Method and device for determining soil humidity of mining area and soil humidity detection system

Abstract

The invention provides a method and a device for determining soil humidity of a mining area and a soil humidity detection system. The method comprises the steps that related data of a mining area are acquired, the related data comprise one or more of meteorological data, geographical environment data and remote sensing image data, the meteorological data comprise one or more of rainfall, environment temperature and environment humidity, and the geographical environment data comprise one or more of soil type, soil texture and terrain; the remote sensing image data is image data of the mining area earth surface obtained through a remote sensing technology; constructing a soil humidity analysis model; and inputting the related data into the soil humidity analysis model to obtain the soil humidity corresponding to the related data. According to the scheme, the problem that the accuracy of the measurement result of the soil humidity in the mining area is poor in the prior art is solved.

Description

Technical Field

[0001] The present application relates to the technical field of coal mine soil detection, and in particular, to a method, device, computer program product and soil moisture detection system for determining soil moisture in a mining area. Background Art

[0002] Accurate measurement of soil moisture in mining areas is of great significance for environmental management and resource development in mining areas. Soil moisture is one of the key indicators for evaluating sustainable land use and water resource management. It not only directly affects processes such as vegetation growth, soil erosion and water cycle in the soil, but is also closely related to the ecological environment and groundwater resources in mining areas. However, traditional methods for calculating soil moisture in mining areas have some challenges and shortcomings.

[0003] First, traditional methods usually use direct measurement or indirect estimation to obtain soil moisture data. These traditional methods often require a lot of field observations and manual sampling, which is time-consuming, labor-intensive and costly. This field observation-based method cannot provide instant and continuous soil moisture information, limiting the real-time nature of mining area environmental management.

[0004] However, the traditional soil moisture calculation method has limited spatial coverage within the mining area and cannot provide comprehensive and continuous soil moisture information. The real soil moisture data is usually concentrated in a limited number of monitoring points and cannot meet the soil moisture requirements in different areas and at different scales within the mining area. In a vast mining environment, this lack of spatial coverage limits the comprehensive assessment and monitoring of soil moisture in the mining area.

[0005] In addition, existing soil moisture calculation methods also have some limitations in feature extraction and data processing. Traditional feature extraction methods are often oversimplified or rely on manually specified features, ignoring the potential patterns and correlations in the data. Since soil moisture data usually has high dimensions and complex spatial correlations, traditional calculation methods have difficulty capturing and utilizing this information, thus affecting the accuracy of the calculation results. Summary of the invention

[0006] The main purpose of the present application is to provide a method, device, computer program product and soil moisture detection system for determining soil moisture in a mining area, so as to at least solve the problem of poor accuracy of the measurement results of soil moisture in a mining area in the prior art.

[0007] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a method for determining soil moisture in a mining area is provided, comprising: obtaining relevant data of the mining area, wherein the relevant data include one or more of meteorological data, geographical environment data and remote sensing image data, the meteorological data include one or more of rainfall, ambient temperature and ambient humidity, the geographical environment data include one or more of soil type, soil texture and topography, and the remote sensing image data is image data of the surface of the mining area obtained by remote sensing technology; constructing a soil moisture analysis model, wherein the soil moisture analysis model is trained by a convolution algorithm and an LSTM algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes historical relevant data obtained within a historical time period and historical soil moisture corresponding to the historical relevant data; inputting the relevant data into the soil moisture analysis model to obtain soil moisture corresponding to the relevant data.

[0008] Optionally, obtaining relevant data of the mining area includes: obtaining initial relevant data of the mining area; performing data preprocessing on the initial relevant data to obtain the relevant data, wherein the data preprocessing method includes one or more of filling missing values, removing noise data, and eliminating abnormal values.

[0009] Optionally, before constructing the soil moisture analysis model, the method also includes: obtaining information of a first network structure, wherein the first network structure is constructed based on a convolution algorithm, and the first network structure is used to extract features of the relevant data by downsampling and convolution; obtaining information of a second network structure, wherein the second network structure is constructed based on an improved LSTM algorithm, and the improved LSTM algorithm is obtained by improving the LSTM algorithm through feature inheritance and feature selective forgetting mechanisms, and the second network structure is used to adjust the degree of retention of the extracted features; coupling the information of the first network structure and the information of the second network structure to obtain the information of the soil moisture analysis model.

[0010] Optionally, obtaining information about the second network structure includes: calculating information about a forget gate according to a first formula, wherein the first formula is: Z f =σ(W f ×[h t-1 ,h t-2 ,...,h t-n ,X t ]+b f ), Z f represents the information of the forget gate, σ represents the sigmoid function, W f represents the weight matrix of the forget gate, h t-n represents the feature vector, X tThe input value at the current moment, b f represents the bias term of the forget gate; and calculates the information of the memory gate according to the second formula, wherein the second formula is: Z i =σ(W i ×[h t-1 ,h t-2 ,...,h t-n ,X t ]+b i ), Z i Represents the information of the memory gate, W i represents the weight matrix of the memory gate, b i represents the bias term of the memory gate.

[0011] Optionally, obtaining the information of the second network structure includes: calculating the parameter of the current input unit state according to a third formula, wherein the third formula is: Z = tanh (W c ×[h t-1 ,h t-2 ,...,h t-n ,X t ]+b c ), Z represents the parameter of the unit state currently input, W c represents the weight matrix of the unit state of the current input, b c The bias term representing the current input unit state; the information of the unit state at the current moment is calculated according to the fourth formula, wherein the fourth formula is: t =Z f ×mean(c t-1 ,c t-2 ,...,c t-n )+Z i ×Z,c t Information indicating the state of the unit at the current moment.

[0012] Optionally, obtaining information about the second network structure includes: calculating information about the output gate according to a fifth formula, wherein the fifth formula is: Z o =σ(W o ×[h t-1 ,h t-2 ,...,h t-n ,X t ]+b o ), Z o Represents the information of the output gate, W o represents the weight matrix of the output gate, b o represents the bias term of the output gate; and calculating the output value of the second network structure according to the sixth formula, wherein the sixth formula is: h t =Zo ×tanc(c t ), h t Represents the output value of the second network structure.

[0013] Optionally, before constructing the soil moisture analysis model, the method further includes: constructing an initial soil moisture analysis model, wherein the initial soil moisture analysis model is trained using a plurality of sets of training data through a convolution algorithm and an LSTM algorithm, and each set of the plurality of training data includes the historical relevant data acquired within the historical time period and the historical soil moisture corresponding to the historical relevant data; and optimizing the initial soil moisture analysis model using Adam optimization and gradient descent to obtain the soil moisture analysis model.

[0014] According to another aspect of the present application, a device for determining soil moisture in a mining area is provided, comprising: a first acquisition unit, used to acquire relevant data of the mining area, wherein the relevant data include one or more of meteorological data, geographical environment data and remote sensing image data, the meteorological data include one or more of rainfall, ambient temperature and ambient humidity, the geographical environment data include one or more of soil type, soil texture and topography, and the remote sensing image data is image data of the surface of the mining area obtained by remote sensing technology; a first construction unit, used to construct a soil moisture analysis model, wherein the soil moisture analysis model is obtained by training with a convolution algorithm and an LSTM algorithm using multiple sets of training data, each set of the multiple sets of training data includes historical relevant data obtained within a historical time period and historical soil moisture corresponding to the historical relevant data; a first processing unit, used to input the relevant data into the soil moisture analysis model to obtain soil moisture corresponding to the relevant data.

[0015] According to another aspect of the present application, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the steps of any one of the methods for determining soil moisture in a mining area are implemented.

[0016] According to another aspect of the present application, a soil moisture detection system is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for determining the soil moisture of any one of the mining areas.

[0017] By applying the technical solution of the present application, a soil moisture analysis model is built. The soil moisture analysis model is trained based on the convolution algorithm and the LSTM algorithm. The convolution algorithm can well extract useful features from the data. The LSTM algorithm can solve the problems of gradient vanishing and gradient explosion that occur in long time series, and can capture the dependency of data in the time series. Therefore, this solution combines the convolution algorithm with the LSTM algorithm to extract effective data features and make full use of the relationship between the time series in the data, thereby ensuring good accuracy of the measurement results. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The drawings constituting part of the present application are used to provide a further understanding of the present application. The exemplary embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0019] Figure 1 A hardware structure block diagram of a mobile terminal for executing a method for determining soil moisture in a mining area provided in an embodiment of the present application is shown;

[0020] Figure 2 A schematic flow chart of a method for determining soil moisture in a mining area provided in accordance with an embodiment of the present application is shown;

[0021] Figure 3 A schematic diagram of the process of soil moisture calculation using the Con-Fisfnet model is shown;

[0022] Figure 4 A schematic diagram showing the overall framework of Fisfnet;

[0023] Figure 5 The schematic diagram of the architecture of the Con-Fisfnet model is shown;

[0024] Figure 6 A structural block diagram of a device for determining soil moisture in a mining area provided according to an embodiment of the present application is shown.

[0025] The above drawings include the following reference numerals:

[0026] 102, processor; 104, memory; 106, transmission device; 108, input and output devices. DETAILED DESCRIPTION

[0027] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0028] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0029] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application 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 comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0030] As introduced in the background technology, the results of the measurement of soil moisture in mining areas in the prior art are less accurate. To solve the above problems, the embodiments of the present application provide a method, device, computer program product and soil moisture detection system for determining soil moisture in a mining area.

[0031] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0032] The method embodiments provided in the embodiments of the present application can be executed in a mobile terminal, a computer terminal or a similar computing device. Taking running on a mobile terminal as an example, Figure 1 FIG. 1 is a hardware structure block diagram of a mobile terminal of a method for determining soil moisture in a mining area according to an embodiment of the present invention. Figure 1 As shown, the mobile terminal may include one or more ( Figure 1 Only one is shown in the figure) a processor 102 (the processor 102 may include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data, wherein the mobile terminal may also include a transmission device 106 and an input / output device 108 for communication functions. It can be understood by those skilled in the art that Figure 1 The structure shown is only for illustration and does not limit the structure of the mobile terminal. Figure 1 More or fewer components as shown, or with Figure 1 Different configurations are shown.

[0033] The memory 104 can be used to store computer programs, for example, software programs and modules of application software, such as the computer program corresponding to the display method of device information in the embodiment of the present invention. The processor 102 executes various functional applications and data processing by running the computer program stored in the memory 104, that is, the above method is implemented. The memory 104 may include a high-speed random access memory, and may also include a non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some examples, the memory 104 may further include a memory remotely arranged relative to the processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the above-mentioned network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof. The transmission device 106 is used to receive or send data via a network. The above-mentioned specific examples of the network may include a wireless network provided by a communication provider of the mobile terminal. In one example, the transmission device 106 includes a network adapter (Network Interface Controller, referred to as NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0034] In this embodiment, a method for determining soil moisture in a mining area is provided that runs on a mobile terminal, a computer terminal or a similar computing device. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in an order different from that shown here.

[0035] Figure 2 FIG. 1 is a flow chart of a method for determining soil moisture in a mining area according to an embodiment of the present application. Figure 2 As shown, the method comprises the following steps:

[0036] Step S201, obtaining relevant data of the mining area, wherein the relevant data includes one or more of meteorological data, geographical environment data and remote sensing image data, the meteorological data includes one or more of rainfall, ambient temperature and ambient humidity, the geographical environment data includes one or more of soil type, soil texture and topography, and the remote sensing image data is image data of the mining area surface obtained by remote sensing technology;

[0037] Specifically, a soil moisture characteristic data set for mining areas can be produced. Long-term series data related to soil moisture calculation (soil moisture measurement data, meteorological data, geographic environment data, remote sensing image data) are collected; soil moisture measurement data is used for model training and testing, and the soil moisture at each measurement point is used as a label for training samples. Soil moisture data can be obtained through long-distance sensors such as satellites or airplanes, and soil moisture can be measured using microwave or infrared radiation and other technologies. Meteorological data include information related to soil moisture, such as rainfall, temperature, and relative humidity. Rising temperature accelerates the evaporation of water in the soil, resulting in a decrease in soil moisture; at lower temperatures, the rate of water evaporation is relatively slow, and soil moisture is relatively high. Rainfall is another important factor, which can increase soil moisture through the nourishment of rainwater, but if the rainfall is too large or frequent and exceeds the water holding capacity of the soil, it will also affect soil moisture. Geographic environment data include geographical environment information related to soil moisture, such as soil type, soil texture, and topography. Through this information, the complex relationship between soil moisture and other environmental factors (such as temperature, topography, etc.) can be captured, and soil moisture can be calculated more accurately. Continuous long-term series data helps the model understand the spatiotemporal evolution of soil moisture. The acquired data are put into the soil moisture calculation database to construct the soil moisture characteristic data set of the mining area.

[0038] Step S202, constructing a soil moisture analysis model, wherein the soil moisture analysis model is trained by a convolution algorithm and an LSTM algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes historical related data obtained in a historical time period and historical soil moisture corresponding to the historical related data;

[0039] Specifically, the convolution algorithm and the LSTM algorithm can be used to perform machine learning training to obtain a soil moisture analysis model.

[0040] Use CNN to extract features of time series data: First, the time series data is converted into a two-dimensional input form, and then features are extracted through convolutional layers and pooling layers. This can effectively capture local patterns and spatial associations in time series data.

[0041] Input the feature sequence extracted by CNN into LSTM: Using the feature sequence extracted by CNN as the input of LSTM can help LSTM better learn the long-term dependencies in time series data. LSTM has mechanisms such as memory units and forget gates, which can effectively capture long-term dependencies in time series data.

[0042] Combine CNN and LSTM for training and prediction: During the training process, CNN and LSTM can be connected together to build an end-to-end model, and the parameters of CNN and LSTM can be updated simultaneously through the back propagation algorithm. During the prediction process, new time series data is input into the model to obtain the prediction results for future moments.

[0043] Step S203: input the above-mentioned relevant data into the above-mentioned soil moisture analysis model to obtain the soil moisture corresponding to the above-mentioned relevant data.

[0044] Specifically, the relevant data can be input into a trained soil moisture analysis model, and then the output of the soil moisture analysis model is obtained, and the soil moisture corresponding to the relevant data is obtained.

[0045] Through this embodiment, a soil moisture analysis model is built. The soil moisture analysis model is trained based on the convolution algorithm and the LSTM algorithm. The convolution algorithm can well extract useful features in the data. The LSTM algorithm can solve the problems of gradient disappearance and gradient explosion that occur in long time series, and can capture the dependency of data in the time series. Therefore, this solution combines the convolution algorithm with the LSTM algorithm to extract effective data features and make full use of the relationship between the time series in the data, thereby ensuring good accuracy of the measurement results.

[0046] In order to ensure the validity and accuracy of the data, the relevant data of the mining area can be obtained through the following steps: obtaining the initial relevant data of the above-mentioned mining area; performing data preprocessing on the above-mentioned initial relevant data to obtain the above-mentioned relevant data, wherein the data preprocessing method includes one or more of filling missing values, removing noise data, and eliminating outliers.

[0047] In this solution, the initial relevant data can be preprocessed to fill in missing values, remove noise data, and eliminate outliers, which can ensure the validity of the data and ensure that the data is relatively accurate.

[0048] Specifically, this scheme can produce a mining area soil moisture characteristic data set; collect data related to soil moisture calculation (mining area remote sensing images, meteorological data, soil moisture measured data, etc.), put the data in the above steps into the soil moisture calculation database, and construct a mining area soil moisture characteristic data set; check the missing values, outliers or noise in the data, fill the missing values ​​by interpolation and other means, and reduce the noise by smoothing or filtering technology, and divide the data into training set, validation set and test set. The ratio of training set, validation set and test set can be 7:1:2.

[0049] In some existing schemes, with the rapid development of remote sensing technology and deep learning, people can obtain a large number of high-resolution remote sensing images, so that the automatic recognition method based on high-resolution remote sensing images gradually replaces the traditional manual method. The soil moisture calculation method that integrates remote sensing technology and deep learning has more advantages than the traditional method. It can make full use of the multidimensional, panoramic and real-time characteristics of remote sensing data, provide accurate and high-resolution soil moisture prediction, and realize automated processing and intelligent analysis, providing a convenient and efficient solution for soil moisture research.

[0050] Among some existing solutions, RNN (recurrent neural network) is a neural network structure commonly used in sequence data processing. In soil moisture calculation, the application of RNN network has some advantages. First, soil moisture is time-series data. RNN network can handle this time series data well and can capture the impact of information from past time steps on the current humidity prediction. RNN network is naturally suitable for processing variable-length sequence data, while the time interval of soil moisture observation is not fixed. There may be data loss at certain time nodes. The application of RNN network in soil moisture calculation has advantages, which can better capture time series relationships, process variable-length sequence data, and model long-term dependencies. It also has good scalability and integration.

[0051] In some existing solutions, there are some shortcomings and defects in using RNN models to calculate soil moisture in mining areas. (1) Long training time: RNN models need to consider long-term dependencies during training, and usually require longer training time to optimize model parameters. This may increase the time cost of model development and deployment. (2) Difficulty in processing contextual information: When processing soil moisture data, RNN models often find it difficult to directly capture the complex relationship between soil moisture and other environmental factors (such as temperature, radiation, etc.). This contextual information may have an important impact on the accurate prediction of soil moisture, and the use of RNN models alone may not be able to fully utilize this additional information. (3) Limited model generalization ability: RNN models are prone to overfitting training data, resulting in decreased prediction performance on new unseen data. If the data used is limited or the data distribution changes, the model may lose accuracy. Therefore, it is necessary to improve and optimize the RNN model and combine it with other models to better calculate soil moisture in mining areas.

[0052] In the specific implementation process, before constructing the soil moisture analysis model, the method also includes the following steps: obtaining information of a first network structure, wherein the first network structure is constructed based on a convolution algorithm, and the first network structure is used to extract features of the related data by downsampling and convolution; obtaining information of a second network structure, wherein the second network structure is constructed based on an improved LSTM algorithm, and the improved LSTM algorithm is obtained by improving the LSTM algorithm through feature inheritance and feature selective forgetting mechanisms, and the second network structure is used to adjust the degree of retention of the extracted features; coupling the information of the first network structure and the information of the second network structure to obtain the information of the soil moisture analysis model.

[0053] In this scheme, the features of the data can be extracted through downsampling and convolution, and then the feature inheritance and feature selective forgetting mechanisms are introduced to improve the LSTM algorithm to obtain the Fisfnet algorithm. The convolution algorithm constructs the first network, and the Fisfnet algorithm constructs the second network. The first network and the second network are coupled, so that the model can extract multi-dimensional features and pay attention to time series data, so as to realize the subsequent accurate prediction of soil moisture.

[0054] Specifically, in this solution, the improved LSTM algorithm is given a name, Fisfnet.

[0055] Specifically, Figure 3 As shown in the figure, firstly, according to the long-term meteorological data, geographical environment data, and real soil moisture measurement values ​​of the mining area, a soil moisture feature set of the mining area is constructed, and the data set is divided into a training set, a test set, and a validation set. Secondly, a Con-Fisfnet network model is built. The convolutional neural network is first applied to the constructed data set for convolution, downsampling, and a series of other operations to extract the multi-dimensional characteristics of the mining area soil. The purpose is to directly capture the complex relationship between soil moisture and other environmental factors (such as temperature, radiation, etc.) when processing soil moisture data, and solve the problem that the traditional model ignores the complex relationship between soil moisture and other environmental factors (such as temperature, radiation, etc.). The soil moisture can be calculated more accurately. At the same time, the application of the Fisfnet model can not only consider the soil moisture characteristics of the previous time, but also consider all the feature details that have appeared in the previous time. Through the unique feature inheritance and selective forgetting method, the generalization ability of the model is improved, and the problem of overfitting in the training process is effectively avoided. Finally, the relevant parameters of the model are continuously adjusted through the test set, and the model effect is continuously optimized, so that it can be better used for wide-area soil moisture calculation in mining areas.

[0056] Specifically, the soil moisture analysis model of this scheme can also be constructed in the following way: the soil moisture analysis model is trained by using multiple sets of training data through convolution algorithm and Fisfnet algorithm (i.e., improved LSTM algorithm), and each set of training data in the above multiple sets of training data includes historical relevant data obtained within the historical time period and historical soil moisture corresponding to the above historical relevant data.

[0057] Specifically, in this scheme, a feature inheritance and selective forgetting neural network (Fisfnet) is built, and the convolutional neural network module is used to extract the multi-dimensional features of the mining area soil, which is combined with the Fisfnet network to form a Con-Fisfnet network to achieve accurate prediction of soil moisture; the soil moisture feature extraction module is built using the Pytoch deep learning framework. Through the convolutional neural network, the soil moisture feature set of the mining area is extracted for multi-scale features of soil moisture through downsampling and convolution operations. The feature inheritance and selective forgetting mechanism is introduced, and the previous time series data is introduced in the forgetting process to construct the Fisfnet model; the convolutional neural network feature extraction module is coupled with the Fisfnet model to create a Con-Fisfnet network model with both feature extraction and inheritance and feature selective forgetting.

[0058] Specifically, in this scheme, we first integrate long-term multi-source remote sensing data such as meteorological data, geographic environment data, and soil moisture measurement data to construct a mining area soil moisture feature data set. Secondly, we build a feature inheritance and selective forgetting neural network (Fisfnet), which not only performs well when considering the current feature value pixel by pixel, but also effectively inherits and selectively forgets the time series data to achieve accurate prediction of soil moisture; this soil moisture calculation method based on feature inheritance and selective forgetting avoids the problem of overfitting of traditional soil moisture prediction models, shows stronger performance in processing long time series data, and can more accurately predict soil moisture at future moments. Finally, based on convolutional neural networks, a new wide-area soil moisture calculation network model called Con-Fisfnet is proposed. This model fully considers the relationship between soil moisture and other related factors, and solves the problem of traditional methods in directly capturing the complex relationship between soil moisture and environmental factors (such as temperature, radiation, etc.). The model has strong generalization ability and high robustness, and can better achieve accurate soil moisture prediction.

[0059] Specifically, in this solution, according to the soil moisture feature extraction module, soil moisture multi-scale feature extraction is performed on the mining area soil moisture feature set through downsampling and convolution operations. The convolutional neural network constructed in the present invention includes 4 convolution layers, 3 pooling layers and one fully connected layer. The established mining area soil feature data set is converted into a 32*32 grayscale image. The first convolution layer includes 6 5*5 convolution kernels with a step size of 1. No padding is performed in this operation to obtain a 28*28*6 feature map. Subsequently, a downsampling operation is performed, using a 2*2 maximum pooling with a step size of 2. The obtained soil moisture feature map (14*14) is halved, which can increase the speed of feature extraction; then The convolution operation is performed again. In this convolution process, the convolution kernel size is 5*5, the step size is 1, the number of output channels is 16, and there is no padding operation. A 10*10*16 feature map is obtained. The downsampling operation is repeated on the new feature map to obtain a 5*5*16 feature map. The fully connected layer pulls each feature map of size 5*5 into a vector of length 400 and connects it through a fully connected layer with 120 neurons. Then, convolution and downsampling operations are performed again to obtain the input of Fisfnet.

[0060] Specifically, feature inheritance and selective forgetting mechanisms are introduced, and the data of previous time series are introduced in the forgetting process to construct a Fisfnet model. In a time series, not all information is equal. The present invention establishes a dynamic neural network, and the value of feature information gradually decays with the advancement of time. A gate (forget gate) is used to control the degree of retention of the information at the previous time to the information at this moment, that is, a selective forgetting process. Another gate (input gate) is used to control how much of the features at the current moment should be inherited, and finally an output gate is used to control the current output value of the Fisfnet model.

[0061] In some embodiments, obtaining the information of the second network structure can be specifically achieved by the following steps: calculating the information of the forget gate according to the first formula, wherein the first formula is: Z f =σ(W f ×[h t-1 ,h t-2 ,...,h t-n ,X t ]+b f ), Z f represents the information of the above forget gate, σ represents the sigmoid function, W f represents the weight matrix of the above forget gate, h t-n represents the feature vector, X t The input value at the current moment, b f represents the bias term of the forget gate; the information of the memory gate is calculated according to the second formula, wherein the second formula is: Zi =σ(W i ×[h t-1 ,h t-2 ,...,h t-n ,X t ]+b i ), Z i Represents the information of the above memory gate, W i represents the weight matrix of the above memory gate, b i represents the bias term of the above memory gate.

[0062] In this scheme, the forget gate can control the degree to which the information at the previous time is retained in the current information, that is, a selective forgetting process, and the memory gate can determine how much of the feature input at the current moment will be retained to the unit state at the current moment, and then accurately determine the promotion value between features, so as to obtain a better LSTM algorithm, so as to facilitate the subsequent processing of time series data based on the improved LSTM algorithm.

[0063] Specifically, the overall framework of Fisfnet is as follows Figure 4 shown.

[0064] Z f is the forget gate, and its calculation formula is shown in the first formula, where [h t-1 ,h t-2 ,...,h t-n ,X t ] means connecting the feature vector of the previous moment and the input value of the current moment into a longer vector, and σ is the sigmoid function, which is used to compress any input to the interval (0,1).

[0065] Where Z i It is the memory gate, which determines how much of the feature input at the current moment will be saved to the unit state at the current moment. Its calculation formula is shown in the second formula, where [h t-1 ,h t-2 ,...,h t-n ,X t ] means connecting the feature vector of the previous moment and the input value of the current moment into a longer vector, and σ is the sigmoid function, which is used to compress any input to the interval (0,1).

[0066] In some embodiments, obtaining the information of the second network structure can be specifically achieved by the following steps: calculating the parameters of the current input unit state according to the third formula, wherein the third formula is: Z = tanh (W c ×[h t-1 ,h t-2 ,...,h t-n ,X t]+b c ), Z represents the parameter of the above unit state currently input, W c represents the weight matrix of the above unit state for the current input, b c The bias term representing the current input unit state; the information of the unit state at the current moment is calculated according to the fourth formula, wherein the fourth formula is: c t =Z f ×mean(c t-1 ,c t-2 ,...,c t-n )+Z i ×Z,c t Information indicating the status of the above unit at the current moment.

[0067] In this scheme, the tanh function can be used to control the impact of input information in the network on the cell state, which can increase the nonlinear expression ability of the model and appropriately scale the activation value, which helps to improve the model's ability to capture and model input data. The cell state can retain information from a long time ago, has the feature of selective forgetting, and can prevent currently irrelevant content from entering the memory. It has the function of feature inheritance, and can obtain a better LSTM algorithm to facilitate the subsequent processing of time series data based on the improved LSTM algorithm.

[0068] Specifically, Z is a parameter used to describe the current input unit state, and its calculation formula is shown in the third formula, where [h t-1 ,h t-2 ,...,h t-n ,X t ] means connecting the previously input feature vector and the input value at the current moment into a longer vector. Tanh refers to the hyperbolic tangent function, also known as the hyperbolic tangent activation function. It is a nonlinear function, commonly used as an activation function in neural networks. It determines the importance of the information input at the current time step in the cell state. The tanh function converts the input value to a range between -1 and 1 to control the impact of the input information on the cell state. In the present invention, the introduction of the tanh function can increase the nonlinear expression ability of the model and appropriately scale the activation value. This helps to improve the Fisfnet model's ability to capture and model input data.

[0069] Specifically, c tis the cell state at the current moment, and its calculation formula is shown in the fourth formula. The cell state at the current moment is the sum of the average value of all cell states at previous times multiplied by the forget gate and the product of the current input cell state multiplied by the memory gate. It can retain information from a long time ago and has the feature of selective forgetting, and can prevent currently irrelevant content from entering the memory and has the function of feature inheritance.

[0070] In some embodiments, obtaining the information of the second network structure can be specifically implemented by the following steps: calculating the information of the output gate according to the fifth formula, wherein the fifth formula is: Z o =σ(W o ×[h t-1 ,h t-2 ,...,h t-n ,X t ]+b o ), Z o Represents the information of the above output gate, W o represents the weight matrix of the above output gate, b o represents the bias term of the output gate; the output value of the second network structure is calculated according to the sixth formula, wherein the sixth formula is: h t =Z o ×tanc(c t ), h t Represents the output value of the above second network structure.

[0071] In this solution, the output gate can be used to control how much of the current feature should be inherited, and then the output value can be calculated. Through this embodiment, a better LSTM algorithm can be obtained, so as to facilitate the subsequent processing of time series data based on the improved LSTM algorithm.

[0072] Specifically, Z o is the output gate, which is used to control the current cell state c t How many outputs are there to the current output value h of the LSTM model? t , its calculation formula is shown in the fifth formula, where [h t-1 ,h t-2 ,...,h t-n ,X t ] means connecting the feature vector of the previous moment and the input value of the current moment into a longer vector, and σ is the sigmoid function, which is used to compress any input to the interval (0,1).

[0073] Specifically, the current output value h tThe calculation formula is shown in the sixth formula. The convolutional neural network model is coupled with the Fisfnet model to create a Con-Fisfnet model with wide-area feature extraction, inheritance, and feature selective forgetting. Figure 5 shown.

[0074] In the specific implementation process, before constructing the soil moisture analysis model, the method further includes the following steps: constructing an initial soil moisture analysis model, wherein the initial soil moisture analysis model is trained by a convolution algorithm and an LSTM algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes the historical related data obtained in the historical time period and the historical soil moisture corresponding to the historical related data; using Adam optimization and gradient descent to optimize the initial soil moisture analysis model to obtain the soil moisture analysis model.

[0075] In this scheme, an initial soil moisture analysis model can be constructed first, and then the Adam optimization and gradient descent methods are used to optimize the initial soil moisture analysis model. The Adam algorithm can quickly converge to the local optimal solution in the early stage of training. The Adam algorithm can adaptively adjust the learning rate of each parameter without manually setting the learning rate. Different learning rates can be used for different parameters. The gradient descent algorithm is the most basic optimization algorithm with simple implementation. Each step is updated based on the gradient of the parameters. Therefore, a soil moisture analysis model with higher accuracy can be obtained through this scheme.

[0076] Specifically, this scheme can use the training data set to train the network model; calculate the wide-area soil moisture of the mining area through the trained network model, and continuously adjust the parameters for optimization according to the calculation results; use the training set in the mining area soil moisture feature set to train the Con-Fisfnet model, with weather data, geographical environment data, and soil moisture test data as input, and soil moisture as output, to build a soil moisture prediction model (i.e., the initial soil moisture analysis model); use the cross entropy loss function in the model training process, and use Adam as the optimizer. The calculation method of the cross entropy loss function is:

[0077]

[0078] L represents the model loss value, which indicates the error size of the model. S represents the number of image objects in the validation set. y represents the actual soil moisture of the pixel. Represents the soil moisture calculated by the model.

[0079] Set the number of epochs for iterative training and perform multiple iterative training. In each iterative training, the gradient descent algorithm is used to reduce the model loss value and adjust the relevant parameters of the Con-Fisfnet model (weights in the process of feature inheritance and selective forgetting, etc.).

[0080] Specifically, the adjusted Con-Fisfnet network model is used to calculate the soil moisture in a large area of ​​the mining area, and the relevant data of the mining area for which soil moisture needs to be calculated (meteorological data, geographical environment data, etc.) are input into the improved Con-Fisfnet model as the model; after completing the soil moisture calculation, the calculation results within the mining area are verified for accuracy to ensure the accuracy of the calculation results.

[0081] In summary, in this scheme, firstly, long-term multi-source remote sensing data such as meteorological data, geographic environment data, and soil moisture measurement data are integrated to construct a mining area soil moisture feature data set. Secondly, a feature inheritance and selective forgetting neural network (Fisfnet) is built. For each pixel, the network not only considers the characteristic value of the position, but also inherits and selectively forgets the value at this moment according to the characteristic value of the previous time series of this position, and calculates the soil moisture through the network. This wide-area soil moisture calculation method based on feature basis and selective forgetting can more accurately calculate and predict the soil moisture at future moments. It has stronger processing ability for long-term series data than traditional methods, and can avoid the problem of overfitting of traditional soil moisture prediction models to a certain extent. Finally, Fisfnet is combined with convolutional neural network to generate a new Con-Fisfnet network model. This model considers the relationship between soil moisture and other related factors, and solves the problem that traditional methods are difficult to directly capture the complex relationship between soil moisture and other environmental factors (such as temperature, radiation, etc.) when processing soil moisture data. The model has strong generalization ability and high robustness, and can be better used for soil moisture calculation and prediction. Any technology that uses similar soil moisture calculation methods is within the scope of protection of this patent.

[0082] The embodiment of the present application also provides a device for determining soil moisture in a mining area. It should be noted that the device for determining soil moisture in a mining area in the embodiment of the present application can be used to execute the method for determining soil moisture in a mining area provided in the embodiment of the present application. The device is used to implement the above-mentioned embodiments and preferred implementation modes, and those that have been described will not be repeated. As used below, the term "module" can implement a combination of software and / or hardware for a predetermined function. Although the device described in the following embodiments is preferably implemented in software, the implementation of hardware, or a combination of software and hardware, is also possible and conceived.

[0083] The following is an introduction to the device for determining soil moisture in a mining area provided in an embodiment of the present application.

[0084] Figure 6 is a structural block diagram of a device for determining soil moisture in a mining area according to an embodiment of the present application. Figure 6 As shown, the device comprises:

[0085] The first acquisition unit 10 is used to acquire relevant data of the mining area, wherein the relevant data includes one or more of meteorological data, geographical environment data and remote sensing image data, the meteorological data includes one or more of rainfall, ambient temperature and ambient humidity, the geographical environment data includes one or more of soil type, soil texture and topography, and the remote sensing image data is image data of the surface of the mining area obtained by remote sensing technology;

[0086] A first construction unit 20 is used to construct a soil moisture analysis model, wherein the soil moisture analysis model is obtained by training with a convolution algorithm and an LSTM algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes historical relevant data obtained in a historical time period and historical soil moisture corresponding to the historical relevant data;

[0087] The first processing unit 30 is used to input the above-mentioned relevant data into the above-mentioned soil moisture analysis model to obtain the soil moisture corresponding to the above-mentioned relevant data.

[0088] Through this embodiment, a soil moisture analysis model is built. The soil moisture analysis model is trained based on the convolution algorithm and the LSTM algorithm. The convolution algorithm can well extract useful features in the data. The LSTM algorithm can solve the problems of gradient disappearance and gradient explosion that occur in long time series, and can capture the dependency of data in the time series. Therefore, this solution combines the convolution algorithm with the LSTM algorithm to extract effective data features and make full use of the relationship between the time series in the data, thereby ensuring good accuracy of the measurement results.

[0089] In order to ensure the data is more effective and accurate, the first acquisition unit includes an acquisition module and a processing module. The acquisition module is used to acquire the initial relevant data of the above-mentioned mining area; the processing module is used to perform data preprocessing on the above-mentioned initial relevant data to obtain the above-mentioned relevant data, wherein the data preprocessing method includes one or more of filling missing values, removing noise data, and eliminating abnormal values.

[0090] In this solution, the initial relevant data can be preprocessed to fill in missing values, remove noise data, and eliminate outliers, which can ensure the validity of the data and ensure that the data is relatively accurate.

[0091] During the specific implementation process, the above-mentioned device also includes a second acquisition unit, a third acquisition unit and a coupling unit. The second acquisition unit is used to obtain information of the first network structure before constructing the soil moisture analysis model, wherein the above-mentioned first network structure is constructed based on a convolution algorithm, and the above-mentioned first network structure is used to extract features of the above-mentioned related data by downsampling and convolution; the third acquisition unit is used to obtain information of the second network structure, wherein the above-mentioned second network structure is constructed based on an improved LSTM algorithm, and the improved LSTM algorithm is obtained by improving the LSTM algorithm through feature inheritance and feature selective forgetting mechanism, and the above-mentioned second network structure is used to adjust the retention degree of the extracted features; the coupling unit is used to couple the information of the above-mentioned first network structure with the information of the above-mentioned second network structure to obtain the information of the above-mentioned soil moisture analysis model.

[0092] In this scheme, the features of the data can be extracted through downsampling and convolution, and then the feature inheritance and feature selective forgetting mechanisms are introduced to improve the LSTM algorithm to obtain the Fisfnet algorithm. The convolution algorithm constructs the first network, and the Fisfnet algorithm constructs the second network. The first network and the second network are coupled, so that the model can extract multi-dimensional features and pay attention to time series data, so as to realize the subsequent accurate prediction of soil moisture.

[0093] In some embodiments, the third acquisition unit includes a first calculation module and a second calculation module, and the first calculation module is used to calculate the information of the forget gate according to the first formula, wherein the first formula is: Z f =σ(W f ×[h t-1 ,h t-2 ,...,h t-n ,X t ]+b f ), Z f represents the information of the above forget gate, σ represents the sigmoid function, W f represents the weight matrix of the above forget gate, h t-n represents the feature vector, X t The input value at the current moment, b f represents the bias term of the forget gate; the second calculation module is used to calculate the information of the memory gate according to the second formula, wherein the second formula is: Z i =σ(W i ×[h t-1 ,h t-2 ,...,h t-n ,X t ]+b i ), Z i Represents the information of the above memory gate, W irepresents the weight matrix of the above memory gate, b i represents the bias term of the above memory gate.

[0094] In this scheme, the forget gate can control the degree to which the information at the previous time is retained in the current information, that is, a selective forgetting process, and the memory gate can determine how much of the feature input at the current moment will be retained to the unit state at the current moment, and then accurately determine the promotion value between features, so as to obtain a better LSTM algorithm, so as to facilitate the subsequent processing of time series data based on the improved LSTM algorithm.

[0095] In some embodiments, the third acquisition unit includes a third calculation module and a fourth calculation module, and the third calculation module is used to calculate the parameters of the current input unit state according to a third formula, wherein the third formula is: Z = tanh (W c ×[h t-1 ,h t-2 ,...,h t-n ,X t ]+b c ), Z represents the parameter of the above unit state currently input, W c represents the weight matrix of the above unit state for the current input, b c The fourth calculation module is used to calculate the information of the above-mentioned unit state at the current moment according to the fourth formula, wherein the above-mentioned fourth formula is: c t =Z f ×mean(c t-1 ,c t-2 ,...,c t-n )+Z i ×Z,c t Information indicating the status of the above unit at the current moment.

[0096] In this scheme, the tanh function can be used to control the impact of input information in the network on the cell state, which can increase the nonlinear expression ability of the model and appropriately scale the activation value, which helps to improve the model's ability to capture and model input data. The cell state can retain information from a long time ago, has the feature of selective forgetting, and can prevent currently irrelevant content from entering the memory. It has the function of feature inheritance, and can obtain a better LSTM algorithm to facilitate the subsequent processing of time series data based on the improved LSTM algorithm.

[0097] In some embodiments, the third acquisition unit includes a fifth calculation module and a sixth calculation module, and the fifth calculation module is used to calculate the information of the output gate according to a fifth formula, wherein the fifth formula is: Z o =σ(W o ×[ht-1 ,h t-2 ,...,h t-n ,X t ]+b o ), Z o Represents the information of the above output gate, W o represents the weight matrix of the above output gate, b o represents the bias term of the output gate; the sixth calculation module is used to calculate the output value of the second network structure according to the sixth formula, wherein the sixth formula is: t =Z o ×tanc(c t ), h t Represents the output value of the above second network structure.

[0098] In this solution, the output gate can be used to control how much of the current feature should be inherited, and then the output value can be calculated. Through this embodiment, a better LSTM algorithm can be obtained, so as to facilitate the subsequent processing of time series data based on the improved LSTM algorithm.

[0099] During the specific implementation process, the above-mentioned device also includes a second construction unit and a second processing unit. The second construction unit is used to construct an initial soil moisture analysis model before constructing the soil moisture analysis model, wherein the above-mentioned initial soil moisture analysis model is obtained by training through a convolution algorithm and an LSTM algorithm using multiple sets of training data, and each set of training data in the above-mentioned multiple sets of training data includes the above-mentioned historical related data obtained in the above-mentioned historical time period and the historical soil moisture corresponding to the above-mentioned historical related data; the second processing unit is used to optimize the above-mentioned initial soil moisture analysis model by adopting Adam optimization and gradient descent to obtain the above-mentioned soil moisture analysis model.

[0100] In this scheme, an initial soil moisture analysis model can be constructed first, and then the Adam optimization and gradient descent methods are used to optimize the initial soil moisture analysis model. The Adam algorithm can quickly converge to the local optimal solution in the early stage of training. The Adam algorithm can adaptively adjust the learning rate of each parameter without manually setting the learning rate. Different learning rates can be used for different parameters. The gradient descent algorithm is the most basic optimization algorithm with simple implementation. Each step is updated based on the gradient of the parameters. Therefore, a soil moisture analysis model with higher accuracy can be obtained through this scheme.

[0101] The device for determining soil moisture in the mining area includes a processor and a memory. The first acquisition unit, the first construction unit, the first processing unit, etc. are all stored in the memory as program units, and the processor executes the program units stored in the memory to implement corresponding functions. The modules are all located in the same processor; or, the modules are located in different processors in any combination.

[0102] The processor includes a kernel, and the kernel retrieves the corresponding program unit from the memory. The kernel can be set to one or more, and the problem of poor accuracy of the measurement result of soil moisture in the mining area in the prior art can be solved by adjusting the kernel parameters.

[0103] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0104] An embodiment of the present invention provides a computer-readable storage medium, which includes a stored program, wherein when the program is run, the device where the computer-readable storage medium is located is controlled to execute the method for determining soil moisture in the mining area.

[0105] Specifically, the method for determining soil moisture in a mining area includes:

[0106] Step S201, obtaining relevant data of the mining area, wherein the relevant data includes one or more of meteorological data, geographical environment data and remote sensing image data, the meteorological data includes one or more of rainfall, ambient temperature and ambient humidity, the geographical environment data includes one or more of soil type, soil texture and topography, and the remote sensing image data is image data of the mining area surface obtained by remote sensing technology;

[0107] Step S202, constructing a soil moisture analysis model, wherein the soil moisture analysis model is trained by a convolution algorithm and an LSTM algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes historical related data obtained in a historical time period and historical soil moisture corresponding to the historical related data;

[0108] Step S203: input the above-mentioned relevant data into the above-mentioned soil moisture analysis model to obtain the soil moisture corresponding to the above-mentioned relevant data.

[0109] An embodiment of the present invention provides a processor, and the processor is used to run a program, wherein the method for determining soil moisture in the mining area is executed when the program is run.

[0110] Specifically, the method for determining soil moisture in a mining area includes:

[0111] Step S201, obtaining relevant data of the mining area, wherein the relevant data includes one or more of meteorological data, geographical environment data and remote sensing image data, the meteorological data includes one or more of rainfall, ambient temperature and ambient humidity, the geographical environment data includes one or more of soil type, soil texture and topography, and the remote sensing image data is image data of the mining area surface obtained by remote sensing technology;

[0112] Step S202, constructing a soil moisture analysis model, wherein the soil moisture analysis model is trained by a convolution algorithm and an LSTM algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes historical related data obtained in a historical time period and historical soil moisture corresponding to the historical related data;

[0113] Step S203: input the above-mentioned relevant data into the above-mentioned soil moisture analysis model to obtain the soil moisture corresponding to the above-mentioned relevant data.

[0114] An embodiment of the present invention provides a device, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, at least the following steps of a method for determining soil moisture in a mining area are implemented:

[0115] Specifically, the method for determining soil moisture in a mining area includes:

[0116] Step S201, obtaining relevant data of the mining area, wherein the relevant data includes one or more of meteorological data, geographical environment data and remote sensing image data, the meteorological data includes one or more of rainfall, ambient temperature and ambient humidity, the geographical environment data includes one or more of soil type, soil texture and topography, and the remote sensing image data is image data of the mining area surface obtained by remote sensing technology;

[0117] Step S202, constructing a soil moisture analysis model, wherein the soil moisture analysis model is trained by a convolution algorithm and an LSTM algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes historical related data obtained in a historical time period and historical soil moisture corresponding to the historical related data;

[0118] Step S203: input the above-mentioned relevant data into the above-mentioned soil moisture analysis model to obtain the soil moisture corresponding to the above-mentioned relevant data.

[0119] The devices in this article can be servers, PCs, PADs, mobile phones, etc.

[0120] A computer program product includes a non-volatile computer-readable storage medium, wherein the non-volatile computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for determining soil moisture in the mining area in each embodiment of the present application are implemented:

[0121] Specifically, the method for determining soil moisture in a mining area includes:

[0122] Step S201, obtaining relevant data of the mining area, wherein the relevant data includes one or more of meteorological data, geographical environment data and remote sensing image data, the meteorological data includes one or more of rainfall, ambient temperature and ambient humidity, the geographical environment data includes one or more of soil type, soil texture and topography, and the remote sensing image data is image data of the mining area surface obtained by remote sensing technology;

[0123] Step S202, constructing a soil moisture analysis model, wherein the soil moisture analysis model is trained by a convolution algorithm and an LSTM algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes historical related data obtained in a historical time period and historical soil moisture corresponding to the historical related data;

[0124] Step S203: input the above-mentioned relevant data into the above-mentioned soil moisture analysis model to obtain the soil moisture corresponding to the above-mentioned relevant data.

[0125] The present application also provides a soil moisture detection system, comprising one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for determining the soil moisture of any one of the above-mentioned mining areas.

[0126] Specifically, the method for determining soil moisture in a mining area includes:

[0127] Step S201, obtaining relevant data of the mining area, wherein the relevant data includes one or more of meteorological data, geographical environment data and remote sensing image data, the meteorological data includes one or more of rainfall, ambient temperature and ambient humidity, the geographical environment data includes one or more of soil type, soil texture and topography, and the remote sensing image data is image data of the mining area surface obtained by remote sensing technology;

[0128] Step S202, constructing a soil moisture analysis model, wherein the soil moisture analysis model is trained by a convolution algorithm and an LSTM algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes historical related data obtained in a historical time period and historical soil moisture corresponding to the historical related data;

[0129] Step S203: input the above-mentioned relevant data into the above-mentioned soil moisture analysis model to obtain the soil moisture corresponding to the above-mentioned relevant data.

[0130] Obviously, those skilled in the art should understand that the above modules or steps of the present invention can be implemented by a general computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, and in some cases, the steps shown or described can be executed in a different order than here, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. Thus, the present invention is not limited to any specific combination of hardware and software.

[0131] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0132] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0133] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0134] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0135] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0136] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0137] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0138] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0139] From the above description, it can be seen that the above embodiments of the present application achieve the following technical effects:

[0140] 1) The method for determining soil moisture in a mining area of ​​the present application builds a soil moisture analysis model. The soil moisture analysis model is trained based on a convolution algorithm and an LSTM algorithm. The convolution algorithm can extract useful features from the data well. The LSTM algorithm can solve the problems of gradient vanishing and gradient explosion that occur in long time series, and can capture the dependency of data in the time series. Therefore, this solution combines the convolution algorithm with the LSTM algorithm to extract effective data features and make full use of the relationship between the time series in the data, thereby ensuring good accuracy of the measurement results.

[0141] 2) The device for determining soil moisture in the mining area of ​​the present application has built a soil moisture analysis model. The soil moisture analysis model is trained based on the convolution algorithm and the LSTM algorithm. The convolution algorithm can well extract useful features in the data. The LSTM algorithm can solve the problems of gradient disappearance and gradient explosion that occur in long time series, and can capture the dependency of data in the time series. Therefore, this solution combines the convolution algorithm and the LSTM algorithm to extract effective data features and make full use of the relationship between the time series in the data, thereby ensuring good accuracy of the measurement results.

[0142] The above description is only the preferred embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

Claims

1. A method for determining soil moisture in a mining area, characterized in that: include: Acquire relevant data of the mining area, wherein the relevant data includes one or more of meteorological data, geographical environment data and remote sensing image data, the meteorological data includes one or more of rainfall, ambient temperature and ambient humidity, the geographical environment data includes one or more of soil type, soil texture and topography, and the remote sensing image data is image data of the surface of the mining area obtained by remote sensing technology; Constructing a soil moisture analysis model, wherein the soil moisture analysis model is trained by a convolution algorithm and an LSTM algorithm using multiple sets of training data, each set of training data in the multiple sets of training data includes historical relevant data obtained within a historical time period and historical soil moisture corresponding to the historical relevant data; The relevant data are input into the soil moisture analysis model to obtain the soil moisture corresponding to the relevant data.

2. The method according to claim 1, characterized in that Obtain relevant data on the mining area, including: Obtaining initial relevant data of the mining area; The initial relevant data is subjected to data preprocessing to obtain the relevant data, wherein the data preprocessing method includes one or more of filling missing values, removing noise data, and eliminating abnormal values.

3. The method according to claim 1, characterized in that: Before constructing the soil moisture analysis model, the method further includes: Acquire information of a first network structure, wherein the first network structure is constructed based on a convolution algorithm, and the first network structure is used to extract features of the relevant data by downsampling and convolution; Acquiring information of a second network structure, wherein the second network structure is constructed based on an improved LSTM algorithm, the improved LSTM algorithm is obtained by improving the LSTM algorithm through feature inheritance and feature selective forgetting mechanisms, and the second network structure is used to adjust the degree of retention of the extracted features; The information of the first network structure and the information of the second network structure are coupled to obtain the information of the soil moisture analysis model.

4. The method according to claim 3, characterized in that Obtain information about the second network structure, including: The information of the forget gate is calculated according to the first formula, wherein the first formula is: f =σ(W f ×[h t-1 ,h t-2 ,...,h t-n ,X t ]+b f ), Z f represents the information of the forget gate, σ represents the sigmoid function, W f represents the weight matrix of the forget gate, h t-n represents the feature vector, X t The input value at the current moment, b f represents the bias term of the forget gate; The information of the memory gate is calculated according to the second formula, wherein the second formula is: i =σ(W i ×[h t-1 ,h t-2 ,...,h t-n ,X t ]+b i ), Z i Represents the information of the memory gate, W i represents the weight matrix of the memory gate, b i represents the bias term of the memory gate.

5. The method according to claim 4, characterized in that Obtain information about the second network structure, including: The parameters of the current input unit state are calculated according to the third formula, wherein the third formula is: Z = tanh (W c ×[h t-1 ,h t-2 ,...,h t-n ,X t ]+b c ), Z represents the parameter of the unit state currently input, W c represents the weight matrix of the unit state of the current input, b c A bias term representing the current input state of the unit; The information of the unit state at the current moment is calculated according to the fourth formula, wherein the fourth formula is: ct = Zf × mean (ct - 1 ,ct- 2 ,...,ct-n)+Zi×Z,c t Information indicating the state of the unit at the current moment.

6. The method according to claim 5, characterized in that Obtain information about the second network structure, including: The information of the output gate is calculated according to the fifth formula, wherein the fifth formula is: Z o =σ(W o ×[h t-1 ,h t-2 ,...,h t-n ,X t ]+b o ), Z o Represents the information of the output gate, W o represents the weight matrix of the output gate, b o represents a bias term of the output gate; The output value of the second network structure is calculated according to the sixth formula, wherein the sixth formula is: ht=Zo×tanc(ct), h t Represents the output value of the second network structure.

7. The method according to any one of claims 1 to 6, characterized in that Before constructing the soil moisture analysis model, the method further includes: Constructing an initial soil moisture analysis model, wherein the initial soil moisture analysis model is trained by a convolution algorithm and an LSTM algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes the historical relevant data obtained in the historical time period and the historical soil moisture corresponding to the historical relevant data; The initial soil moisture analysis model is optimized by using Adam optimization and gradient descent to obtain the soil moisture analysis model.

8. A device for determining soil moisture in a mining area, characterized in that: include: A first acquisition unit is used to acquire relevant data of the mining area, wherein the relevant data includes one or more of meteorological data, geographical environment data and remote sensing image data, the meteorological data includes one or more of rainfall, ambient temperature and ambient humidity, the geographical environment data includes one or more of soil type, soil texture and topography, and the remote sensing image data is image data of the surface of the mining area obtained by remote sensing technology; A first construction unit is used to construct a soil moisture analysis model, wherein the soil moisture analysis model is trained by a convolution algorithm and an LSTM algorithm using multiple sets of training data, and each set of training data in the multiple sets of training data includes historical relevant data obtained in a historical time period and historical soil moisture corresponding to the historical relevant data; The first processing unit is used to input the relevant data into the soil moisture analysis model to obtain the soil moisture corresponding to the relevant data.

9. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method for determining soil moisture in a mining area as described in any one of claims 1 to 7 are implemented.

10. A soil moisture detection system, characterized in that: include: One or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and are configured to be executed by the one or more processors, and the one or more programs include a method for determining soil moisture in a mining area according to any one of claims 1 to 7.

Images