deviation of corresponding feature sets,/>And/>Second eigenvector/>, respectivelyMean and standard deviation of corresponding feature sets,/>Representing the position-by-position evolution of the feature vector, and/>Is a logarithm based on 2,/>Is the optimized downhole temperature-downhole pressure significant timing feature vector.

Here, in order to promote the mapping effect of the sequence of the downhole temperature-downhole pressure significant time sequence feature vector and the downhole temperature-downhole pressure time sequence associated feature vector to the fusion feature distribution domain under the feature fusion scene, on the basis that the traditional weighted fusion mode has limitation on deducing the semantic space evolution diffusion mode based on feature superposition, the fusion effect of the sequence of the downhole temperature-downhole pressure significant time sequence feature vector and the downhole temperature-downhole pressure time sequence associated feature vector is promoted by adopting a mode of combining a low-order superposition fusion mode and a high-order superposition fusion mode of a space and simulating the evolution center and the evolution track through the statistical feature interaction relation of the sequence of the downhole temperature-downhole pressure significant time sequence feature vector and the downhole temperature-downhole pressure time sequence associated feature vector, so that the semantic space evolution diffusion under the fusion scene is reconstructed based on the asynchronous evolution under the action of different evolution diffusion velocity fields, and the projection effect in the same high-dimensional feature space is effectively promoted. In this way, the optimized downhole temperature-downhole pressure significant time sequence characteristic vector is further obtainedBy classifying the obtained classification results by the classifier-based monitoring result generator, the accuracy of the obtained classification results can be improved.

Specifically, in step S152, the optimized downhole temperature-downhole pressure significant timing feature vector is input to a classifier-based monitoring result generator to obtain the monitoring result, where the monitoring result is used to indicate whether an anomaly exists. It should be appreciated that a classifier is a machine learning model for classifying sample data that is capable of accurately assigning input data points into preset categories through supervised training. In the technical scheme of the application, the monitoring result generator based on the classifier can classify the input optimized downhole temperature-downhole pressure significant time sequence feature vector by using a supervised learning method, so as to judge whether the feature mode in the optimized downhole temperature-downhole pressure significant time sequence feature vector accords with the normal condition. Therefore, the analysis result of the underground temperature and pressure condition can be rapidly and accurately generated through the monitoring result generator based on the classifier, so that related departments can take preventive measures in time, further deterioration of the problem is prevented, the monitoring efficiency is improved, and the labor cost is reduced.

FIG. 6 is a flow chart of inputting the optimized downhole temperature-downhole pressure salient timing feature vector into a classifier-based monitoring result generator to obtain the monitoring result in an artificial intelligence-based SAGD downhole temperature-pressure monitoring method according to an embodiment of the present application. As shown in fig. 6, the step S152 includes: s1521, performing full-connection coding on the optimized downhole temperature-downhole pressure significant time sequence feature vector by using a full-connection layer of the classifier to obtain a full-connection coding feature vector; s1522, inputting the fully-connected coding feature vector into a Softmax classification function of the classifier to obtain probability values of the optimized downhole temperature-downhole pressure significant time sequence feature vector belonging to various classification labels, wherein the classification labels comprise normal and abnormal; and S1523, determining the classification label corresponding to the maximum probability value as the classification result.

Further, the artificial intelligence-based SAGD downhole temperature and pressure monitoring method further comprises the following steps: training the time sequence association mode feature extractor based on the multi-scale neighborhood feature extraction network, the feature saliency device based on the autocorrelation attention network and the monitoring result generator based on the classifier; the training step comprises the steps of obtaining training data, wherein the training data comprises the following steps: a time series of training downhole temperature values and a time series of training downhole pressure values; performing data preprocessing on the time sequence of the training downhole temperature value and the time sequence of the training downhole pressure value to obtain a sequence of training downhole temperature-downhole pressure time sequence correlation vectors; the sequence of the training downhole temperature-downhole pressure time sequence correlation vector is processed through a time sequence correlation mode feature extractor based on a multi-scale neighborhood feature extraction network to obtain the sequence of the training downhole temperature-downhole pressure time sequence correlation feature vector; inputting the sequence of the training downhole temperature-downhole pressure time sequence correlation feature vectors into a feature salizer based on an autocorrelation attention network to obtain training downhole temperature-downhole pressure time sequence feature vectors; performing feature distribution optimization on the training downhole temperature-downhole pressure significant time sequence feature vector to obtain a training optimized downhole temperature-downhole pressure significant time sequence feature vector; inputting the training optimized downhole temperature-downhole pressure significant time sequence feature vector into a classifier-based monitoring result generator to obtain a classification loss function value; training the time sequence association mode feature extractor based on the multi-scale neighborhood feature extraction network, the feature saliency device based on the autocorrelation attention network and the monitoring result generator based on the classifier by using the classification loss function value, wherein when the time sequence feature vector with the temperature-pressure saliency in the well after training optimization is subjected to classification iteration by the monitoring result generator based on the classifier, the time sequence feature vector with the temperature-pressure saliency in the well after training optimization is optimized.

In the technical solution of the present application, each training downhole temperature-downhole pressure time-series correlation feature vector in the sequence of training downhole temperature-downhole pressure time-series correlation feature vectors expresses a multi-scale local time-series neighborhood higher order correlation feature of downhole temperature and downhole pressure in a local time-series neighborhood determined by matrix splitting, so that after the sequence of training downhole temperature-downhole pressure time-series correlation feature vectors is input into a feature salizer based on an autocorrelation attention network, the training downhole temperature-downhole pressure salient time-series feature vectors can perform local time-series distribution self-reinforcement based on an overall higher order time-series correlation feature distribution in each local time-series domain, but this also makes the training downhole temperature-downhole pressure salient time-series feature vectors deviate from the time-series correlation feature distribution of the local time-series temporal under the global time-series, thereby, it is desirable to optimize the training downhole temperature salient feature vectors by further fusing the training downhole temperature-downhole pressure salient time-series feature vectors and the training downhole temperature-downhole pressure time-series correlation feature vectors.

Meanwhile, considering different characteristic representation modes based on local time domain autocorrelation attention enhancement of the training downhole temperature-downhole pressure significant time sequence characteristic vector and the sequence of the training downhole temperature-downhole pressure time sequence correlation characteristic vector, the fused training optimized downhole temperature-downhole pressure significant time sequence characteristic vector also has discretized local characteristic distribution, so that the convergence effect of the training optimized downhole temperature-downhole pressure significant time sequence characteristic vector to a quasi probability density space is affected when the training optimized downhole temperature-downhole pressure significant time sequence characteristic vector is classified by a monitoring result generator based on a classifier. Therefore, the application optimizes the training optimized downhole temperature-downhole pressure significant time sequence feature vector when the training optimized downhole temperature-downhole pressure significant time sequence feature vector is subjected to classification iteration through a classifier-based monitoring result generator.

Accordingly, in one example, optimizing the trained post-optimization downhole temperature-downhole pressure salient timing feature vector each time the trained post-optimization downhole temperature-downhole pressure salient timing feature vector is classified and iterated by a classifier-based monitoring result generator includes: optimizing the training optimized downhole temperature-downhole pressure significant time sequence feature vector by the following optimization formula, wherein the optimization formula is as follows: ; wherein/> Representing the significant time sequence characteristic vector of the downhole temperature and the downhole pressure after training and optimizingAnd/>The time sequence characteristic vector/>, respectively, of the temperature-pressure significance in the underground well after the training optimization(1 /)And/>Characteristic value of location,/>Representing a first intermediate matrix,/>Representing a second intermediate matrix,/>Representing/> -of the first intermediate matrixCharacteristic value of location,/>Representing/> -of the second intermediate matrixCharacteristic value of location,/>Representing addition by location,/>Representing matrix multiplication,/>And representing the optimized training, optimized and obvious time sequence characteristic vector of the downhole temperature and the downhole pressure.

That is, the post-training optimized downhole temperature-downhole pressure salient timing feature vector is introducedIs used as an external information source to carry out retrieval enhancement of feature vectors so as to avoid the phenomenon that the local overflow information distribution leads to the training optimized downhole temperature-downhole pressure significant time sequence feature vector/>, based on local statistics intensive information structuringThereby obtaining the training optimized downhole temperature-downhole pressure significant timing feature vector/>Information credible response reasoning based on local distribution group dimension retention to obtain a downhole temperature-downhole pressure significant time sequence characteristic vector/>, after training optimizationTrusted distribution response in probability density space based on discretized local feature distribution, thereby improving probability density space convergence effect to improve downhole temperature-downhole pressure significant time sequence feature vector after training optimizationThe training speed and the accuracy of the training result of the classification regression training are carried out by a monitoring result generator based on the classifier.

In summary, the SAGD downhole temperature and pressure monitoring method based on artificial intelligence according to the embodiment of the present application is illustrated, which adopts artificial intelligence technology based on deep learning to monitor and analyze downhole temperature and pressure, and digs out time sequence association change rules between downhole temperature and pressure, so as to intelligently identify abnormal downhole temperature and pressure conditions. Therefore, automatic monitoring of underground temperature and pressure can be realized, manual intervention is reduced, monitoring efficiency is improved, early warning can be timely sent out when abnormal conditions are found, workers are helped to quickly respond, and the possibility of accidents is reduced.

FIG. 7 is a block diagram of an artificial intelligence based SAGD downhole temperature and pressure monitoring system according to an embodiment of the present application. As shown in fig. 7, an artificial intelligence based SAGD downhole temperature and pressure monitoring system 100 according to an embodiment of the present application comprises: a temperature and pressure data acquisition module 110 for acquiring a time series of downhole temperature values and a time series of downhole pressure values acquired by the temperature sensor and the pressure sensor; a data preprocessing module 120, configured to perform data preprocessing on the time sequence of downhole temperature values and the time sequence of downhole pressure values to obtain a sequence of downhole temperature-downhole pressure time sequence correlation vectors; a multi-scale feature extraction module 130, configured to perform multi-scale feature extraction on the sequence of downhole temperature-downhole pressure timing related vectors to obtain a sequence of downhole temperature-downhole pressure timing related feature vectors; a feature saliency module 140 for inputting the sequence of downhole temperature-downhole pressure time series correlation feature vectors into an autocorrelation-attention-network-based feature saliency to obtain a downhole temperature-downhole pressure saliency time series feature vector; the monitoring result generating module 150 is configured to determine a monitoring result based on the downhole temperature-downhole pressure significant timing feature vector.

Here, it will be understood by those skilled in the art that the specific operations of the respective modules in the above-described artificial intelligence-based SAGD downhole temperature and pressure monitoring system have been described in detail in the above description of the artificial intelligence-based SAGD downhole temperature and pressure monitoring method with reference to fig. 1 to 6, and thus, repetitive descriptions thereof will be omitted.

The basic principles of the present invention have been described above in connection with specific embodiments, but it should be noted that the advantages, benefits, effects, etc. mentioned in the present invention are merely examples and not intended to be limiting, and these advantages, benefits, effects, etc. are not to be construed as necessarily possessed by the various embodiments of the invention. Furthermore, the particular details of the above-described embodiments are for purposes of illustration and understanding only, and are not intended to limit the invention to the particular details described above, but are not necessarily employed.

In the foregoing embodiments, the descriptions of the embodiments are emphasized, and in part, not described or illustrated in any particular embodiment, reference is made to the related descriptions of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods may be implemented in other ways. For example, the system embodiments described above are merely illustrative, and for example, the module division is merely a logical function division, and other manners of division may be implemented in practice. The modules described as separate components may or may not be physically separate, and components shown as modules may or may not be physical units, may be located in one place, or may be distributed over multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

It will be evident to those skilled in the art that the invention is not limited to the details of the foregoing illustrative embodiments, and that the present invention may be embodied in other specific forms without departing from the spirit or essential characteristics thereof. The present embodiments are, therefore, to be considered in all respects as illustrative and not restrictive, the scope of the invention being indicated by the appended claims rather than by the foregoing description, and all changes which come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein. Any reference signs in the claims shall not be construed as limiting the claim concerned.

Furthermore, it is evident that the word "comprising" does not exclude other elements or steps, and that the singular does not exclude a plurality. The units recited in the system claims may also be implemented by means of software or hardware.

Finally, it should be noted that the foregoing description has been presented for the purposes of illustration and description. Furthermore, the foregoing embodiments are merely for illustrating the technical scheme of the present invention, and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it will be understood by those skilled in the art that modifications and equivalents may be made thereto without departing from the spirit and scope of the technical scheme of the present invention.

Claims (9)

1. An artificial intelligence-based SAGD downhole temperature and pressure monitoring method is characterized by comprising the following steps: acquiring a time sequence of downhole temperature values and a time sequence of downhole pressure values acquired by a temperature sensor and a pressure sensor; performing data preprocessing on the time sequence of the downhole temperature value and the time sequence of the downhole pressure value to obtain a sequence of downhole temperature-downhole pressure time sequence correlation vectors; performing multi-scale feature extraction on the sequence of downhole temperature-downhole pressure time sequence correlation vectors to obtain a sequence of downhole temperature-downhole pressure time sequence correlation feature vectors; inputting the sequence of downhole temperature-downhole pressure time sequence correlation feature vectors into a feature salizer based on an autocorrelation attention network to obtain a downhole temperature-downhole pressure time sequence feature vector; determining a monitoring result based on the downhole temperature-downhole pressure significant timing feature vector; wherein inputting the sequence of downhole temperature-downhole pressure timing related feature vectors into an autocorrelation-attention-network-based feature salizer to obtain a downhole temperature-downhole pressure saliency timing feature vector, comprising: processing the sequence of downhole temperature-downhole pressure timing correlation feature vectors in an autocorrelation attention formula to obtain the downhole temperature-downhole pressure significance timing feature vector; wherein the autocorrelation attention formula is: ; wherein/> For/>Score of attention, v-For/>Score of attention, v-A/>, in a sequence of time-series-correlated eigenvectors for the downhole temperature-downhole pressureCharacteristic vector related to downhole temperature and downhole pressure time sequence,/>For/>Characteristic vector related to downhole temperature and downhole pressure time sequence,/>Is a weight coefficient matrix,/>Is a weight coefficient vector,/>Is bias vector,/>Representing hyperbolic tangent function processing,/>Expressed as/>Exponential function processing for the base,/>For/>Attention weighting coefficient,/>For/>Attention weighting coefficient,/>For the length of the sequence of downhole temperature-downhole pressure timing related feature vectors,Representing a cascade function,/>Is the downhole temperature-downhole pressure significant timing feature vector.

2. The artificial intelligence based SAGD downhole temperature and pressure monitoring method according to claim 1, wherein data preprocessing the time series of downhole temperature values and the time series of downhole pressure values to obtain a sequence of downhole temperature-downhole pressure timing correlation vectors comprises: data normalization is carried out on the time sequence of the downhole temperature value and the time sequence of the downhole pressure value to obtain a downhole temperature time sequence input vector and a downhole pressure time sequence input vector; and performing association coding and data structure adjustment on the downhole temperature time sequence input vector and the downhole pressure time sequence input vector to obtain a sequence of the downhole temperature-downhole pressure time sequence association vector.

3. The artificial intelligence based SAGD downhole temperature and pressure monitoring method according to claim 2, wherein correlating the downhole temperature timing input vector and the downhole pressure timing input vector and data structure adjustments to obtain the sequence of downhole temperature-downhole pressure timing correlation vectors comprises: calculating a sample covariance correlation matrix of the downhole temperature time sequence input vector relative to the downhole pressure time sequence input vector to obtain a downhole temperature-downhole pressure time sequence correlation matrix; and carrying out matrix splitting on the downhole temperature-downhole pressure time sequence correlation matrix to obtain a sequence of the downhole temperature-downhole pressure time sequence correlation vector.

4. The artificial intelligence based SAGD downhole temperature and pressure monitoring method according to claim 3, wherein calculating a sample covariance correlation matrix of the downhole temperature timing input vector relative to the downhole pressure timing input vector to obtain a downhole temperature-downhole pressure timing correlation matrix comprises: calculating a sample covariance correlation matrix of the downhole temperature timing input vector relative to the downhole pressure timing input vector with a sample covariance correlation formula to obtain the downhole temperature-downhole pressure timing correlation matrix; the sample covariance correlation formula is as follows: ; wherein, Inputting a vector for the downhole temperature timing,/>Transposed vector of input vector for downhole temperature timing,/>Inputting a vector for the downhole pressure timing,/>Transposed vector of the downhole pressure timing input vector,/>And (3) the downhole temperature-downhole pressure time sequence correlation matrix is obtained.

5. The artificial intelligence based SAGD downhole temperature and pressure monitoring method as claimed in claim 4, wherein the multi-scale feature extraction of the sequence of downhole temperature-downhole pressure timing correlation vectors to obtain the sequence of downhole temperature-downhole pressure timing correlation feature vectors comprises: and the sequence of downhole temperature-downhole pressure time sequence correlation vectors is obtained through a time sequence correlation mode feature extractor based on a multi-scale neighborhood feature extraction network.

6. The artificial intelligence based SAGD downhole temperature and pressure monitoring method as claimed in claim 5, wherein determining the monitoring result based on the downhole temperature-downhole pressure salient timing feature vector comprises: performing feature distribution optimization on the downhole temperature-downhole pressure significant time sequence feature vector to obtain an optimized downhole temperature-downhole pressure significant time sequence feature vector; and inputting the optimized downhole temperature-downhole pressure significant time sequence feature vector into a classifier-based monitoring result generator to obtain the monitoring result, wherein the monitoring result is used for indicating whether an abnormality exists.

7. The artificial intelligence based SAGD downhole temperature and pressure monitoring method as claimed in claim 6, wherein the optimized downhole temperature-downhole pressure salient timing feature vector is input to a classifier based monitoring result generator to obtain the monitoring result, the monitoring result is used for indicating whether an abnormality exists, and the method comprises: performing full-connection coding on the optimized downhole temperature-downhole pressure significant time sequence feature vector by using a full-connection layer of the classifier to obtain a full-connection coding feature vector; inputting the fully-connected coding feature vector into a Softmax classification function of the classifier to obtain probability values of the optimized downhole temperature-downhole pressure significant time sequence feature vector belonging to various classification labels, wherein the classification labels comprise normal and abnormal; and determining the classification label corresponding to the maximum probability value as the classification result.

8. The artificial intelligence based SAGD downhole temperature and pressure monitoring method as claimed in claim 7, further comprising: training the time sequence association mode feature extractor based on the multi-scale neighborhood feature extraction network, the feature saliency device based on the autocorrelation attention network and the monitoring result generator based on the classifier; wherein the training step comprises: acquiring training data, the training data comprising: a time series of training downhole temperature values and a time series of training downhole pressure values; performing data preprocessing on the time sequence of the training downhole temperature value and the time sequence of the training downhole pressure value to obtain a sequence of training downhole temperature-downhole pressure time sequence correlation vectors; the sequence of the training downhole temperature-downhole pressure time sequence correlation vector is processed through a time sequence correlation mode feature extractor based on a multi-scale neighborhood feature extraction network to obtain the sequence of the training downhole temperature-downhole pressure time sequence correlation feature vector; inputting the sequence of the training downhole temperature-downhole pressure time sequence correlation feature vectors into a feature salizer based on an autocorrelation attention network to obtain training downhole temperature-downhole pressure time sequence feature vectors; performing feature distribution optimization on the training downhole temperature-downhole pressure significant time sequence feature vector to obtain a training optimized downhole temperature-downhole pressure significant time sequence feature vector; inputting the training optimized downhole temperature-downhole pressure significant time sequence feature vector into a classifier-based monitoring result generator to obtain a classification loss function value; training the time sequence association mode feature extractor based on the multi-scale neighborhood feature extraction network, the feature saliency device based on the autocorrelation attention network and the monitoring result generator based on the classifier by using the classification loss function value, wherein when the time sequence feature vector with the temperature-pressure saliency in the well after training optimization is subjected to classification iteration by the monitoring result generator based on the classifier, the time sequence feature vector with the temperature-pressure saliency in the well after training optimization is optimized.

9. An artificial intelligence based SAGD downhole temperature and pressure monitoring system, comprising: the temperature and pressure data acquisition module is used for acquiring a time sequence of downhole temperature values and a time sequence of downhole pressure values acquired by the temperature sensor and the pressure sensor; the data preprocessing module is used for preprocessing data of the time sequence of the downhole temperature value and the time sequence of the downhole pressure value to obtain a sequence of downhole temperature-downhole pressure time sequence correlation vectors; the multi-scale feature extraction module is used for carrying out multi-scale feature extraction on the sequence of the downhole temperature-downhole pressure time sequence correlation vectors so as to obtain the sequence of the downhole temperature-downhole pressure time sequence correlation feature vectors; the characteristic salifying module is used for inputting the sequence of the downhole temperature-downhole pressure time sequence correlation characteristic vectors into a characteristic salizer based on an autocorrelation attention network so as to obtain a downhole temperature-downhole pressure time sequence characteristic vector; the monitoring result generation module is used for determining a monitoring result based on the downhole temperature-downhole pressure significant time sequence feature vector; wherein, the characteristic salifying module is used for: processing the sequence of downhole temperature-downhole pressure timing correlation feature vectors in an autocorrelation attention formula to obtain the downhole temperature-downhole pressure significance timing feature vector; wherein the autocorrelation attention formula is: ; wherein/> For/>Score of attention, v-For/>Score of attention, v-A/>, in a sequence of time-series-correlated eigenvectors for the downhole temperature-downhole pressureCharacteristic vector related to downhole temperature and downhole pressure time sequence,/>For/>Characteristic vector related to downhole temperature and downhole pressure time sequence,/>Is a weight coefficient matrix,/>Is a weight coefficient vector,/>As a result of the offset vector,Representing hyperbolic tangent function processing,/>Expressed as/>Exponential function processing for the base,/>For/>Attention weighting coefficient,/>For/>Attention weighting coefficient,/>For the length of the sequence of downhole temperature-downhole pressure timing correlation feature vectors,/>Representing a cascade function,/>Is the downhole temperature-downhole pressure significant timing feature vector.