Intelligent ground source heat pump management method, system and device

Abstract

The invention discloses an intelligent ground source heat pump management method, system and device, and relates to the technical field of ground source heat pumps, and the intelligent ground source heat pump management method comprises the steps that the injection temperature characteristic and the injection flow characteristic of an inlet of a geothermal well, the first temperature characteristic and the first pressure characteristic of an outlet of the geothermal well and the corresponding time characteristic are obtained; normalizing the injection temperature feature, the injection flow feature and the time feature to obtain an injection data set, and normalizing the corresponding first temperature feature and the first pressure feature to obtain a first outlet data set; inputting the injection data set and the first outlet data set into a preset large model, and training the large model to predict the temperature and pressure of the outlet of the geothermal well; after the training of the large model is completed, inputting the injection port data set into the trained large model, and outputting a second outlet data set by the large model; and carrying out inverse normalization on the second outlet data set to obtain a second temperature characteristic and a second pressure characteristic of the geothermal well outlet predicted by the large model, thereby accurately predicting the temperature and the pressure of the geothermal well outlet.

Description

Technical Field

[0001] The present invention relates to the technical field of ground source heat pumps, and in particular to an intelligent ground source heat pump management method, system and device. Background Art

[0002] The ground source heat pump is an efficient, energy-saving and environmentally friendly air conditioning system that utilizes shallow underground geothermal resources. The ground source heat pump can transfer energy from low-temperature heat sources to high-temperature heat sources by inputting a small amount of high-quality energy. After decades of development in my country, ground source heat pumps are widely used in heating systems. At present, the ground source heat pump energy is managed through intelligent ground source heat pump management devices.

[0003] In existing geothermal heat pump systems, in order to accurately describe the complex thermal and fluid transfer process between the injection well and the production well, it is usually necessary to establish complex partial differential equations and perform fine grid division and time discretization. This method not only increases the complexity of the calculation model, but also puts higher requirements on computing resources, and cannot calculate and predict the geothermal well outlet temperature and pressure with high accuracy, generalization ability and robustness.

[0004] In addition, when managing the geothermal heat pump energy, the operator is sometimes required to go to the operation room to adjust the parameters of the geothermal heat pump. However, the control panel of the existing geothermal heat pump is generally set on the main box on the top of the geothermal heat pump. When the geothermal heat pump is running, heat is easily accumulated on its surface. The operator cannot avoid the need to get close to the geothermal heat pump when adjusting the parameters. The operator is prone to contact with the geothermal heat pump and cause injury. Summary of the invention

[0005] The present invention provides an intelligent ground source heat pump management method, system and device, which can accurately and quickly predict the temperature and pressure of a geothermal well outlet.

[0006] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0007] Intelligent ground source heat pump management method, including:

[0008] S1: Obtaining the injection temperature characteristics and injection flow characteristics of the geothermal well inlet, the first temperature characteristics and the first pressure characteristics of the geothermal well outlet, and the corresponding time characteristics;

[0009] S2: normalizing the injection temperature characteristic, injection flow characteristic and time characteristic to obtain an injection data set, and normalizing the corresponding first temperature characteristic and first pressure characteristic to obtain a first outlet data set;

[0010] S3: inputting the injection data set and the first outlet data set into a preset large model and training the large model to predict the temperature and pressure of the geothermal well outlet;

[0011] S4: After the training of the large model is completed, the injection port data set is input into the trained large model, and the large model outputs a second outlet data set;

[0012] S5: Inverse normalize the second outlet data set to obtain the second temperature characteristic and the second pressure characteristic of the geothermal well outlet predicted by the large model.

[0013] Preferably, the normalized expression is Among them, the X norm represents the normalized features, Indicates the injection temperature characteristics of the geothermal well inlet, Indicates the injection flow characteristics of the geothermal well inlet, represents the geothermal well outlet temperature characteristics, represents the geothermal well outlet pressure characteristics, t represents the time characteristics, X represents the variables in the data set, X min and X max are the minimum and maximum values ​​of the features respectively, and the X norm The range is between 0 and 1.

[0014] Preferably, the large model adopts an artificial neural network framework, and the large model includes an input layer, a hidden layer and an output layer. The input layer is provided with three nodes, including injection temperature characteristics, injection flow characteristics and corresponding time characteristics; the hidden layer includes a first hidden layer, a second hidden layer, a third hidden layer, a fourth hidden layer and a fifth hidden layer. The first hidden layer includes 8 neurons and is connected to the input layer, the second hidden layer includes 16 neurons and is connected to the first hidden layer, the third hidden layer includes 32 neurons and is connected to the second hidden layer, the fourth hidden layer includes 16 neurons and is connected to the third hidden layer, and the fifth hidden layer includes 8 neurons and is connected to the fourth hidden layer; the output layer includes 2 neurons, which are used to output the second temperature characteristic and the second pressure characteristic to predict the temperature and pressure at the outlet of the geothermal well.

[0015] Preferably, the hidden layer uses ReLU as the activation function, expressed as ReLU(x)=max(0,x).

[0016] Preferably, the large modeling adopts mean square error (MSE) as the loss function, and the mean square error measures the error of the model by calculating the square difference between the predicted value and the real-time value, and its expression is as follows:

[0017] in, and They are the real-time injection temperature characteristics and injection flow characteristics of the geothermal well outlet, and is the predicted value of the second temperature characteristic and the second pressure characteristic of the geothermal well calculated by the large model, and n is the number of samples; a smaller MSE value indicates that the prediction of the large model is accurate.

[0018] Preferably, the large model training step includes:

[0019] S1: Forward propagation, passing the input data to the model, calculating the output layer by layer, and finally getting the prediction result. The mathematical expression of forward propagation is as follows:

[0020] a (l) =f(W (l) a (l-1) +b (l) )

[0021] Where a(l) represents the activation value of the lth layer, W(l) and b(l) are the weight and bias of the lth layer respectively, and f is the activation function;

[0022] S2: Calculate the loss and use the loss function (MSE) to calculate the error between the predicted value and the real-time value;

[0023] S3: Back propagation, calculate the partial derivative of the weight of each layer according to the loss, and use the chain rule to calculate the gradient layer by layer; the mathematical expression of the back propagation is as follows:

[0024]

[0025] Among them, L is the loss function and is the gradient of the lth layer.

[0026] S4: Weight update: Use the optimizer to update the weight of the network based on the calculated gradient to minimize the value of the loss function. The mathematical expression of the weight update is:

[0027] Among them, η is the learning rate.

[0028] Preferably, after the training of the large model is completed, the real-time features and the training features are normalized in the same manner, and the normalized data is input into the trained large model. The output expression of the trained large model is:

[0029]

[0030] in, represents the normalized second temperature characteristic of the geothermal well outlet predicted by the large model, represents the normalized second pressure characteristic of the geothermal well outlet predicted by the large model, t new Represents the real-time or large-model training geothermal well inlet injection temperature characteristics, Represents the temporal features of real-time or large model training, Represents the real-time or large model training geothermal well inlet injection flow characteristics.

[0031] Preferably, the inverse normalization expression of the large model is:

[0032] in, The second temperature characteristic representing the physical quantity at the geothermal well outlet predicted by the large model, A second pressure characteristic representing a physical quantity at the geothermal well outlet predicted by the large model.

[0033] The present invention also provides an intelligent ground source heat pump management system, comprising the following modules:

[0034] Acquisition module: obtain the injection temperature characteristics and injection flow characteristics of the geothermal well inlet, the first temperature characteristics and the first pressure characteristics of the geothermal well outlet, and the corresponding time characteristics;

[0035] Preprocessing module: normalizing the injection temperature feature, injection flow rate feature and time feature to obtain an injection data set, and normalizing the corresponding first temperature feature and first pressure feature to obtain a first outlet data set;

[0036] Large model module: inputs the injection data set into a preset large model, and outputs a second outlet data set for predicting the temperature and pressure of the geothermal well outlet;

[0037] Inverse processing module: inverse normalizes the second outlet data set to obtain the predicted second temperature characteristic and second pressure characteristic of the geothermal well outlet.

[0038] The present invention also provides an intelligent ground source heat pump management device, comprising a ground source heat pump body, a main box and an intelligent energy management module, wherein the main box is arranged on the top of the ground source heat pump body, a control box is arranged on the front of the main box, a protective flip assembly is arranged on the bottom of the control box, a protective cover is arranged on one side of the protective flip assembly, a control panel is arranged in the inner cavity of the control box, moving components are arranged on both the left and right sides of the control panel, and a heat recovery component is arranged on the left side of the ground source heat pump body;

[0039] The protective flip assembly includes a chassis, the chassis is fixedly connected to the bottom of the control box, the inner cavity of the chassis is fixedly connected to a dual-axis motor, the output shaft of the dual-axis motor is fixedly connected to a first gear, the surface of the first gear is meshed with a second gear, and the axis of the second gear is fixedly connected to a rotating shaft;

[0040] The moving assembly includes a mounting plate, the mounting plate is fixed on the left and right sides of the control panel, a small motor is fixedly connected to the top of the mounting plate, an output shaft of the small motor is fixedly connected to a third gear, and a main body tooth plate is meshed on the surface of the third gear;

[0041] The intelligent energy management module includes a control center, sensor nodes, a data acquisition and analysis module, an intelligent control algorithm, a user interaction interface and sensor nodes.

[0042] The present invention has the following beneficial effects:

[0043] The present invention adopts a coordination method of normalization processing, inverse normalization processing and a large model, so as to more accurately predict the geothermal well outlet temperature and pressure, thereby improving the safety and efficiency of production.

[0044] In addition, the present invention adopts a method of normalizing data features, which can not only approach the accuracy of numerical simulation, but also significantly reduce the calculation time.

[0045] In addition, the large model of the present invention uses machine learning technology to efficiently and accurately approximate the input-output relationship in large-scale dynamic systems. In the application of ground source heat pumps, the heat flow process between injection wells and production wells is highly nonlinear, requiring complex coupled simulation of multiple physical fields, thereby saving time and computing resources. The advantages are more prominent.

[0046] In addition, the present invention adopts an artificial neural network framework to automatically learn the nonlinear mapping relationship of the underground model through data, without relying on a simplified physical model, and can achieve the calculation and prediction of geothermal well outlet temperature and pressure with high accuracy, generalization ability and robustness.

[0047] In addition, the present invention is designed with a protective flip component. When no parameter adjustment is performed, the protective cover can protect the control panel to prevent the external environment from causing damage to the control panel. When the parameters are adjusted, the protective flip component drives the protective cover to rotate ninety degrees, and then the moving component drives the control panel to move on the protective cover, so that the control panel is close to the operator and there is a certain distance between the operator and the ground source heat pump body. This solves the problem in the prior art that the operator cannot avoid the need to get close to the ground source heat pump when adjusting the parameters, and the operator is easily exposed to the ground source heat pump and caused injury.

[0048] In addition, the present invention facilitates the introduction of external air into the inner cavity of the control box through the cooperation of the ventilator, the heat dissipation fan and the dustproof net, thereby facilitating the cooling of the internal electronic components and improving the heat dissipation effect. The dustproof net can prevent external dust from entering the control box, and the temperature sensor facilitates real-time monitoring of the inner cavity temperature of the control box, thereby facilitating the control of the working efficiency of the corresponding heat dissipation fan through the control panel to ensure the heat dissipation effect of the control box. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for describing the embodiments are briefly introduced below.

[0050] Figure 1 It is a structural stereogram of the intelligent ground source heat pump management device;

[0051] Figure 2 A three-dimensional diagram of the structure when used in an intelligent ground-source heat pump management device;

[0052] Figure 3 It is a three-dimensional diagram of the control box structure in the intelligent ground source heat pump management device;

[0053] Figure 4 It is a three-dimensional diagram of the structure of the mobile components in the intelligent ground source heat pump management device;

[0054] Figure 5 It is a three-dimensional diagram of the first gear structure in the intelligent ground source heat pump management device;

[0055] Figure 6 It is a three-dimensional diagram of the structure of the heat recovery component in the intelligent ground source heat pump management device;

[0056] Figure 7 It is a three-dimensional diagram of the water tank structure in the intelligent ground source heat pump management device;

[0057] Figure 8 Schematic diagram of the large model artificial neural network framework.

[0058] In the attached figure:

[0059] 1. Ground source heat pump body; 2. Main chassis; 3. Control box; 4. Protective cover; 5. Control panel; 6. Chassis; 7. Temperature sensor; 8. Dual-axis motor; 9. First gear; 10. Second gear; 11. Rotating shaft; 12. Mounting plate; 13. Small motor; 14. Third gear; 15. Main gear plate; 16. Support pulley; 17. Slide; 18. Groove; 19. Ventilator; 20. Cooling fan; 21. Dustproof net; 22. Cooling slot; 23. Observation window; 24. Water tank; 25. First drain pipe; 26. Heat recovery box; 27. Second drain pipe; 28. Booster pump; 29. ​​Three-way valve; 30. Water temperature sensor; 31. Reflux pipe; 32. Heat recovery pipe; 33. Mounting sleeve; 34. One-way valve; 35. Circulation motor; 36. Blade. DETAILED DESCRIPTION

[0060] The technical solutions in the embodiments of the present invention will be described below in conjunction with the drawings in the embodiments of the present invention. The described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0061] This embodiment provides an intelligent ground source heat pump management method, which inputs the geothermal well inlet data set and outlet data set into a preset large model and trains the large model. After the large model is trained, the temperature and pressure of the geothermal well outlet can be predicted, thereby improving production safety and management capabilities.

[0062] The intelligent ground source heat pump management method comprises:

[0063] S1: Obtaining the injection temperature characteristics and injection flow characteristics of the geothermal well inlet, the first temperature characteristics and the first pressure characteristics of the geothermal well outlet, and the corresponding time characteristics;

[0064] S2: normalizing the injection temperature characteristic, injection flow characteristic and time characteristic to obtain an injection data set, and normalizing the corresponding first temperature characteristic and first pressure characteristic to obtain a first outlet data set;

[0065] S3: inputting the injection data set and the first outlet data set into a preset large model and training the large model to predict the temperature and pressure of the geothermal well outlet;

[0066] S4: After the training of the large model is completed, the injection port data set is input into the trained large model, and the large model outputs a second outlet data set;

[0067] S5: Inverse normalize the second outlet data set to obtain the second temperature characteristic and the second pressure characteristic of the geothermal well outlet predicted by the large model.

[0068] This embodiment adopts a combination of normalization processing, inverse normalization processing and a large model, so that the geothermal well outlet temperature and pressure can be predicted more accurately, thereby improving the safety and efficiency of production.

[0069] Preferably, the normalized expression is Among them, the X norm represents the normalized features, Indicates the injection temperature characteristics of the geothermal well inlet, Indicates the injection flow characteristics of the geothermal well inlet, represents the geothermal well outlet temperature characteristics, represents the geothermal well outlet pressure characteristics, t represents the time characteristics, X represents the variables in the data set, X min and X max are the minimum and maximum values ​​of the features respectively, and the X normThe range is between 0 and 1. This embodiment adopts a method of normalizing data features, which can not only approach the accuracy of numerical simulation, but also significantly reduce the calculation time, reduce the scale difference between features, and ensure that the model can be stably trained.

[0070] Furthermore, the X norm represents the normalized feature, that is, the X norm is a data set, the X norm The dataset is divided into a training set and a test set. The training set is used to learn the neural network parameters, that is, to adjust the network weights to capture the inherent laws in the data; the test set is used to verify the performance of the model on unseen data.

[0071] Preferably, Figure 8 As shown, the large model adopts an artificial neural network framework, and the large model includes an input layer, a hidden layer and an output layer. The input layer is provided with three nodes, including injection temperature characteristics, injection flow characteristics and corresponding time characteristics; the hidden layer includes a first hidden layer, a second hidden layer, a third hidden layer, a fourth hidden layer and a fifth hidden layer. The first hidden layer includes 8 neurons and is connected to the input layer, the second hidden layer includes 16 neurons and is connected to the first hidden layer, the third hidden layer includes 32 neurons and is connected to the second hidden layer, the fourth hidden layer includes 16 neurons and is connected to the third hidden layer, and the fifth hidden layer includes 8 neurons and is connected to the fourth hidden layer; the output layer includes 2 neurons, which are used to output the second temperature characteristic and the second pressure characteristic, and predict the temperature and pressure of the geothermal well outlet. This embodiment adopts an artificial neural network framework, automatically learns the nonlinear mapping relationship of the underground model through data, does not need to rely on a simplified physical model, and can realize the calculation and prediction of the geothermal well outlet temperature and pressure with high accuracy, generalization ability and robustness.

[0072] Preferably, the hidden layer uses ReLU as the activation function, expressed as ReLU(x)=max(0,x). ReLU can effectively solve the gradient vanishing problem and help the large model learn complex nonlinear relationships. Through layer-by-layer calculation and weight update, the large model can gradually learn the input (injected temperature Injection flow and time t) and output (outlet temperature and outlet pressure ) to achieve accurate prediction.

[0073] Preferably, the large modeling adopts mean square error (MSE) as the loss function, and the mean square error measures the error of the model by calculating the square difference between the predicted value and the real-time value, and its expression is as follows: in, and They are the real-time injection temperature characteristics and injection flow characteristics of the geothermal well outlet, and is the predicted value of the second temperature characteristic and the second pressure characteristic of the geothermal well calculated by the large model, and n is the number of samples; a smaller MSE value indicates that the prediction of the large model is accurate. Thus, the large model can measure the deviation between the model prediction value and the true value.

[0074] Preferably, the large model training step includes:

[0075] S1: Forward propagation, passing the input data to the model, calculating the output layer by layer, and finally getting the prediction result. The mathematical expression of forward propagation is as follows:

[0076] a (l) =f(W (l) a (l-1) +b (l) )

[0077] Where a(l) represents the activation value of the lth layer, W(l) and b(l) are the weight and bias of the lth layer respectively, and f is the activation function;

[0078] S2: Calculate the loss and use the loss function (MSE) to calculate the error between the predicted value and the real-time value;

[0079] S3: Back propagation, calculate the partial derivative of the weight of each layer according to the loss, and use the chain rule to calculate the gradient layer by layer; the mathematical expression of the back propagation is as follows:

[0080]

[0081] Among them, L is the loss function and is the gradient of the lth layer.

[0082] S4: Weight update: Use the optimizer to update the weight of the network based on the calculated gradient to minimize the value of the loss function. The mathematical expression of the weight update is:

[0083] Among them, η is the learning rate.

[0084] This embodiment uses machine learning technology to efficiently and accurately approximate the input-output relationship in a large-scale dynamic system. In the application of ground source heat pumps, the heat flow process between the injection well and the production well is highly nonlinear, and complex coupling simulation of multiple physical fields is required, so that the advantages of saving time and computing resources are more prominent. In addition, for each iteration of the large model training process, the model will optimize the loss function, and the prediction error of the large model will gradually decrease by continuously adjusting the weights of the network. In order to monitor the performance during the training process, the loss values ​​of the training set and the validation set are calculated, and the loss change curve is plotted. These curves help analyze the training status of the large model, identify whether there is an overfitting or underfitting problem, and adjust the model parameters or structure accordingly.

[0085] Furthermore, in order to optimize the parameters of the large model, the Adam optimizer (Adaptive Moment Estimation) is used. The Adam optimizer is an adaptive learning rate optimization algorithm that combines the advantages of momentum and RMSProp and can achieve fast and stable convergence in most cases.

[0086] Preferably, after the training of the large model is completed, the real-time features and the training features are normalized in the same manner, and the normalized data is input into the trained large model. The output expression of the trained large model is:

[0087]

[0088] in, represents the normalized second temperature characteristic of the geothermal well outlet predicted by the large model, represents the normalized second pressure characteristic of the geothermal well outlet predicted by the large model, t new Represents the real-time or large-model training geothermal well inlet injection temperature characteristics, Represents the temporal features of real-time or large model training, Represents the real-time or large model training geothermal well inlet injection flow characteristics.

[0089] Preferably, the inverse normalization expression of the large model is:

[0090] in, The second temperature characteristic representing the physical quantity at the geothermal well outlet predicted by the large model, A second pressure characteristic representing a physical quantity at the geothermal well outlet predicted by the large model.

[0091] This embodiment also provides an intelligent ground source heat pump management system, including the following modules:

[0092] Acquisition module: obtain the injection temperature characteristics and injection flow characteristics of the geothermal well inlet, the first temperature characteristics and the first pressure characteristics of the geothermal well outlet, and the corresponding time characteristics;

[0093] Preprocessing module: normalizing the injection temperature feature, injection flow rate feature and time feature to obtain an injection data set, and normalizing the corresponding first temperature feature and first pressure feature to obtain a first outlet data set;

[0094] Large model module: inputs the injection data set into a preset large model, and outputs a second outlet data set for predicting the temperature and pressure of the geothermal well outlet;

[0095] Inverse processing module: inverse normalizes the second outlet data set to obtain the predicted second temperature characteristic and second pressure characteristic of the geothermal well outlet.

[0096] See also Figure 1-Figure 7 The present embodiment further provides an intelligent ground source heat pump management device, comprising a ground source heat pump body 1 and a main box 2, the main box 2 is arranged on the top of the ground source heat pump body 1, the control box 3 is arranged on the front of the main box 2, the bottom of the control box 3 is arranged with a protective flip assembly, one side of the protective flip assembly is arranged with a protective cover 4, the inner cavity of the control box 3 is arranged with a control panel 5, the left and right sides of the control panel 5 are arranged with moving components, and the left side of the ground source heat pump body 1 is arranged with a heat recovery component;

[0097] The protective flip assembly includes a chassis 6, which is fixedly connected to the bottom of the control box 3, and the inner cavity of the chassis 6 is fixedly connected to a dual-axis motor 8, and the output shaft of the dual-axis motor 8 is fixedly connected to a first gear 9, and the surface of the first gear 9 is meshed with a second gear 10, and the axis of the second gear 10 is fixedly connected to a rotating shaft 11;

[0098] The moving component includes a mounting plate 12, which is fixed on the left and right sides of the control panel 5. A small motor 13 is fixedly connected to the top of the mounting plate 12, and an output shaft of the small motor 13 is fixedly connected to a third gear 14, and a main body tooth plate 15 is meshed on the surface of the third gear 14.

[0099] Specifically: During daily operation of the ground source heat pump body 1, the control panel 5 displays the operating parameters of the ground source heat pump body 1 through the observation window 23, which is convenient for the operator to observe. When it is necessary to get close to the ground source heat pump body 1 to adjust the parameters, the protective flip assembly drives the protective cover 4 to rotate ninety degrees, and then the moving assembly drives the control panel 5 to move on the protective cover 4, so that the control panel 5 is close to the operator, so that there is a certain distance between the operator and the ground source heat pump body 1, which solves the problem in the prior art that the operator cannot avoid getting close to the ground source heat pump when adjusting the parameters, and the operator is easy to touch the ground source heat pump and cause injury.

[0100] See also Figure 1-Figure 7 , one end of the rotating shaft 11 is fixedly connected to one side of the protective cover 4, the number of the rotating shafts 11 is two, one side of the mounting plate 12 is fixedly connected with a supporting pulley 16, and the other side of the supporting pulley 16 is slidably connected with a slide groove 17, the slide groove 17 is arranged on the inner side of the protective cover 4, the inner side of the protective cover 4 is provided with a groove 18, one side of the main tooth plate 15 is fixedly connected to the inner cavity of the groove 18, the inner cavity of the control box 3 is provided with an auxiliary tooth plate, the left and right sides of the control box 3 are provided with a ventilator 19, the inner cavity of the ventilator 19 is provided with a heat dissipation fan 20, the inner cavity of the ventilator 19 is provided with a dustproof net 21, the left and right sides of the control box 3 are provided with a heat dissipation groove 22, the heat dissipation groove 22 is inclined at a certain angle to the horizontal, the front of the protective cover 4 is provided with an observation window 23, the top and bottom of the protective cover 4 are both rounded, the top of the inner cavity of the control box 3 is provided with a temperature sensor 7, the output end of the temperature sensor 7 is electrically connected to the input end of the control panel 5, the heat recovery component It includes a left water tank 24 fixed to the main case 2, a first drain pipe 25 is arranged at the bottom of the water tank 24, the other end of the first drain pipe 25 is connected to a heat recovery tank 26, the bottom of the heat recovery tank 26 is connected to a second drain pipe 27, a pressure pump 28, a three-way valve 29 and a water temperature sensor 30 are arranged on the surface of the first drain pipe 25 from top to bottom, one side of the three-way valve 29 is connected to a return pipe 31, a heat recovery pipe 32 is arranged in the inner cavity of the water tank 24, a mounting sleeve 33 is arranged on one side of the heat recovery pipe 32, one side of the heat recovery pipe 32 is closely attached to the surface of the ground source heat pump body 1, the other end of the second drain pipe 27 is connected to the back of the water tank 24, the other end of the return pipe 31 is connected to the left side of the water tank 24, a one-way valve 34 is sleeved on the surface of the return pipe 31, the heat recovery pipe 32 is arranged in a U shape in the inner cavity of the water tank 24, a circulation motor 35 is fixedly connected to the back of the water tank 24, and a blade 36 is fixedly connected to the output shaft of the circulation motor 35.

[0101] Specifically: by designing two rotating shafts 11, the stability of the protective cover 4 during rotation can be increased. By cooperating with the supporting pulley 16 and the slide groove 17, it is convenient to limit the control panel 5, making it more stable during the forward movement. By designing the auxiliary tooth plate, it can be ensured that the control panel 5 can be separated from the main tooth plate 15 during the process of moving inward to the innermost side, avoiding interference with the protective cover 4. By cooperating with the ventilator 19, the heat dissipation fan 20 and the dustproof net 21, it is convenient to pass the external air into the inner cavity of the control box 3, so as to cool down the internal electronic components and improve the The dustproof net 21 can prevent external dust from entering the control box 3. The heat dissipation slot 22 is designed to be inclined at a certain angle to the horizontal, which facilitates the discharge of gas inside the control box 3. At the same time, it can also prevent external gas from entering the inner cavity of the control box 3 through the heat dissipation slot 22. The observation window 23 is designed to facilitate the display of parameters of the ground source heat pump body 1 during daily operation. The top and bottom of the protective cover 4 are designed with rounded corners to ensure that the top and bottom of the protective cover 4 will not interfere with the control box 3 when it is turned over. The temperature sensor 7 is designed to facilitate the temperature of the inner cavity of the control box 3. Real-time monitoring is carried out to facilitate the control of the working efficiency of the corresponding cooling fan through the control panel 5. The heat generated by the ground source heat pump body 1 heats the circulating water in the inner cavity of the heat recovery tank 26. The booster pump 28 drives the circulating water to circulate in the water tank 24 and the heat recovery tank 26. The high-temperature circulating water heats the heat recovery pipe 32 to generate high-temperature gas inside it, which is then transported to the required heating pipeline. The water temperature sensor 30 can detect the water temperature of the circulating water in the first drain pipe 25. When its temperature is lower than the set temperature, it indicates that it needs to enter the heat recovery tank 26 for heating. The three-way valve 29 can directly transport it to the heat recovery box 26. When its temperature is higher than the set temperature, it means that it is still worth using again. The three-way valve 29 allows the circulating water to flow into the water tank 24 again through the reflux pipe 31 for secondary utilization until its temperature is lower than the set value, thereby improving its thermal energy utilization rate. By designing the heat recovery component, it is convenient to collect the heat generated on the surface of the ground source heat pump body 1 when it is running. The heat energy collected by the heat recovery component is used to heat and insulate the pipes of the ground source heat pump body 1 in the form of hot steam, which can reduce energy loss.

[0102] See also Figure 1-Figure 7 ,The intelligent energy management module includes the control center, sensor nodes, data ,acquisition and analysis module, intelligent control algorithm, user ,interface and sensor nodes.

[0103] By adopting the above technical solutions, the control center: as the core of the whole system, the control center is responsible for receiving data from each sensor node and processing it through the data acquisition and analysis module. It uses high-performance computers or servers to ensure the real-time and accuracy of data processing; sensor nodes: these nodes are distributed in key positions of the ground source heat pump system, such as underground heat exchangers, heat pump units, pipeline systems, etc., and are used to monitor the system's key parameters such as temperature, pressure, flow, etc. in real time. The sensor nodes communicate with the control center through wired or wireless methods to ensure real-time transmission of data; data acquisition and analysis module: this module is responsible for collecting data from sensor nodes, and preprocessing, cleaning and verification to ensure the accuracy and integrity of the data. Subsequently, advanced data analysis algorithms, such as machine learning and deep learning, are used to conduct in-depth analysis of the data to explore the operating laws and potential problems of the system; intelligent control algorithm: based on the results of the data acquisition and analysis module, the intelligent control algorithm can automatically adjust the operating status of the ground source heat pump system, such as the opening / closing of the heat pump unit, the speed of the water pump, the underground The heat exchange efficiency of the heat exchanger, etc., in order to maximize the utilization of energy and optimize the operation of the system. The algorithm can also adjust the system strategy in advance according to factors such as weather forecasts and user behavior predictions to ensure efficient operation under different working conditions. User interaction interface: provides an intuitive and easy-to-use user interface, allowing users to view the system's operating status, energy consumption data, alarm information, etc. in real time. Users can also set system parameters, adjust operating strategies, and even perform remote control and fault diagnosis through the interface. In addition, the user interaction interface can also provide energy consumption reports, energy-saving suggestions and other functions to help users better understand and manage their energy use; the intelligent ground source heat pump management device provided by the present invention realizes comprehensive and intelligent management of the ground source heat pump system through an integrated control center, high-precision sensor nodes, powerful data acquisition and analysis capabilities, intelligent control algorithms and friendly user interaction interfaces. The system can not only significantly improve energy utilization efficiency and reduce operating costs, but also improve user comfort and satisfaction, providing new ideas and methods for the development and application of ground source heat pump technology.

[0104] The working principle of the present invention is as follows: during daily operation of the ground source heat pump body 1, the control panel 5 displays the operating parameters of the ground source heat pump body 1 through the observation window 23, which is convenient for the operator to observe. When it is necessary to approach the ground source heat pump body 1 to adjust the parameters, the output shaft of the dual-axis motor 8 drives the rotating shaft 11 to rotate through the first gear 9 and the second gear 10, and the rotating shaft 11 drives the protective cover 4 to rotate ninety degrees. Then the output shaft of the small motor 13 rotates on the tooth plate through the third gear 14 to drive the mounting plate 12 to move, and the mounting plate 12 drives the control panel 5 to move outward, so that the control panel 5 is close to the operator, so that there is a certain distance between the operator and the ground source heat pump body 1, which solves the problem in the prior art that the operator cannot avoid the need to approach the ground source heat pump when adjusting the parameters, and the operator is easy to contact the ground source heat pump and cause injury. The temperature sensor 7 is convenient for real-time monitoring of the inner cavity temperature of the control box 3, so that it is convenient for it to control the working efficiency of the corresponding cooling fan through the control panel 5, and the cooling fan 20 passes the outside air into the control box The inner cavity of the control box 3 is convenient for cooling the internal electronic components and improving the heat dissipation effect. The dustproof net 21 can prevent external dust from entering the control box 3 to ensure the heat dissipation effect of the control box 3. By designing one side of the heat recovery box 26 to be close to the surface of the ground source heat pump body 1, the heat generated when the ground source heat pump body 1 is running can be efficiently recovered. By designing a one-way valve 34, the one-way valve 34 allows the liquid to flow in one direction, which can prevent the high-temperature circulating water in the water tank 24 from flowing directly into the heat recovery box 26 through the return pipe 31. By designing the heat recovery pipe 32 to be arranged in a U shape in the inner cavity of the water tank 24, the time for the gas to flow inside the water tank 24 can be increased, and the efficiency ratio of heat recovery can be further increased. Through the cooperation of the circulation motor 35 and the blades 36, the output shaft of the circulation motor 35 drives the blades 36 to rotate, and the blades 36 drive the water flow, so that the circulating water temperature in the inner cavity of the water tank 24 is within a rough range, avoiding the problem that the temperature of the gas discharged from the guide column heat recovery pipe 32 is high and low due to different internal water temperatures.

[0105] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to specific implementation methods. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that technicians in the relevant technical field can better understand and utilize the present invention.

Claims

1. Intelligent ground source heat pump management method, characterized in that: include: S1: Obtaining the injection temperature characteristics and injection flow characteristics of the geothermal well inlet, the first temperature characteristics and the first pressure characteristics of the geothermal well outlet, and the corresponding time characteristics; S2: normalizing the injection temperature characteristic, injection flow characteristic and time characteristic to obtain an injection data set, and normalizing the corresponding first temperature characteristic and first pressure characteristic to obtain a first outlet data set; S3: inputting the injection data set and the first outlet data set into a preset large model and training the large model to predict the temperature and pressure of the geothermal well outlet; S4: After the training of the large model is completed, the injection port data set is input into the trained large model, and the large model outputs a second outlet data set; S5: Inverse normalize the second outlet data set to obtain the second temperature characteristic and the second pressure characteristic of the geothermal well outlet predicted by the large model.

2. The intelligent ground source heat pump management method according to claim 1, characterized in that: The normalized expression is Among them, the X norm represents the normalized features, Indicates the injection temperature characteristics of the geothermal well inlet, Indicates the injection flow characteristics of the geothermal well inlet, represents the geothermal well outlet temperature characteristics, represents the geothermal well outlet pressure characteristics, t represents the time characteristics, X represents the variables in the data set, X min and X max are the minimum and maximum values ​​of the features respectively, and the X norm The range is between 0 and 1.

3. The intelligent ground source heat pump management method according to claim 1, characterized in that: The large model adopts an artificial neural network framework, and the large model includes an input layer, a hidden layer and an output layer. The input layer is provided with three nodes, including injection temperature characteristics, injection flow characteristics and corresponding time characteristics; the hidden layer includes a first hidden layer, a second hidden layer, a third hidden layer, a fourth hidden layer and a fifth hidden layer. The first hidden layer includes 8 neurons and is connected to the input layer, the second hidden layer includes 16 neurons and is connected to the first hidden layer, the third hidden layer includes 32 neurons and is connected to the second hidden layer, the fourth hidden layer includes 16 neurons and is connected to the third hidden layer, and the fifth hidden layer includes 8 neurons and is connected to the fourth hidden layer; the output layer includes 2 neurons, which are used to output the second temperature characteristic and the second pressure characteristic to predict the temperature and pressure of the geothermal well outlet.

4. The intelligent ground source heat pump management method according to claim 3, characterized in that: The hidden layer uses ReLU as the activation function, and the expression is ReLU(x)=max(0,x).

5. The intelligent ground source heat pump management method according to claim 1, characterized in that: The large modeling adopts mean square error (MSE) as the loss function. The mean square error measures the error of the model by calculating the square difference between the predicted value and the real-time value. The expression is as follows: in, and They are the real-time injection temperature characteristics and injection flow characteristics of the geothermal well outlet, and is the predicted value of the second temperature characteristic and the second pressure characteristic of the geothermal well calculated by the large model, and n is the number of samples; a smaller MSE value indicates that the prediction of the large model is accurate.

6. The intelligent ground source heat pump management method according to claim 1, characterized in that: The large model training step includes: S1: Forward propagation, passing the input data to the model, calculating the output layer by layer, and finally getting the prediction result. The mathematical expression of forward propagation is as follows: a (l) =f(W (l) a (l-1) +b (l) ) Where a(l) represents the activation value of the lth layer, W(l) and b(l) are the weight and bias of the lth layer respectively, and f is the activation function; S2: Calculate the loss and use the loss function (MSE) to calculate the error between the predicted value and the real-time value; S3: Back propagation, calculate the partial derivative of the weight of each layer according to the loss, and use the chain rule to calculate the gradient layer by layer; the mathematical expression of the back propagation is as follows: Among them, L is the loss function and is the gradient of the lth layer. S4: Weight update: Use the optimizer to update the weight of the network based on the calculated gradient to minimize the value of the loss function. The mathematical expression of the weight update is: Among them, η is the learning rate.

7. The intelligent ground source heat pump management method according to claim 6, characterized in that: After the training of the large model is completed, the real-time features and the training features are normalized in the same way, and the normalized data is input into the trained large model. The output expression of the trained large model is: in, represents the normalized second temperature characteristic of the geothermal well outlet predicted by the large model, represents the normalized second pressure characteristic of the geothermal well outlet predicted by the large model, t new Represents the real-time or large-model training geothermal well inlet injection temperature characteristics, Represents the temporal features of real-time or large model training, Represents the real-time or large model training geothermal well inlet injection flow characteristics.

8. The intelligent ground source heat pump management method according to claim 7, characterized in that: The expression of the inverse normalization of the large model is: in, The second temperature characteristic representing the physical quantity at the geothermal well outlet predicted by the large model, A second pressure characteristic representing a physical quantity at the geothermal well outlet predicted by the large model.

9. An intelligent ground source heat pump management system, used to implement the intelligent ground source heat pump management method according to any one of claims 1 to 8, characterized in that: Includes the following modules: Acquisition module: obtain the injection temperature characteristics and injection flow characteristics of the geothermal well inlet, the first temperature characteristics and the first pressure characteristics of the geothermal well outlet, and the corresponding time characteristics; Preprocessing module: normalizing the injection temperature feature, injection flow rate feature and time feature to obtain an injection data set, and normalizing the corresponding first temperature feature and first pressure feature to obtain a first outlet data set; Large model module: inputs the injection data set into a preset large model, and outputs a second outlet data set for predicting the temperature and pressure of the geothermal well outlet; Inverse processing module: inverse normalizes the second outlet data set to obtain the predicted second temperature characteristic and second pressure characteristic of the geothermal well outlet.

10. An intelligent ground source heat pump management device, comprising a ground source heat pump body (1), a main box (2) and an intelligent energy management module, characterized in that: A main box (2) is arranged on the top of the ground-source heat pump body (1), a control box (3) is arranged on the front of the main box (2), a protective flip assembly is arranged on the bottom of the control box (3), a protective cover (4) is arranged on one side of the protective flip assembly, a control panel (5) is arranged in the inner cavity of the control box (3), and movable components are arranged on both the left and right sides of the control panel (5), and a heat recovery component is arranged on the left side of the ground-source heat pump body (1); The protective flip assembly comprises a chassis (6), the chassis (6) is fixedly connected to the bottom of the control box (3), the inner cavity of the chassis (6) is fixedly connected to a dual-axis motor (8), the output shaft of the dual-axis motor (8) is fixedly connected to a first gear (9), the surface of the first gear (9) is meshed with a second gear (10), and the axis of the second gear (10) is fixedly connected to a rotating shaft (11); The moving assembly comprises a mounting plate (12), the mounting plate (12) being fixed on the left and right sides of the control panel (5), the top of the mounting plate (12) being fixedly connected to a small motor (13), the output shaft of the small motor (13) being fixedly connected to a third gear (14), the surface of the third gear (14) being meshed with a main body tooth plate (15); The intelligent energy management module includes a control center, sensor nodes, a data acquisition and analysis module, an intelligent control algorithm, a user interaction interface and sensor nodes.

Images