A geothermal well temperature sequence prediction method

Abstract

The application discloses a geothermal well temperature sequence prediction method, belongs to the technical field of geothermal energy development, and can solve the problem that the existing prediction method is single in seasonal modeling and is difficult to process multi-period superposition effect. The method comprises the following steps: S1, constructing a daily cycle term and an annual cycle term of the geothermal well temperature, and determining a seasonal term of the geothermal well temperature according to the daily cycle term and the annual cycle term; S2, constructing an autoregressive term, a moving average term and an external variable term of the geothermal well temperature, and constructing a regression model of the geothermal well temperature according to the autoregressive term, the moving average term, the external variable term and the seasonal term; S3, training the regression model by using a historical temperature sequence and a historical external variable sequence of the geothermal well, and obtaining a prediction model; and S4, inputting a current temperature sequence and a current external variable sequence of the geothermal well into the prediction model to obtain a temperature prediction value of the geothermal well at a next moment. The application is used for the prediction of the geothermal well temperature.

Description

Technical Field

[0001] This invention relates to a method for predicting geothermal well temperature sequences, belonging to the field of geothermal energy development technology. Background Technology

[0002] In the process of geothermal energy development and utilization, wellbore temperature is an important indicator for judging the energy status of the reservoir, optimizing the mining plan, controlling the water flow rate and power generation / heating efficiency. Accurate prediction of well temperature changes can not only guide production scheduling, but also assist in the power adjustment of fans (pumps / heat pumps), the selection of downhole equipment and maintenance plans. Existing geothermal well temperature prediction methods are mainly divided into three categories: numerical simulation based on physical mechanisms, time series model based methods and pure data-driven machine learning methods. Specifically, (1) physical mechanism model: by establishing the heat transfer-seepage coupling equation of the geothermal reservoir, combined with geological parameters (such as rock thermal conductivity and permeability) to simulate the dynamics of wellbore temperature. For example, the heat conduction equation is solved by the finite element method, or the deep geothermal field is predicted by combining magnetotelluric inversion data. (2) time series model: ARMA / ARIMA model is used to describe the autocorrelation and trend of temperature series. Some studies introduce a single seasonal term (such as annual cycle) or external variable (such as flow rate), but the lag order of external variables is fixed and the number is limited. (3) Data-driven machine learning models use algorithms such as LSTM and random forest to learn the nonlinear relationship between historical temperature and influencing factors. Some studies introduce wavelet decomposition to preprocess high-frequency noise. In the process of use, pure data-driven LSTM predicts geothermal production capacity and ignores physical boundary constraints.

[0003] However, the above prediction methods usually have the following problems: (1) The seasonal modeling is too simple and it is difficult to handle the superimposed effects of multiple cycles. Existing models usually only consider a single cycle (such as an annual cycle or a daily cycle), while geothermal well temperature is affected by both diurnal production scheduling (24h) and annual climate fluctuations (8760h). The single model cannot capture the multi-scale cycle coupling effect, resulting in significant fitting bias. (2) Insufficient use of external variables and rigid lag mechanism. Existing ARMAX models only introduce 1-2 external variables (such as flow rate or air temperature), and the lag order is fixed. In fact, well temperature is driven by multiple factors (pumping flow rate, well pressure, surface air temperature, power generation load), and the time delay of each factor is different (such as air temperature lags by 6h, load lags by 0h). Ignoring the dynamic lag relationship between variables weakens the physical interpretability of the model. (3) Static parameters cannot adapt to dynamic geological and equipment conditions. Traditional models use fixed parameters or fixed time windows for updates, without considering the performance drift caused by formation permeability decay, equipment scaling, etc. For example, the hot washing cycle of oil wells needs to be dynamically adjusted due to the decrease in formation pressure, but the static model cannot respond in real time. (4) Anomaly detection depends on thresholds and has poor robustness. Existing methods (such as wavelet threshold denoising) require preset fixed thresholds, which cannot adapt to the dynamic changes in data distribution. Geothermal well data often produce outliers due to sensor failures or short-term geological events (such as fracture blockage). Fixed thresholds are prone to misjudgment or missed judgment, which contaminates the training data. Summary of the Invention

[0004] This invention provides a method for predicting geothermal well temperature sequences, which can solve the problem that existing prediction methods have a single seasonal model and are difficult to handle multi-period superposition effects.

[0005] This invention provides a method for predicting geothermal well temperature sequences, the method comprising:

[0006] S1. Construct the daily and annual cycle terms of geothermal well temperature, and determine the seasonality term of geothermal well temperature based on the daily and annual cycle terms.

[0007] S2. Construct the autoregressive term, moving average term, and external variable term for geothermal well temperature, and construct a regression model for geothermal well temperature based on the autoregressive term, the moving average term, the external variable term, and the seasonality term;

[0008] S3. Train the regression model using the historical temperature sequence and historical external variable sequence of geothermal wells to obtain the prediction model;

[0009] S4. Input the current temperature sequence and the current external variable sequence of the geothermal well into the prediction model to obtain the predicted temperature value of the geothermal well at the next moment.

[0010] Optionally, the external variable for constructing the geothermal well temperature in S2 specifically includes:

[0011] Identify the multiple external variables affecting the temperature of geothermal wells and determine the lag order of each external variable;

[0012] Based on multiple external variables and their lag orders, an external variable term for geothermal well temperature is constructed.

[0013] Optionally, the objective function of the prediction model is a weighted least squares objective function; the weights in the objective function are exponentially decaying weights.

[0014] Optionally, after S3, the method further includes:

[0015] A time rolling window is determined, and the prediction model is used to determine the predicted temperature sequence of the geothermal well within the time rolling window;

[0016] The predicted temperature sequence is used to update the historical temperature sequence within the time rolling window to obtain the updated temperature sequence. The prediction model is then trained using the updated temperature sequence and the historical external variable sequence to obtain the updated model.

[0017] Accordingly, S4 specifically involves inputting the current temperature sequence and the current external variable sequence of the geothermal well into the update model to obtain the predicted temperature value of the geothermal well at the next moment.

[0018] Optionally, updating the historical temperature sequence within the time rolling window using the predicted temperature sequence to obtain the updated temperature sequence specifically includes:

[0019] Obtain the residual sequence of the historical temperature sequence and the predicted temperature sequence within the time rolling window, and determine the confidence interval based on the residual sequence;

[0020] Temperature values ​​that exceed the confidence interval in the historical temperature sequence are removed to obtain an updated temperature sequence.

[0021] Optionally, determining the confidence interval based on the residual sequence specifically includes:

[0022] The weighted residual mean and weighted residual variance are determined based on the residual sequence and the exponentially decaying weights.

[0023] The upper and lower thresholds are determined based on the weighted residual mean and the weighted residual variance, and the upper and lower thresholds are combined to form a confidence interval.

[0024] Optionally, determining the confidence interval based on the residual sequence specifically includes:

[0025] Determine the moving mean and standard deviation of the residuals based on the residual sequence;

[0026] The upper and lower thresholds are determined based on the moving mean and standard deviation of the residuals, and the upper and lower thresholds are combined to form a confidence interval.

[0027] Optionally, both the daily periodic term and the annual periodic term are in the form of double harmonic sine and cosine.

[0028] Optionally, prior to S3, the method further includes:

[0029] Historical temperature sequences and historical external variable sequences of geothermal wells are collected, and the historical temperature sequences and historical external variable sequences are preprocessed.

[0030] Optionally, the preprocessing includes interpolation, denoising, and smoothing.

[0031] The beneficial effects that this invention can produce include:

[0032] The geothermal well temperature sequence prediction method provided by this invention introduces daily and annual cycle terms into the regression model, and simultaneously captures the response of diurnal and annual cycles to geothermal well temperature changes. This "dual seasonal" harmonic modeling method can overcome the defect of insufficient fitting of existing single seasonal models when multiple cycles are superimposed at the same time, thereby achieving high-precision prediction.

[0033] The geothermal well temperature sequence prediction method provided by this invention not only presents a mathematical model but also provides a detailed algorithm flow from data acquisition, preprocessing, parameter estimation to prediction output. In implementation, this invention offers optional modules (such as anomaly detection and weighted adaptive calorimetry) and a default core module (ARMAX + seasonality), which can be freely tailored according to engineering needs to achieve the best balance between performance and computational efficiency. It provides flexible solutions for different industry needs (small-scale shallow wells and large-scale deep wells), expanding the method's applicability. Attached Figure Description

[0034] Figure 1 A flowchart of a geothermal well temperature sequence prediction method provided in an embodiment of the present invention. Detailed Implementation

[0035] The present invention will now be described in detail with reference to the embodiments, but the present invention is not limited to these embodiments.

[0036] This invention provides a method for predicting geothermal well temperature sequences, such as... Figure 1 As shown, the method includes:

[0037] S1. Construct the daily and annual periodic terms of geothermal well temperature, and determine the seasonality term of geothermal well temperature based on the daily and annual periodic terms.

[0038] In one embodiment of the present invention, both the daily periodic term and the annual periodic term are in the form of double harmonic sine and cosine.

[0039] Specifically, daily cycle items The expression is:

[0040] ;

[0041] Annual cycle item The expression is:

[0042] ;

[0043] According to the daily cycle item With annual periodic items Constructing seasonal items Specifically, it takes the form of:

[0044] ;

[0045] In the above formula, For timestamps (hours); The length of the daily cycle ( ); The length of the annual cycle ( ); The harmonic order can be set to 1 to 3 to capture the main cycle; , , and All of these are seasonal parameters that are yet to be estimated.

[0046] Optionally, in another embodiment of the present invention, both the daily periodic term and the annual periodic term are in sinusoidal form.

[0047] Specifically, daily cycle items The expression is:

[0048] ;

[0049] Annual cycle item The expression is:

[0050] ;

[0051] According to the daily cycle item With annual periodic items Constructing seasonal items Specifically, it takes the form of:

[0052] ;

[0053] In the above formula, For timestamps (hours); and All of these are seasonal parameters that are yet to be estimated.

[0054] Using a sinusoidal seasonal term simplifies the model and reduces the number of parameters, but it is slightly inferior to the double harmonic sine and cosine form in terms of fitting accuracy. It is suitable for shallow geothermal systems where the requirements for seasonal variation are not so strict.

[0055] S2. Construct the autoregressive term, moving average term, and external variable term for geothermal well temperature, and construct a regression model for geothermal well temperature based on the autoregressive term, moving average term, external variable term, and seasonality term.

[0056] In this embodiment of the invention, the autoregressive term The expression is:

[0057] ;

[0058] in, These are the autoregressive coefficients; for Temperature value of a geothermal well.

[0059] Moving average term The expression is:

[0060] ;

[0061] in, The moving average coefficient; It is a white noise sequence.

[0062] The aforementioned external variables for constructing geothermal well temperature specifically include:

[0063] First, identify the multiple external variables that affect the temperature of the geothermal well and determine the lag order of each external variable; then, construct the external variable term for the temperature of the geothermal well based on the multiple external variables and their lag orders.

[0064] external variable The expression is:

[0065] ;

[0066] Among them, external variables Include:

[0067] Wellhead pumping flow rate (unit: m³ / h);

[0068] Wellhead production pressure (unit: MPa);

[0069] Surface air temperature (unit: °C);

[0070] : Heating / Power Generation Load Factor (dimensionless);

[0071] Other optional external variables include: wellhead pressure, well inclination, injection temperature, wellhead mud specific gravity, wellbore inclination angle variation, and injection water temperature, which can be expanded to [specific parameters] depending on site conditions. indivual.

[0072] For the corresponding external variable Linear regression coefficients with different lag orders; lag order The time delay of the physical influence of external variables can be determined. For example, the effect of air temperature on well temperature lags by 6 to 12 hours, and this can be set. (6 hours per stage).

[0073] Compared to existing methods that only introduce a single or a few external variables with fixed lags, this invention allows for flexible selection of external variables. External variables and their lag orders This enables a comprehensive interpretation of well temperature changes based on multi-source information.

[0074] In another embodiment of the invention, when the external variable term Includes only one external variable, such as only the pump flow rate. or surface temperature When, external variables The expression is:

[0075] .

[0076] Using a single external variable can greatly simplify data collection and model complexity, making it suitable for situations where data sources are scarce.

[0077] Based on the autoregressive term Moving average term external variable terms and seasonal items Constructing a regression model for geothermal well temperature Its expression is:

[0078] .

[0079] S3. Train the regression model using the historical temperature sequence and historical external variable sequence of geothermal wells to obtain the prediction model.

[0080] The objective function of the prediction model is a weighted least squares objective function; the weights in the objective function are exponentially decaying weights.

[0081] Furthermore, prior to S3, the method also includes:

[0082] Historical temperature and external variable sequences of geothermal wells are collected and preprocessed. Preprocessing may include interpolation, denoising, and smoothing.

[0083] In this invention, the specific process of model training may include the following steps:

[0084] (1) Data acquisition and preprocessing.

[0085] (1.1) Continuous temperature signals at the wellhead are collected using a downhole real-time thermometer, with a sampling interval of [missing information]. , , Or at higher frequencies, recorded as This refers to the historical temperature sequence.

[0086] (1.2) Use a wellhead flow meter and pressure gauge to simultaneously record the pumping flow rate. (Unit: m³ / h), Production Pressure (Unit: MPa), etc.; Surface air temperature is collected using ground weather stations or integrated weather sensors. (Unit: °C); By utilizing the geothermal system's power generation / heating load acquisition equipment, the heating / power generation load coefficient can be obtained. (Normalized dimensionless [0, 1]); the summation is recorded as an external variable. That is, the historical sequence of external variables.

[0087] (1.3) Upload the above data to the ground monitoring center for format standardization, missing value detection and marking. If there is short-term data loss or abnormal changes caused by thermometer drift, temporary interpolation or manual completion can be used. The pre-processed sequence can also be pre-processed with "denoising" or "smoothing". Specifically: if the noise is small, a moving average (window width 3-5 points) can be used for coarse filtering; if the noise is large, wavelet thresholding can be used for denoising (such as db4 wavelet, hard threshold).

[0088] (2) Initialization of parameters for the regression model.

[0089] (2.1) Based on cross-correlation analysis, determine the lag order of the main external variables, specifically:

[0090] If the maximum correlation lag between pumping flow rate and well temperature is approximately Then let .

[0091] If the maximum correlation hysteresis between production pressure and well temperature is approximately Then let .

[0092] If the air temperature lags behind the well temperature by approximately Then let .

[0093] If the load and well temperature are not significantly correlated, a preliminary assumption can be made. .

[0094] (2.2) Selection of the initial order of the regression model: usually selected If the data volume is large, you can try... .

[0095] (2.3) Initialize the parameters of the regression model Specifically, the parameters can be set randomly or based on experience; include: , , ,Will Initially set to a small random number or an initial value determined based on the results of a simple linear regression.

[0096] (3) Construct the response vector and objective function.

[0097] (3.1) Set the length of the time scrolling window Typically, hourly data from the most recent 3 to 6 months are selected, for example, if the sampling frequency... Then we can set or .

[0098] (3.2) For each moment within the time scrolling window Assign exponential decay weights:

[0099] ;

[0100] in, for Exponentially decaying weights at time points, This is the attenuation coefficient.

[0101] (3.3) Constructing the response vector: For each ( from arrive ),

[0102] Based on the historical temperature series of collected geothermal wells, the autoregressive term vector of the model input is determined. .

[0103] Determine the moving average term vector ,initial You can set it to 0 first.

[0104] Based on the collected historical external variable sequences, determine the external variable term vector for the model input. .

[0105] Determine the seasonal term vector .

[0106] Input the above response vector into the following regression model:

[0107] .

[0108] (3.4) Construct the weighted least squares objective function The expression is as follows:

[0109] .

[0110] This invention assigns higher weights to recent data, enabling model parameters to respond in real time to changes in formation conditions and equipment performance drift, thus avoiding the drawbacks of parameter failure under fixed parameters or fixed windows.

[0111] (3.5) Numerical optimization algorithms such as Iterative Weighted Least Squares (IRLS) or quasi-Newton methods are used to optimize the parameters. The solution process continues until the objective function converges or the preset number of iterations is reached, ultimately yielding the result at time [time value missing]. Model parameter estimation .

[0112] S4. Input the current temperature sequence and current external variable sequence of the geothermal well into the prediction model to obtain the predicted temperature value of the geothermal well at the next moment.

[0113] At any moment After all parameters have been updated, you can directly proceed with future updates. Time period ( Can be , , Perform rolling forecasts:

[0114] ;

[0115] right Replace with "previous prediction value" in sequence Then make further predictions to form a rolling prediction sequence.

[0116] Preferably, after S3, the method further includes:

[0117] (1) Determine the time rolling window and use the prediction model to determine the predicted temperature sequence of the geothermal well within the time rolling window.

[0118] (2) The historical temperature sequence within the time rolling window is updated using the predicted temperature sequence to obtain the updated temperature sequence, and the prediction model is trained using the updated temperature sequence and the historical external variable sequence to obtain the updated model.

[0119] Accordingly, S4 specifically involves inputting the current temperature sequence and the current external variable sequence of the geothermal well into the update model to obtain the predicted temperature value of the geothermal well at the next moment.

[0120] Specifically, the updated temperature sequence is obtained by using the historical temperature sequence within the time rolling window updated by the predicted temperature sequence, including:

[0121] First, obtain the residual sequences of the historical temperature series and the predicted temperature series within the time rolling window, and determine the confidence interval based on the residual sequences; then, remove the temperature values ​​in the historical temperature series that exceed the confidence interval to obtain the updated temperature series.

[0122] The residual sequences of the historical and predicted temperature sequences obtained above within the time rolling window can be specifically calculated using the following formula:

[0123] ;

[0124] in, It is a residual sequence; For historical temperature sequences; To predict temperature sequences.

[0125] The above-mentioned determination of confidence intervals based on residual sequences specifically includes:

[0126] The weighted residual mean and weighted residual variance are determined based on the residual sequence and the exponentially decaying weights.

[0127] The upper and lower thresholds are determined based on the weighted residual mean and weighted residual variance, and the upper and lower thresholds are combined to form a confidence interval.

[0128] Specifically, the weighted residual mean is calculated using the following formula. and weighted residual variance :

[0129] , ;

[0130] at this time, .

[0131] The confidence interval is [ ].

[0132] If at a certain moment If the temperature exceeds the above confidence interval, the point is considered an outlier, and its temperature value is removed from the historical temperature sequence and will not be used in the next round of parameter updates.

[0133] This invention calculates the weighted residual mean within a rolling time window. and weighted residual variance With threshold Anomalies are detected and removed from the historical temperature sequence, preventing them from being included in the next round of parameter updates. Compared to methods that only use wavelet thresholding or fixed thresholding for removal, this invention can dynamically determine anomalies by combining prediction results, avoiding the model being dragged back by outliers.

[0134] In another embodiment of the present invention, determining the confidence interval based on the residual sequence specifically includes:

[0135] Determine the moving mean and standard deviation of the residuals based on the residual sequence;

[0136] The upper and lower thresholds are determined based on the moving mean and standard deviation of the residuals, and the upper and lower thresholds are combined to form a confidence interval.

[0137] For example, directly calculate the recent Time period (e.g.) Moving mean of residuals in the residual series With residual standard deviation ,like If an outlier is detected, that moment is identified as an anomaly, and its temperature value is removed from the historical temperature series, preventing it from being used in the next round of parameter updates. This method does not require exponential weighting and is suitable for scenarios with relatively low frequency of abnormal fluctuations and relatively stable data.

[0138] After removing outliers from the historical temperature sequence, an updated temperature sequence is obtained.

[0139] Furthermore, the attenuation coefficient can be determined based on the updated temperature sequence and the predicted temperature sequence. The calculation formula is as follows:

[0140] .

[0141] After obtaining the updated temperature sequence and attenuation coefficient Then, the parameters of the prediction model can be estimated again, and the process can be repeated until all residuals are within the confidence interval or the maximum number of iterations is reached.

[0142] Finally, the predicted temperature sequence and confidence interval are output to the ground monitoring center for real-time reference by the power generation / heating system. Based on the predicted temperature sequence, future load dispatch suggestions, heat allocation plans, or maintenance warnings can be generated. Model parameters can also be periodically saved for quarterly / annual review and analysis; and the system can be periodically recalibrated. Model structure, etc.

[0143] Compared with the prior art, the present invention has the following significant advantages:

[0144] (1) High-precision prediction.

[0145] This invention utilizes a combination of "dual seasonal harmonics + multi-source external variable fusion + adaptive weighted updating" to simultaneously capture diurnal and annual cycles, multi-source physical driving factors, and real-time updates, enabling the model to respond more quickly and accurately to geothermal well temperature changes. For example, in In short-term forecasts, the root mean square error (RMSE) decreased from 1.5°C in conventional ARMA to 0.4°C; in 7-day medium-term forecasts, the RMSE decreased from 1.5°C to 0.6°C.

[0146] (2) Strong robustness and strong anomaly handling capability.

[0147] This invention employs "dynamic residual statistical anomaly detection + local interpolation correction". Based on the dynamic determination of residuals combined with bilinear interpolation correction, the model is more robust when facing outliers. It can automatically filter out sudden anomalies such as well shutdown and maintenance, and thermometer drift, and avoid anomalies affecting subsequent model parameters. Experiments have shown that it can still maintain high prediction accuracy when the occurrence rate of anomalies is about 1% to 2%.

[0148] (3) It has strong self-adaptability and can adapt to long-term non-stationary changes.

[0149] This invention is based on the fundamental principle that exponential weighting makes new data have a greater impact and old data decay rapidly, and parameters can be dynamically adjusted, making it suitable for slow and non-stationary systems. By adopting "rolling window exponential weighted least squares online update", it can capture parameter drift caused by formation thermal reservoir decay and equipment aging in real time, so that parameters are updated smoothly over time throughout the year. A single model can maintain stable prediction results in different time periods without human intervention.

[0150] (5) The project has strong usability and scalability.

[0151] This invention significantly improves the accuracy and robustness of geothermal well temperature prediction. Furthermore, the parameters involved in the method can be flexibly adjusted according to actual engineering conditions. It possesses good engineering applicability and can provide reliable predictive references for geothermal power generation and geothermal heating systems, helping to further improve resource utilization efficiency and reduce operation and maintenance costs.

[0152] The above description is merely a few embodiments of this application and is not intended to limit this application in any way. Although this application discloses preferred embodiments as described above, it is not intended to limit this application. Any changes or modifications made by those skilled in the art without departing from the scope of the technical solution of this application using the disclosed technical content are equivalent to equivalent implementation cases and fall within the scope of the technical solution.

Claims

1. A method for predicting geothermal well temperature sequences, characterized in that, The method includes: S1. Construct the daily and annual periodic terms of geothermal well temperature, and determine the seasonality of geothermal well temperature based on the daily and annual periodic terms; both the daily and annual periodic terms are in double harmonic sine and cosine form. The daily cycle item The expression is: ； The annual periodic item The expression is: ； According to daily cycle items With annual periodic items Constructed seasonal items The expression is: ； In the formula, For timestamps; The length of the daily cycle; The length of the annual cycle; For harmonic order; , , and All of these are seasonal parameters to be estimated; S2. Construct the autoregressive term and moving average term for geothermal well temperature, identify multiple external variables affecting geothermal well temperature, and determine the lag order of each external variable; based on the multiple external variables and their lag orders, construct the external variable term for geothermal well temperature, and construct a regression model for geothermal well temperature based on the autoregressive term, the moving average term, the external variable term, and the seasonality term. Among them, the external variable term The expression is: ； In the formula, For external variables; For the corresponding external variable Linear regression coefficients with different lag orders ; This represents the total number of external variables. The lag order; S3. Train the regression model using the historical temperature sequence and historical external variable sequence of geothermal wells to obtain the prediction model; S4. Input the current temperature sequence and the current external variable sequence of the geothermal well into the prediction model to obtain the predicted temperature value of the geothermal well at the next moment.

2. The method according to claim 1, characterized in that, The objective function of the prediction model is a weighted least squares objective function; the weights in the objective function are exponentially decaying weights.

3. The method according to claim 2, characterized in that, Following S3, the method further includes: A time rolling window is determined, and the prediction model is used to determine the predicted temperature sequence of the geothermal well within the time rolling window; The predicted temperature sequence is used to update the historical temperature sequence within the time rolling window to obtain the updated temperature sequence. The prediction model is then trained using the updated temperature sequence and the historical external variable sequence to obtain the updated model. Accordingly, S4 specifically involves inputting the current temperature sequence and the current external variable sequence of the geothermal well into the update model to obtain the predicted temperature value of the geothermal well at the next moment.

4. The method according to claim 3, characterized in that, The step of updating the historical temperature sequence within the time rolling window using the predicted temperature sequence to obtain the updated temperature sequence specifically includes: Obtain the residual sequence of the historical temperature sequence and the predicted temperature sequence within the time rolling window, and determine the confidence interval based on the residual sequence; Temperature values ​​that exceed the confidence interval in the historical temperature sequence are removed to obtain an updated temperature sequence.

5. The method according to claim 4, characterized in that, The step of determining the confidence interval based on the residual sequence specifically includes: The weighted residual mean and weighted residual variance are determined based on the residual sequence and the exponentially decaying weights. The upper and lower thresholds are determined based on the weighted residual mean and the weighted residual variance, and the upper and lower thresholds are combined to form a confidence interval.

6. The method according to claim 4, characterized in that, The step of determining the confidence interval based on the residual sequence specifically includes: Determine the moving mean and standard deviation of the residuals based on the residual sequence; The upper and lower thresholds are determined based on the moving mean and standard deviation of the residuals, and the upper and lower thresholds are combined to form a confidence interval.

7. The method according to claim 1, characterized in that, Prior to S3, the method further includes: Historical temperature sequences and historical external variable sequences of geothermal wells are collected, and the historical temperature sequences and historical external variable sequences are preprocessed.

8. The method according to claim 7, characterized in that, The preprocessing includes interpolation, noise reduction, and smoothing.

Images