Method and device for monitoring water flux of variable saturated soil from soil moisture content data
By constructing a computing space-time interval and training a neural network framework, combining weak priors and strong prior knowledge, the problem of soil water flux monitoring in complex situations where unsaturated and saturated states alternately is solved, and the soil water flux is achieved Effective monitoring.
Patent Information
- Application Number
- CN202410639918.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-22
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-05-22
AI Technical Summary
The existing soil water flux monitoring methods cannot effectively solve the problem of water flux monitoring in complex situations where soil alternates between unsaturated and saturated states, and lack the ability to reason for boundary extrapolation, so it is impossible to achieve water flux in the complete soil profile. monitor.
By constructing a computed spatiotemporal interval, a pre-trained neural network framework is trained based on the soil moisture content data set, and retrained in local and global intervals, combining weak priors and strong prior knowledge to generate soil water flux data.
Soil water flux monitoring is achieved in the situation of alternating changes of soil without saturation, meeting the monitoring needs of variable saturated soil water flux and avoiding the problem of fragmentation between initial and boundary conditions.
Smart Images

Figure CN118467890B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of soil water migration monitoring, and in particular to a method and device for monitoring variable saturated soil water flux from soil moisture content data. Background Art
[0002] The monitoring of soil water flux plays a huge role in agricultural production, ecological construction, water resources management, and geological and meteorological disaster forecasting. However, there is no direct method to measure soil water flux in reality, nor is there a universal water flux monitoring device. The water and soil environment in various regions of my country varies greatly, which not only has an urgent need for soil water flux monitoring, but also brings huge challenges to universal soil water flux monitoring.
[0003] Soil water migration is a highly nonlinear process that involves the interaction of soil hydraulic conductivity, water potential, moisture content and other factors. Currently, the method for monitoring soil water flux can use the instantaneous profile method, but its practical application and development are limited by reasons such as the mismatch of the measured soil constitutive relationship scale, the inconsistency between the control equation and the actual situation, sampling disturbance and complex noise interference; at the same time, the thermal pulse method also has great potential, but due to soil heterogeneity, soil surface temperature drift and other reasons, this method cannot monitor the water flux of the entire soil profile and lacks the ability to extrapolate the boundary. More importantly, the current water flux monitoring work is often divided into two major modes: soil saturated flow and unsaturated flow. The existing methods cannot solve the problem of water flux monitoring in complex situations where the soil alternates between unsaturated and saturated states. Summary of the invention
[0004] In view of this, the present application provides a method and device for monitoring variable saturated soil water flux from soil moisture content data, which can realize soil water flux monitoring under variable saturation scenarios.
[0005] In a first aspect, the present application provides a method for monitoring saturated soil water flux from soil moisture content data, comprising:
[0006] Constructing a calculation space-time interval based on a soil moisture data set, wherein the calculation space-time interval is divided into a global calculation space-time interval and a local calculation space-time interval, wherein the elements constituting the soil moisture data set are soil moisture and its corresponding time information and space information, and the calculation space-time interval is formed by the distribution range of the time information and the distribution range of the space information;
[0007] Constructing a pre-trained neural network framework for variable saturated soil water flux, and training the pre-trained neural network framework based on weak prior knowledge constraints and within the local computational time and space interval;
[0008] Retraining the trained pre-trained neural network framework based on strong prior knowledge constraints and within the local computational time and space interval;
[0009] Constructing a neural network framework for variable saturated soil water flux, and training the neural network framework in the global calculation time and space interval according to the retraining result of the pretrained neural network framework;
[0010] The designated spatiotemporal position belonging to the global calculation spatiotemporal interval is input into the trained neural network framework to generate the soil water flux corresponding to the designated spatiotemporal position.
[0011] Optionally, locally calculate the spatiotemporal interval D p For z∈[0, z inter ], where z inter It is expressed as:
[0012]
[0013] Where #{·} is the counting function; θ s is the saturated volumetric moisture content of soil, θ(t, z) is the volumetric moisture content of soil, min〔.〕is the minimum value, t is the time information, z is the spatial information, D g The global calculation time interval.
[0014] Optionally, the measurement points within the calculation time-space interval are expressed in the following formula:
[0015]
[0016] P m_g ={(t j , z j )|χ D (t, z)=1}, j=1, 2, 3..., N m_g ;
[0017] P a_p , P a_g =LHS(D p , N a_p ), LHS(D g , N a_g );
[0018] In the above formula, P m_p , P a_p are all sets of measurement points in the local computation space, N m_p For the set P m_p The number of elements, N a_p For the set P a_g The number of elements, P m_g , P a_gare all sets of measurement points in the global computation space, N m_g For the set P m_g The number of elements, N a_g For the set P a_g The number of elements of Ω (·) is the characteristic function of spacetime Ω; LHS(Ω,N) is N times of super Latin cube sampling in spacetime Ω.
[0019] Optionally, the pre-trained neural network framework is expressed in the following formula:
[0020]
[0021]
[0022]
[0023] In the formula, A pre-trained neural network framework for matrix potential; Pre-trained neural network framework for soil moisture content; Pre-trained neural network framework for soil hydraulic conductivity, is the weak prior knowledge constraint of soil hydraulic conductivity; t is the time information, z is the spatial information; W, B are the update parameters that need to be updated in the corresponding network;
[0024] And / or, the loss function corresponding to training the pre-trained neural network framework is expressed by the following formula,
[0025] Loss = w RRE L RRE +w θ L θ ;
[0026]
[0027]
[0028] Where Loss is the loss function of the pre-trained neural network framework; w RRE is the RRE residual coefficient; L RRE is the RRE collocation residual; w θ is the moisture content prediction error coefficient; L θ is the moisture content prediction error; l(·) is the error function; (t i ,z i )=P m_p is the local calculation time and space interval measurement point; (t j ,z j )=P a_p Assign points to local computational time and space intervals;
[0029] And / or, the update parameters corresponding to training the pre-trained neural network framework are expressed in the following formula:
[0030]
[0031] In the above formula, W, B are updated parameters, and Loss is the neural network loss function.
[0032] Optionally, a strong prior knowledge constraint on soil moisture content It is expressed in the following formula:
[0033]
[0034]
[0035] In the above formula, θ r is the residual volume moisture content of soil; K S_sqr , n is the parameter to be updated;
[0036] And / or, the update parameters corresponding to the retraining pre-trained neural network framework are expressed as follows:
[0037]
[0038] In the above formula, W, B, K S_sqr , n is the update parameter corresponding to the retraining pre-trained neural network framework, and Loss is the loss function of the pre-trained neural network.
[0039] Optionally, the neural network framework is expressed in the following formula:
[0040]
[0041]
[0042]
[0043] In the formula, It is a matrix potential neural network framework; is the soil moisture content neural network; is the soil hydraulic conductivity neural network; t is the time information, z is the space information; W, B, K S_sqr , n is the parameter that needs to be updated for the corresponding network; K S_sqr , n is the initial position.
[0044] Optionally, the soil water flux corresponding to the specified time and space position is generated, which is represented by the following formula:
[0045]
[0046] In the formula, is the soil water flux, is the matrix potential neural network framework, t ori is the time information expressed in coordinates, z ori It is the spatial information expressed in coordinates.
[0047] In a second aspect, the present application provides a device for monitoring saturated soil water flux from soil moisture content data, comprising:
[0048] A construction module is used to construct a calculation space-time interval based on a soil moisture data set, wherein the calculation space-time interval is divided into a global calculation space-time interval and a local calculation space-time interval, wherein the elements constituting the soil moisture data set are soil moisture and its corresponding time information and space information, and the calculation space-time interval is formed by the distribution range of the time information and the distribution range of the space information;
[0049] A first training module is used to construct a pre-trained neural network framework for variable saturated soil water flux, and train the pre-trained neural network framework based on weak prior knowledge constraints and within the local calculation time and space interval;
[0050] A retraining module, for retraining the trained pre-trained neural network framework based on strong prior knowledge constraints and within the local computing space-time interval;
[0051] A second training module is used to construct a neural network framework for variable saturated soil water flux, and train the neural network framework in the global calculation time and space interval according to the retraining result of the pre-trained neural network framework;
[0052] A generation module is used to input a designated spatiotemporal position belonging to the global calculation spatiotemporal interval into a trained neural network framework to generate a soil water flux corresponding to the designated spatiotemporal position.
[0053] In a third aspect, the present invention provides a computer-readable storage medium, comprising a program, which, when executed on a computer, enables the computer to execute the method as described above.
[0054] In a fourth aspect, the present application provides an execution device, comprising a processor and a memory, wherein the processor is coupled to the memory;
[0055] The memory is used to store programs;
[0056] The processor is used to execute the program in the memory, so that the execution device executes the above method.
[0057] In the method disclosed in the present application, a sequential training neural network framework in which weak prior constraints and strong prior constraints are interconnected is adopted to reconcile the contradiction between noisy data and incomplete prior knowledge, and to give play to the synergistic and complementary role of the two, thereby eliminating the need to monitor the initial and boundary conditions in the scenario, thereby realizing the monitoring of soil water flux in a situation where the soil is in an alternating unsaturated-saturated state, and meeting the needs of variable saturated soil water flux monitoring. BRIEF DESCRIPTION OF THE DRAWINGS
[0058] The technical solution and other beneficial effects of the present application will be made apparent by describing in detail the specific implementation methods of the present application in conjunction with the accompanying drawings.
[0059] Figure 1 An operational flow chart of a method for monitoring variable saturated soil water flux from soil moisture content data provided by an exemplary embodiment is shown.
[0060] Figure 2 An operation flow chart based on a relatively wide range of usage scenarios provided by an exemplary embodiment is shown.
[0061] Figure 3 A schematic diagram of a neural network sequential training framework structure for calculating variable saturated soil water flux provided by an exemplary embodiment is shown.
[0062] Figure 4 A comparison diagram of forward modeling results of soil moisture field using the physical information neural network of the present application and related technologies provided by an exemplary embodiment is shown.
[0063] Figure 5 A comparison chart of soil water flux field prediction results using the physical information neural network of the present application and related technologies provided by an exemplary embodiment is shown.
[0064] Figure 6 A comparison diagram of soil hydraulic conductivity curve inversion using the present application and a traditional physical information neural network provided by an exemplary embodiment is shown.
[0065] Figure 7 A comparison diagram between the soil boundary water flux calculated using the present application and the true value provided by an exemplary embodiment is shown.
[0066] Figure 8 A block diagram of a device for monitoring saturated soil water flux from soil moisture content data provided by an exemplary embodiment is shown.
[0067] Fig. 9 A block diagram of the execution device provided by an exemplary embodiment is shown. DETAILED DESCRIPTION
[0068] In order to make the above-mentioned purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are described in detail below in conjunction with the accompanying drawings. In the following description, many specific details are set forth to facilitate a full understanding of the present application. However, the present application can be implemented in many other ways different from those described herein, and those skilled in the art can make similar improvements without violating the connotation of the present application, so the present application is not limited by the specific embodiments disclosed below.
[0069] The terms "first", "second", etc. in the specification and claims of this application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments described here can be implemented in an order other than that illustrated or described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or modules is not necessarily limited to those steps or modules clearly listed, but may include other steps or modules that are not clearly listed or inherent to these processes, methods, products or devices. The naming or numbering of steps in this application does not mean that the steps in the method flow must be executed in the time / logical sequence indicated by the naming or numbering. The process steps that have been named or numbered can change the execution order according to the technical purpose to be achieved, as long as the same or similar technical effects can be achieved. The division of units in this application is a logical division. There may be other division methods when it is implemented in actual applications. For example, multiple units can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, and the indirect coupling or communication connection between units can be electrical or other similar forms, which are not limited in this application. In addition, the units or sub-units described as separate components may or may not be physically separated, may or may not be physical units, or may be distributed in multiple circuit units, and some or all of the units may be selected according to actual needs to achieve the purpose of the present application.
[0070] Please refer to Figure 1 ,in Figure 1 A flow chart of a method for monitoring saturated soil water flux from soil moisture content data provided by an exemplary embodiment is shown. The method is implemented through steps 102-108.
[0071] In step 102, a calculation space-time interval is constructed based on the soil moisture data set, and the calculation space-time interval is divided into a global calculation space-time interval and a local calculation space-time interval, wherein the elements of the soil moisture data set are the soil moisture content and its corresponding time information and space information, and the calculation space-time interval is formed by the distribution range of the time information and the distribution range of the space information.
[0072] It should be understood that, since soil moisture content varies with time and spatial position (such as soil depth position, etc.), the elements of the soil moisture data set are soil moisture content and its corresponding time information and spatial information.
[0073] Here, the soil moisture content may be expressed in, but is not limited to, soil volume moisture content (mm / mm).
[0074] Here, the time information may be, but is not limited to, time expressed in coordinate form, i.e., time coordinate t ori , based on the implementation method in which the time information is a time coordinate, the distribution range of the time information can be exemplified as t∈[0,5].
[0075] Spatial information can be, but is not limited to, spatial position expressed in coordinate form, i.e., spatial coordinate z ori , based on the implementation method in which the spatial information is a spatial coordinate, the distribution range of the spatial information can be exemplified as z∈[0,1].
[0076] Please refer to Figure 2 Before constructing the operation of calculating the spatiotemporal interval, the soil moisture data set can also be normalized to eliminate the dimension effect. As an exemplary form, the data normalization of the soil moisture data set is performed using the following formula: ori ∈[-0.05, -0.25, -0.45, -0.65, -0.85], and t ori ∈[0, 0.1736111] as an example,
[0077] t=5·(t ori -min(t ori )) / (max(t ori )-min(t ori )) (1);
[0078] z=1·(z ori -min(z ori )) / (max(z ori )-min(z ori )) (2);
[0079] In formulas (1)-(2), t is the normalized time information, z is the normalized spatial information, and t ori is the time coordinate, z ori is the space coordinate.
[0080] It is worth adding that the above-mentioned calculation time-space interval is used to define the range of measurement points required for training the pre-trained neural network framework, that is, to generate the measurement points required for training the pre-trained neural network framework within the calculation time-space interval.
[0081] It can be easily imagined from the word "global" in the term "global calculation space-time interval" and the word "local" in the term "local calculation space-time interval" that the distribution range of time information and the distribution range of spatial information in the local calculation space-time interval fall within the time information distribution range and the spatial information distribution range of the global calculation space-time interval respectively.
[0082] As a demonstration of the degree of “locality” in the local computation space, the local computation spacetime interval D p For z∈[0, z inter ], where z inter It is expressed as:
[0083]
[0084] Where #{·} is the counting function; θ s is the saturated volumetric moisture content of soil, θ(t, z) is the volumetric moisture content of soil, min〔.〕is the minimum value, t is the time information, z is the spatial information, D g The global calculation time interval.
[0085] As an exemplary embodiment, the measurement points in the calculation time-space interval are expressed in the following formula:
[0086]
[0087] P m_g ={(t j , z j )|χ D (t, z)=1}, j=1, 2, 3..., N m_g (5);
[0088] P a_p , P a_g =LHS(D p , N a_p ), LHS(D g , N a_g ) (6);
[0089] In the above formula, P m_p , P a_pare all sets of measurement points in the local computation space, N m_p For the set P m_p The number of elements, N a_p For the set P a_g The number of elements, P m_g , P a_g are all sets of measurement points in the global computation space, N m_g For the set P m_g The number of elements, N a_g For the set P a_g The number of elements of Ω (·) is the indicative function of spacetime Ω; LHS(Ω, N) is N times of super Latin cube sampling in spacetime Ω.
[0090] In step 104, a pre-trained neural network framework for variable saturated soil water flux is constructed, and the pre-trained neural network framework is trained based on weak prior knowledge constraints and within the local calculation time and space interval.
[0091] Please refer to Figure 3 As an exemplary embodiment, the pre-trained neural network framework is pre-trained in the form expressed by the following formula: Figure 3 Part A,
[0092]
[0093]
[0094]
[0095] In formulas (7)-(9), A pre-trained neural network framework for matrix potential; Pre-trained neural network framework for soil moisture content; Pre-trained neural network framework for soil hydraulic conductivity, is the weak prior knowledge constraint of soil hydraulic conductivity; t is the time information, z is the spatial information; W, B are the update parameters that need to be updated in the corresponding network.
[0096] In the process of training the pre-trained neural network framework, it can be understood that the pre-trained neural network framework performs forward propagation and uses the neural network self-differentiation method to determine the loss function, which can be expressed by the following formula:
[0097] Loss = w RRE L RRE +w θ L θ (10);
[0098]
[0099]
[0100] Where Loss is the loss function of the pre-trained neural network framework; w RRE is the RRE residual coefficient; L RRE is the RRE collocation residual; w θ is the moisture content prediction error coefficient; L θ is the moisture content prediction error; l(·) is the error function; (t i ,z i )=P m_p is the local calculation time and space interval measurement point; (t j ,z j )=P a_p Assign points to a spacetime interval for local computations.
[0101] In the process of training the pre-trained neural network framework, the update parameters are expressed in the following formula:
[0102]
[0103] In the above formula, W, B are updated parameters, and Loss is the neural network loss function.
[0104] In step 106, the trained pre-trained neural network framework is retrained based on the strong prior knowledge constraint and within the local computational spatiotemporal interval.
[0105] It can be understood that here, the strong prior knowledge constraint retraining the trained pre-trained neural network framework is to replace the weak prior knowledge in the pre-trained neural network framework with strong prior knowledge. The strong prior knowledge can be prior knowledge in a strong nonlinear form.
[0106] For details, please refer to Figure 3 Part b of this step 108 use Instead of step 106 in It can be expressed as:
[0107]
[0108]
[0109] In the above formula, θ r is the residual volume moisture content of soil; K S_sqr , n is the parameter to be updated;
[0110] It can be understood that in the process of retraining the pre-trained neural network framework, the pre-trained neural network framework performs forward propagation and uses the neural network self-differentiation method to determine the loss function. This process can also be expressed by formulas (10)-(12), which will not be repeated here.
[0111] As an exemplary embodiment, the update parameters corresponding to the retraining pre-trained neural network framework are expressed in the following formula:
[0112]
[0113] In the above formula, W, B, K S_sqr , n is the update parameter corresponding to the retraining pre-trained neural network framework, and Loss is the loss function of the pre-trained neural network.
[0114] In step 108, a neural network framework of variable saturated soil water flux is constructed, and the neural network framework is trained in the global calculation space-time interval according to the retraining result of the pre-trained neural network framework.
[0115] Please refer to Figure 3 In part c, the neural network framework is expressed by the following formula:
[0116]
[0117]
[0118]
[0119] In the formula, It is the matrix potential neural network framework (e.g. 6*20); is a soil moisture content neural network (e.g., a 1*10 monotonic neural network); is the soil hydraulic conductivity neural network; t is the time information, z is the space information; W, B, K S_sqr , n is the parameter that needs to be updated for the corresponding network; K S_sqr , n is the initial position.
[0120] It can be understood that in the process of training the neural network framework, the neural network framework performs forward propagation and uses the neural network self-differentiation method to determine the loss function. This process can also be represented by formulas (10)-(12) without making some adjustments: (t i ,z i )=P m_g For global calculation of time-space interval measurement points; (t j ,z j )=P a_gThe time-space interval is assigned to the global calculation. Then the parameters are updated according to formula (16).
[0121] In step 110, the designated spatiotemporal position belonging to the global calculation spatiotemporal interval is input into a trained neural network framework to generate the soil water flux corresponding to the designated spatiotemporal position.
[0122] Specifically, choose any t ori ∈(0,0.1736111),z ori ∈[0,-1] time coordinates and space coordinates are input into the model. According to the soil constitutive relations shown in equations (17), (18) and (19) after training, the soil water flux at any position is calculated and can be expressed as:
[0123]
[0124] In the formula, is the soil water flux, is the matrix potential neural network framework, t ori is the time information expressed in coordinates, z ori It is the spatial information expressed in coordinates.
[0125] Taking the soil water flux monitoring in a large soil column with a height of 1 meter and an inner diameter of 30 cm as an example, the soil column is filled with sand and the above-mentioned method of monitoring the variable saturated soil water flux from the soil moisture content data is used. The test results are as follows:
[0126] In addition, this embodiment compares the present invention with the traditional physical information neural network in terms of the ability to coordinate and complement noisy data and incomplete prior knowledge and the ability to predict soil water flux. Figure 4 , Figure 5 As shown, the present invention has more continuous and reasonable results when forward modeling the soil moisture field and predicting the soil water flux field, and greatly avoids the overfitting phenomenon caused by the conflict between data and prior knowledge.
[0127] At the same time, in this embodiment, the boundary conditions of the soil column are controlled, and the real values of the water flux at the upper and lower boundaries are obtained as a comparison of the prediction results of the present invention. The results are as follows: Figure 7 As shown, it can be seen that the present invention has a higher accuracy for flux monitoring.
[0128] Please refer to Figure 8 , which shows a block diagram of the device for monitoring saturated soil water flux from soil moisture content data provided by the present application. The device 200 includes:
[0129] A construction module 202 is used to construct a calculation space-time interval based on a soil moisture data set, wherein the calculation space-time interval is divided into a global calculation space-time interval and a local calculation space-time interval, wherein the elements of the soil moisture data set are soil moisture and its corresponding time information and space information, and the calculation space-time interval is formed by a distribution range of the time information and a distribution range of the space information;
[0130] A first training module 204 is used to construct a pre-trained neural network framework for variable saturated soil water flux, and train the pre-trained neural network framework based on weak prior knowledge constraints and within the local calculation time and space interval;
[0131] A retraining module 206, for retraining the trained pre-trained neural network framework based on strong prior knowledge constraints and within the local computational spatiotemporal interval;
[0132] A second training module 208, for constructing a neural network framework for variable saturated soil water flux, and training the neural network framework in the global calculation time and space interval according to the retraining result of the pretrained neural network framework;
[0133] The generation module 210 is used to input the designated spatiotemporal position belonging to the global calculation spatiotemporal interval into the trained neural network framework to generate the soil water flux corresponding to the designated spatiotemporal position.
[0134] Since the above methods have been discussed in detail, the specific implementation of the above modules will not be repeated here.
[0135] Next, an execution device provided by an embodiment of the present application is introduced. Fig. 9 , Fig. 9 This is a schematic diagram of a structure of an execution device provided in an embodiment of the present application. The execution device 300 can be specifically manifested as an autonomous driving vehicle, a mobile phone, a tablet, a laptop computer, a desktop computer, a monitoring data processing device, etc., which is not limited here. Figure 1 The functions of the execution device in the corresponding embodiment. Specifically, the execution device 300 includes: a receiver 301, a transmitter 302, a processor 303 and a memory 304 (wherein the number of the processor 303 in the execution device 300 can be one or more, Fig. 9 In the example of FIG. 3 , the processor 303 may include an application processor 3031 and a communication processor 3032. In some embodiments of the present application, the receiver 301, the transmitter 302, the processor 303 and the memory 304 may be connected via a bus or other means.
[0136] The memory 304 may include a read-only memory and a random access memory, and provides instructions and data to the processor 303. A portion of the memory 304 may also include a non-volatile random access memory (NVRAM). The memory 304 stores processor and operation instructions, executable modules or data structures, or subsets thereof, or extended sets thereof, wherein the operation instructions may include various operation instructions for implementing various operations.
[0137] The processor 303 controls the operation of the execution device. In a specific application, the various components of the execution device are coupled together through a bus system, wherein the bus system includes not only a data bus but also a power bus, a control bus, and a status signal bus, etc. However, for the sake of clarity, various buses are referred to as bus systems in the figure.
[0138] The method disclosed in the above embodiment of the present application can be applied to the processor 303, or implemented by the processor 303. The processor 303 can be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the hardware integrated logic circuit or software instructions in the processor 303. The above processor 303 can be a general processor, a digital signal processor (digital signal processing, DSP), a microprocessor or a microcontroller, and can further include an application specific integrated circuit (application specific integrated circuit, ASIC), a field programmable gate array (field-programmable gate array, FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components. The processor 303 can implement or execute the various methods, steps and logic block diagrams disclosed in the embodiments of the present application. The general processor can be a microprocessor or the processor can also be any conventional processor, etc. The steps of the method disclosed in the embodiment of the present application can be directly embodied as a hardware decoding processor to execute, or it can be executed by a combination of hardware and software modules in the decoding processor. The software module may be located in a storage medium mature in the art, such as a random access memory, a flash memory, a read-only memory, a programmable read-only memory, or an electrically erasable programmable memory, a register, etc. The storage medium is located in the memory 304, and the processor 303 reads the information in the memory 304 and completes the steps of the above method in combination with its hardware.
[0139] The receiver 301 can be used to receive input digital or character information and generate signal input related to the relevant settings and function control of the execution device. The transmitter 302 can be used to output digital or character information through the first interface; the transmitter 302 can also be used to send instructions to the disk group through the first interface to modify the data in the disk group; the transmitter 302 can also include a display device such as a display screen.
[0140] In the embodiment of the present application, the processor 303 is used to execute Figure 1 The specific manner in which the application processor 3031 in the processor 303 performs the above steps is the same as that in the present application. Figure 1 The corresponding method embodiments are based on the same concept, and the technical effects they bring are the same as those in this application. Figure 1 The corresponding method embodiments are the same. For specific contents, please refer to the description in the method embodiments shown above in this application, which will not be repeated here.
[0141] The above is only a preferred specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by any technician familiar with the technical field within the technical scope disclosed in the present application should be covered within the protection scope of the present application.
Claims
1. A method for monitoring water flux in variably saturated soils from soil moisture content data, It is characterized in that include: Constructing a calculation space-time interval based on a soil moisture data set, wherein the calculation space-time interval is divided into a global calculation space-time interval and a local calculation space-time interval, wherein the elements of the soil moisture data set are soil moisture and its corresponding time information and space information, and the calculation space-time interval is formed by a distribution range of the time information and a distribution range of the space information; Local calculation time and space interval D p For z∈[0, z inter ], where z inter It is expressed as: Where #{·} is the counting function; θ s is the saturated volumetric moisture content of soil, θ(t, z) is the volumetric moisture content of soil, min〔.〕is the minimum value, t is the time information, z is the spatial information, D g It is the global calculation time interval; The measurement points within the calculation time and space interval are expressed by the following formula: P m_g ={(t j ,z j )|χ D (t,z)=1},j=1,2,3...,N m_g ; P a_p ,P a_g =LHS(D p ,N a_p ),LHS(D g ,N a_g ); In the above formula, P m_p , P a_p are all sets of measurement points in the local computation space, N m_p For the set P m_p The number of elements, N a_p , is the set P a_g The number of elements, P m_g , P a_g are all sets of measurement points in the global computation space, N m_g For the set P m_g The number of elements, N a_g For the set P a_g The number of elements of Ω (·) is the characteristic function of space-time Ω; LtS(Ω, N) is N times of super Latin cube sampling in space-time Ω; Constructing a pre-trained neural network framework for variable saturated soil water flux, and training the pre-trained neural network framework based on weak prior knowledge constraints and within the local computational time and space interval; The pre-trained neural network framework is expressed in the following formula: In the formula, A pre-trained neural network framework for matrix potential; Pre-trained neural network framework for soil moisture content; Pre-trained neural network framework for soil hydraulic conductivity, is the weak prior knowledge constraint of soil hydraulic conductivity; t is the time information, z is the spatial information; W, B are the update parameters that need to be updated in the corresponding network; And / or, the loss function corresponding to training the pre-trained neural network framework is expressed by the following formula, Loss=w RRE L RRE +w θ L θ ; Where Loss is the loss function of the pre-trained neural network framework; w RRE is the RRE residual coefficient; L RRE is the RRE collocation residual; w θ is the moisture content prediction error coefficient; L θ is the moisture content prediction error; l(·) is the error function; (t i ,z i )=P m_p is the local calculation time and space interval measurement point; (t j ,z j )=P a_p Assign points to local computational time and space intervals; And / or, the update parameters corresponding to training the pre-trained neural network framework are expressed in the following formula: In the above formula, W, B are updated parameters, and Loss is the neural network loss function; Retraining the trained pre-trained neural network framework based on strong prior knowledge constraints and within the local computational time and space interval; Strong prior knowledge constraints on soil moisture content It is expressed in the following formula: In the above formula, θ r is the residual volume moisture content of soil; K S_sqr , n is the parameter to be updated; And / or, the update parameters corresponding to the retraining pre-trained neural network framework are expressed as follows: In the above formula, W, B, K S_sqr , n is the update parameter corresponding to the retraining pre-trained neural network framework, and Loss is the loss function of the pre-trained neural network; Constructing a neural network framework for variable saturated soil water flux, and training the neural network framework in the global calculation time and space interval according to the retraining result of the pretrained neural network framework; The neural network framework is expressed in the following formula: In the formula, It is a matrix potential neural network framework; is the soil moisture content neural network; is the soil hydraulic conductivity neural network; t is the time information, z is the space information; W, B, K S_sqr , n is the parameter that needs to be updated for the corresponding network; K S_sqr , n is the initial position; Inputting the designated spatiotemporal position belonging to the global calculation spatiotemporal interval into the trained neural network framework to generate the soil water flux corresponding to the designated spatiotemporal position; Generate the soil water flux corresponding to the specified time and space position, which is expressed by the following formula: In the formula, is the soil water flux, is the matrix potential neural network framework, t ori is the time information expressed in coordinates, z ori It is the spatial information expressed in coordinates.
2. A device for monitoring variable saturated soil water flux from soil moisture content data, applied to the method for monitoring variable saturated soil water flux from soil moisture content data as claimed in claim 1, It is characterized in that include: A construction module is used to construct a calculation space-time interval based on a soil moisture data set, wherein the calculation space-time interval is divided into a global calculation space-time interval and a local calculation space-time interval, wherein the elements of the soil moisture data set are soil moisture and its corresponding time information and space information, and the calculation space-time interval is formed by a distribution range of the time information and a distribution range of the space information; A first training module is used to construct a pre-trained neural network framework for variable saturated soil water flux, and train the pre-trained neural network framework based on weak prior knowledge constraints and within the local calculation time and space interval; A retraining module, for retraining the trained pre-trained neural network framework based on strong prior knowledge constraints and within the local computing space-time interval; A second training module is used to construct a neural network framework for variable saturated soil water flux, and train the neural network framework in the global calculation time and space interval according to the retraining result of the pre-trained neural network framework; A generation module is used to input a designated spatiotemporal position belonging to the global calculation spatiotemporal interval into a trained neural network framework to generate a soil water flux corresponding to the designated spatiotemporal position.
3. A computer-readable storage medium, It is characterized in that The invention comprises a program, which, when being run on a computer, causes the computer to execute the method as claimed in claim 1.
4. An execution device, It is characterized in that comprising a processor and a memory, wherein the processor is coupled to the memory; The memory is used to store programs; The processor is configured to execute the program in the memory so that the execution device executes the method according to claim 1.
