In-situ leaching mine pumping balance prediction system and prediction method

Abstract

The present invention discloses a system and method for predicting the pumping and injection balance of an in-situ leaching mine. The system includes a mine basic information management subsystem and an in-situ leaching mine pumping and injection balance prediction subsystem. The mine basic information management subsystem is used to collect and store mine geological data, hydrological data, and pumping and injection hole data. The in-situ leaching mine pumping and injection balance prediction subsystem includes a data preprocessing module and a pumping and injection balance prediction module. The data preprocessing module is used to perform stream processing on data from the mine basic information management subsystem. The pumping and injection balance prediction module is constructed based on a neural network and is used to predict the pumping and injection balance point. The pumping and injection balance prediction module inputs data from the data preprocessing module and outputs a pumping and injection balance point prediction signal. The present invention monitors pumping and injection hole information, pumping and injection fluid change information, pumping and injection fluid layout in the mining area, and groundwater flow in real time, improving the accuracy and real-time performance of the model prediction.

Description

Technical Field

[0001] The present invention relates to a mine pumping and injection balance prediction method, and in particular to an in-situ leaching mine pumping and injection balance prediction system and prediction method. Background Art

[0002] The balance of pumping and injection in a mine is the key to ensuring the normal production of an in-situ leaching mine, and the balance of pumping and injection is mainly achieved through drilling. The purpose of ensuring the balance of pumping and injection in an in-situ leaching well field is: (1) to control the leaching solution from flowing out of the ore body to avoid large-scale groundwater pollution; (2) to reduce or eliminate the loss of leaching solution; (3) to reduce or eliminate the metal loss caused by the loss of leaching solution; (4) to increase the economic benefits of the enterprise; (5) to create good conditions for the later groundwater treatment. In actual in-situ leaching production, the balance of pumping and injection has two meanings. On the one hand, it is to control the total amount of liquid pumped and injected in the entire mining area to be basically equal to the amount of liquid injected, so as to ensure that the injected leaching solution is not lost and the extracted leaching solution is not diluted; on the other hand, it is to control the local balance of pumping and injection in each pumping and injection unit to prevent uneven distribution of leaching solution and the generation of "solution dead corners". According to the principle of groundwater dynamics, in order to achieve the purpose of pumping and injection balance, the pumping volume is often controlled to be 1-3% greater than the injection volume, forming a large precipitation funnel in the mining area and controlling the migration of leaching solution only within the mining area. If too much liquid is pumped out, the leachate will be diluted and more ground treatment facilities will be required.

[0003] At present, there is no corresponding mine pumping and injection balance prediction method to predict the equilibrium state reached by mine pumping and injection. Due to reasons such as the location of the detection equipment and the lag of detection data, the leaching liquid is sometimes unevenly distributed, resulting in "solution dead corners". Sometimes the amount of liquid pumped is too much, which dilutes the leaching liquid and causes low pumping and injection efficiency. Summary of the Invention

[0004] The present invention provides a system and method for predicting the pumping balance of an in-situ leaching mine in order to solve the technical problems existing in the known technology.

[0005] The technical solution adopted by the present invention to solve the technical problems existing in the known technology is: a pumping and injection balance prediction system for an in-situ leaching mine, which includes a mine basic information management subsystem and an in-situ leaching mine pumping and injection balance prediction subsystem; the mine basic information management subsystem is used to collect and store mine geological data, hydrological data and pumping and injection hole data; the in-situ leaching mine pumping and injection balance prediction subsystem includes a data preprocessing module and a pumping and injection balance prediction module; the data preprocessing module is used to perform stream processing on data from the mine basic information management subsystem; the pumping and injection balance prediction module is constructed based on a neural network and is used to predict the pumping and injection balance point; the pumping and injection balance prediction module inputs data from the data preprocessing module and outputs a pumping and injection balance point prediction signal.

[0006] Furthermore, it also includes a display subsystem, which inputs and displays data from the mine basic information management subsystem and the in-situ leaching mine pumping and injection balance prediction subsystem.

[0007] The present invention also provides a method for predicting the pumping and injection balance of an in-situ leaching mine. The method sets up an in-situ leaching mine pumping and injection balance prediction system, and the in-situ leaching mine pumping and injection balance prediction system sets up a mine basic information management subsystem and an in-situ leaching mine pumping and injection balance prediction subsystem; the mine basic information management subsystem is used to collect and store mine geological data, hydrological data and pumping and injection hole data; the in-situ leaching mine pumping and injection balance prediction subsystem sets up a data preprocessing module and a pumping and injection balance prediction module; the data preprocessing module is used to perform stream processing on data from the mine basic information management subsystem; a pumping and injection balance prediction module is constructed based on a neural network, and the pumping and injection balance prediction module is used to predict the pumping and injection balance point; the pumping and injection balance prediction module inputs data from the data preprocessing module and outputs a pumping and injection balance point prediction signal.

[0008] Furthermore, the in-situ leaching mine pumping and injection balance prediction system is also provided with a display subsystem, so that the display subsystem inputs and displays data from the mine basic information management subsystem and the in-situ leaching mine pumping and injection balance prediction subsystem.

[0009] Furthermore, the in-situ leaching mine pumping and injection balance prediction system is built based on the BeeGo framework, using the Go language as the back-end development language, using the front-end development software to realize interface display and two-dimensional graphics drawing, and building a three-dimensional model based on the WebGL framework.

[0010] Furthermore, the ECharts framework is used to render and display the prediction results.

[0011] Furthermore, HTML5 and javascript technologies are used to realize interface display and two-dimensional graphics drawing.

[0012] Furthermore, the pumping and injection balance prediction module is constructed based on the Adaptive-ONN neural network.

[0013] Furthermore, using the pumping balance prediction module to predict the pumping balance point includes the following method steps:

[0014] Set the following prediction function F(x):

[0015]

[0016]

[0017]

[0018] h (1) =x (4)

[0019] Use Adaptive algorithm to update the classifier weight α (l) :

[0020]

[0021] Online gradient descent algorithm updates the classifier parameters Θ (l) :

[0022]

[0023] Update weight W (l) :

[0024]

[0025] Where:

[0026] L represents the number of classifiers in the model. There are L+1 classifiers in total. The last classifier in the model is a weighted combination of all the previous classifiers. l represents the sequence number of the classifier in the model.

[0027] t represents the model iteration number;

[0028] x represents the input data of the model;

[0029] y t Represents the prediction result of the model's t-th iteration;

[0030] α represents the weight of the classifier in the model, α (l) represents the weight of the lth classifier; represents the weight of the lth classifier in the tth iteration of the model; represents the weight of the lth classifier in the t+1th iteration of the model;

[0031] h represents the prediction result of the feature classifier learned by the hidden layer in the model during training. (l) represents the prediction result of the lth feature classifier in the hidden layer; h (l-1) represents the prediction result of the l-1th feature classifier in the hidden layer; h (1) Represents the prediction result of the first feature classifier of the hidden layer;

[0032] f represents the final prediction result of the feature classifier learned by the hidden layer in the model during the training process, f (l) Represents the final prediction result of the lth feature classifier in the hidden layer;

[0033] Θ represents the parameters of the model hidden layer classifier, Θ (l) represents the parameters of the lth classifier in the hidden layer of the model, Represents the parameters of the lth classifier in the hidden layer of the model in the tth iteration of the model; Represents the parameters of the lth classifier in the hidden layer of the model in the t+1th iteration of the model;

[0034] W represents the updated weight of the model, W (l) Represents the weight corresponding to the lth classifier in the model, W t (l) Represents the weight corresponding to the lth classifier in the tth iteration of the model; Represents the weight corresponding to the lth classifier in the t+1th iteration of the model;

[0035] Softmax represents the activation function;

[0036] σ represents the activation function;

[0037] η is the learning rate of the model;

[0038] β represents the discount factor, which ranges from 0 to 1; Represents the final discount factor of the model;

[0039] represents the cross entropy loss function.

[0040] Furthermore, the data preprocessing module normalizes the data and then inputs the data into the pumping balance prediction module.

[0041] The advantages and positive effects of the present invention are as follows: the present invention's in-situ leaching mine pumping and injection balance prediction system builds the overall system framework based on the BeeGo framework, uses the Go language as the back-end development system language, utilizes the front-end development technology HTML5 and JavaScript technology to realize interface display and two-dimensional graphics drawing, and builds a three-dimensional model based on the WebGL framework. The present invention can improve the accuracy and response time of pumping and injection balance prediction, and has a faster response time than the traditional pumping and injection balance prediction system. In addition, the present invention can be independently formed into a module and can also be integrated into a mine production system. The system provided by the present invention is simple to operate and easy for users to operate. The present invention monitors the information of the extraction hole, the information of the injection hole, the change information of the extraction and injection liquid, the layout of the extraction and injection liquid in the mining area, and the groundwater flow in real time, thereby improving the accuracy and real-time performance of the model prediction. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a structural schematic diagram of an in-situ leaching mine pumping and injection balance prediction system of the present invention.

[0043] Figure 2 This is a schematic diagram of data transmission within an in-situ leaching mine pumping and injection balance prediction system of the present invention.

[0044] Figure 3 The present invention discloses a work flow chart of an in-situ leaching mine pumping and injection balance prediction system.

[0045] Figure 4 This is a workflow diagram of a pumping and injection balance prediction module of the present invention.

[0046] Figure 5 It is a schematic diagram of the structure of a pumping and injection balance prediction module of the present invention.

[0047] Figure 6 It is a logical architecture diagram of an in-situ leaching mine pumping and injection balance prediction system of the present invention. DETAILED DESCRIPTION

[0048] To further understand the content, features and effects of the present invention, the following embodiments are listed and described in detail with reference to the accompanying drawings:

[0049] The Chinese meanings of the English words, phrases and abbreviations in this application are as follows:

[0050] BeeGo: Beego is an HTTP framework for rapidly developing Go applications. It can be used to quickly develop various applications such as APIs, Web applications, and backend services. It is a RESTful framework.

[0051] HTML5: HTML5 is the latest revision of HTML, finalized by the World Wide Web Consortium (W3C) in October 2014. Its goal is to replace the HTML 4.01 and XHTML 1.0 standards, established in 1999, in order to keep web standards current with the rapid development of internet applications.

[0052] JavaScript: JavaScript is standardized by ECMA (European Computer Manufacturers Association) through ECMAScript. It is a multi-paradigm high-level interpreted programming language based on prototypes and first-class functions. It supports object-oriented programming, imperative programming and functional programming, and provides methods to manipulate text, arrays, dates, regular expressions, etc.

[0053] WebGL: WebGL is a JavaScript API for rendering interactive 2D and 3D graphics in any compatible web browser without the use of plugins. WebGL is fully integrated into the browser's web standards, enabling GPU-accelerated use of image processing and effects as part of the web page's canvas. WebGL elements can be added to other HTML elements and blended with other parts of the web page or its background. WebGL programs consist of control code written in JavaScript and shader code written in the OpenGL Shading Language (GLSL), a language similar to C or C++, which runs on the computer's graphics processing unit (GPU).

[0054] MongoDB: MongoDB lies somewhere between relational and non-relational databases, offering the richest functionality and being the most relational. It supports a very loose data structure, using a BSON format similar to JSON, allowing it to store relatively complex data types. Mongo's greatest feature is its powerful query language, whose syntax is somewhat similar to object-oriented query languages. It can perform almost all of the same single-table query functions as relational databases and also supports data indexing.

[0055] Adaptive-ONN: An adaptive online neural network algorithm whose maximum network mechanism prevents the model from becoming overly bloated, and whose sparsification mechanism improves the sparsity of the model. These mechanisms can accelerate the model's response and convergence speed, allowing it to quickly respond to online environmental changes.

[0056] Go: Go (also known as Golang) is a statically strongly typed, compiled, concurrent, and garbage-collected programming language developed by Google. Based on the Inferno operating system, Go was officially launched in November 2009 as an open-source project, supporting Linux, macOS, Windows, and other operating systems.

[0057] ECharts: ECharts is a JavaScript-based data visualization library that provides intuitive, vivid, interactive, and customizable data visualization charts. ECharts was originally open-sourced by the Baidu team and donated to the Apache Foundation in early 2018, becoming an ASF incubation-level project.

[0058] MVC: The MVC pattern (Model–View–Controller) is a software architecture pattern used in software engineering that divides a software system into three basic components: Model, View, and Controller. The MVC pattern was first proposed by Trygve Reenskaug in 1978 and is a software architecture developed by Xerox PARC in the 1980s for the Smalltalk programming language. The MVC pattern aims to implement dynamic programming, simplifying subsequent program modifications and extensions and enabling the reuse of program components. Furthermore, this pattern simplifies program complexity and makes program structure more intuitive. By separating its essential components, a software system also assigns them appropriate functionality.

[0059] See Figures 1 to 6 A pumping and injection balance prediction system for an in-situ leaching mine includes a mine basic information management subsystem and an in-situ leaching mine pumping and injection balance prediction subsystem; the mine basic information management subsystem is used to collect and store mine geological data, hydrological data and pumping and injection hole data; the in-situ leaching mine pumping and injection balance prediction subsystem includes a data preprocessing module and a pumping and injection balance prediction module; the data preprocessing module is used to perform stream processing on data from the mine basic information management subsystem; the pumping and injection balance prediction module is constructed based on a neural network and is used to predict the pumping and injection balance point; the pumping and injection balance prediction module inputs data from the data preprocessing module and outputs a pumping and injection balance point prediction signal.

[0060] The mine basic information management subsystem can include a geological model unit, a groundwater model unit, and a 3D borehole data unit. The geological model unit can be used to collect and store mine geological data; the groundwater model unit can be used to collect and store hydrological data; and the 3D borehole data unit can be used to collect and store pumping and injection well data.

[0061] The mine basic information management subsystem manages mine-related information, such as the mine's geological model, groundwater model, and three-dimensional drilling data. The mine basic information management subsystem is the basis of the in-situ leaching mine pumping and injection balance prediction subsystem, collecting data and controlling the pumping and injection balance point.

[0062] Preferably, a display subsystem may also be included, which can input and display data from the mine basic information management subsystem and the in-situ leaching mine pumping and injection balance prediction subsystem.

[0063] The present invention also provides a method for predicting the pumping and injection balance of an in-situ leaching mine. The method sets up an in-situ leaching mine pumping and injection balance prediction system, and the in-situ leaching mine pumping and injection balance prediction system sets up a mine basic information management subsystem and an in-situ leaching mine pumping and injection balance prediction subsystem; the mine basic information management subsystem is used to collect and store mine geological data, hydrological data and pumping and injection hole data; the in-situ leaching mine pumping and injection balance prediction subsystem sets up a data preprocessing module and a pumping and injection balance prediction module; the data preprocessing module is used to perform stream processing on data from the mine basic information management subsystem; a pumping and injection balance prediction module is constructed based on a neural network, and the pumping and injection balance prediction module is used to predict the pumping and injection balance point; the pumping and injection balance prediction module inputs data from the data preprocessing module and outputs a pumping and injection balance point prediction signal.

[0064] The mine basic information management subsystem can set up geological model units, groundwater model units and three-dimensional drilling data units.

[0065] Preferably, the in-situ leaching mine pumping and injection balance prediction system may also be provided with a display subsystem, which may enable the display subsystem to input and display data from the mine basic information management subsystem and the in-situ leaching mine pumping and injection balance prediction subsystem.

[0066] Preferably, the display subsystem may use the ECharts system to render and display the prediction results.

[0067] Preferably, the in-situ leaching mine pumping balance prediction system can be built based on the BeeGo framework, the Go language can be used as the back-end development system, the front-end development system can be used to implement interface display and two-dimensional graphics drawing, and the three-dimensional model can be built based on the WebGL framework.

[0068] Preferably, HTML5 and javascript technologies can be used to implement interface display and two-dimensional graphics drawing.

[0069] Preferably, the pumping-injection balance prediction module can be constructed based on an Adaptive-ONN neural network. Historical data can be collected to create samples for training and testing the Adaptive-ONN neural network. The trained Adaptive-ONN neural network then inputs data from the data preprocessing module and outputs a pumping-injection balance prediction signal.

[0070] Preferably, using the pumping balance prediction module to predict the pumping balance point may include the following method steps:

[0071] The method of using the pumping balance prediction module to predict the pumping balance point includes the following steps:

[0072] Set the following prediction function F(x):

[0073]

[0074]

[0075]

[0076] h (1) =x (4)

[0077] Use Adaptive algorithm to update the classifier weight α (l) :

[0078]

[0079] Online gradient descent algorithm updates the classifier parameters Θ (l) :

[0080]

[0081] Update weight W (l) :

[0082]

[0083] Where:

[0084] L represents the number of classifiers in the model. There are L+1 classifiers in total. The last classifier in the model is a weighted combination of all the previous classifiers. l represents the sequence number of the classifier in the model.

[0085] t represents the model iteration number;

[0086] x represents the input data of the model;

[0087] y t Represents the prediction result of the model's t-th iteration;

[0088] α represents the weight of the classifier in the model, α (l) represents the weight of the lth classifier; represents the weight of the lth classifier in the tth iteration of the model; represents the weight of the lth classifier in the t+1th iteration of the model;

[0089] h represents the prediction result of the feature classifier learned by the hidden layer in the model during training. (l) represents the prediction result of the lth feature classifier in the hidden layer; h (l-1) represents the prediction result of the l-1th feature classifier in the hidden layer; h (1) Represents the prediction result of the first feature classifier of the hidden layer;

[0090] f represents the final prediction result of the feature classifier learned by the hidden layer in the model during the training process, f (l) Represents the final prediction result of the lth feature classifier in the hidden layer;

[0091] Θ represents the parameters of the model hidden layer classifier, Θ (l) represents the parameters of the lth classifier in the hidden layer of the model, Represents the parameters of the lth classifier in the hidden layer of the model in the tth iteration of the model; Represents the parameters of the lth classifier in the hidden layer of the model in the t+1th iteration of the model;

[0092] W represents the updated weight of the model, W (l) Represents the weight corresponding to the lth classifier in the model, W t (l) Represents the weight corresponding to the lth classifier in the tth iteration of the model; Represents the weight corresponding to the lth classifier in the t+1th iteration of the model;

[0093] Softmax represents the activation function;

[0094] σ represents an activation function, such as sigmoid, tanh, ReLU, etc., which represents a feedforward step;

[0095] η is the learning rate of the model. The learning rate is a very important parameter. A reasonable learning rate can make the model converge to the minimum point rather than the local optimal point or saddle point.

[0096] β represents the discount factor, which is between 0 and 1 and is used to determine the weight α caused by the loss (l) reduction; Represents the final discount factor of the model;

[0097] Represents the cross entropy loss function, the goal is to make the model parameters W t Apply online gradient descent on this loss function.

[0098] Preferably, the data preprocessing module can be built based on the Kafka stream processing system.

[0099] Preferably, the data preprocessing module can normalize the data and then input the data into the pumping balance prediction module.

[0100] The following is a preferred embodiment of the present invention to further illustrate the workflow and working principle of the present invention:

[0101] A pumping and injection balance prediction system for in-situ leaching mines includes a mine basic information management subsystem, an in-situ leaching mine pumping and injection balance prediction subsystem, and a display subsystem. The mine basic information management subsystem is used to collect and store mine geological data, hydrological data, and pumping and injection hole data. The in-situ leaching mine pumping and injection balance prediction subsystem includes a data preprocessing module and a pumping and injection balance prediction module. The data preprocessing module is used to stream data from the mine basic information management subsystem. The pumping and injection balance prediction module is built based on a neural network and is used to predict the pumping and injection balance point. The pumping and injection balance prediction module inputs data from the data preprocessing module and outputs a pumping and injection balance point prediction signal. The display subsystem inputs and displays data from the mine basic information management subsystem and the in-situ leaching mine pumping and injection balance prediction subsystem. The data preprocessing module is built based on the Kafka stream processing system. The mine basic information management subsystem uses a MongoDB database to store data. The pumping and injection balance prediction module is built based on the Adaptive-ONN neural network. The display subsystem uses ECharts to render and display prediction results.

[0102] The in-situ leaching mine pumping balance prediction system builds the overall system framework based on the BeeGo framework, uses the Go system as the back-end development system, uses the front-end development technologies such as HTML5 system and JavaScript system to realize interface display and two-dimensional graphics drawing, and builds a three-dimensional model based on the WebGL system.

[0103] The workflow of the in-situ leaching mine pumping and injection balance prediction system is as follows:

[0104] 1. Initialization:

[0105] Initialization mainly includes starting the Kafka master server, loading the corresponding configuration file, starting the data producer client and consumer client; establishing a connection between the producer client and the Kafka server; and establishing a connection between the consumer client and the Kafka server.

[0106] 2. Data collection, storage and consumption:

[0107] The collected data is aggregated to the Kafka master server. The server stores the data and establishes a corresponding data producer queue for consumption by the data consumer queue. Figure 2 As shown:

[0108] 3. Data preprocessing:

[0109] Before model training and prediction, the collected data needs to be normalized. The idea of ​​normalization is to find the extreme values ​​(including minimum and maximum values) in the data and then convert the original data into data between the interval [0, 1]. The calculation formula is as follows:

[0110] Where X represents the original input data, X′ represents the normalized data, X′∈[0,1], X MIN Represents the minimum sample value in the original data, X MAX Represents the maximum value of the sample in the original data.

[0111] The data X′ input to the model is in the format:

[0112] X′=[x1,x2,...,x n ,x n+1 ,x n+2 ,...,x n+m ];

[0113] Where n is the number of injection holes, m is the number of extraction holes, and m+n is the dimension of X′.

[0114] 4. Train the withdrawal and injection balance prediction module, and use the trained withdrawal and injection balance prediction module to predict the withdrawal and injection balance point. The method of using the withdrawal and injection balance prediction module to predict the withdrawal and injection balance point includes the following steps:

[0115] Set the following prediction function F(x):

[0116]

[0117]

[0118]

[0119] h (1) =x (4)

[0120] Use Adaptive algorithm to update the classifier weight α (l) :

[0121]

[0122] Online gradient descent algorithm updates the classifier parameters Θ (l) :

[0123]

[0124] Update weight W (l) :

[0125]

[0126] 5. Output and visualization:

[0127] The data format of the model's output prediction result is:

[0128] Y′=[y1,y2,...,y n ,y n+1 ,y n+2 ,...,y n+m ]

[0129] Where n is the number of injection holes, m is the number of extraction holes, and m+n is the dimension of X′.

[0130] 7. Front-end visualization:

[0131] Finally, ECharts is used to render and display the prediction results output by the model on the client.

[0132] You can choose a suitable development system according to your needs. This invention uses Go as the main language for the back-end development system, uses HTML5 for front-end development, uses JavaScript to implement page interaction, builds 3D scenes based on WebGL framework technology, uses Three.js technology based on WebGL framework to build 3D models, uses HTML5 and JavaScript to implement 2D graphics drawing, and uses MongoDB database to store data. The system architecture is designed according to the MVC framework model. The logical architecture diagram of this system is shown below. Figure 6 shown.

[0133] The in-situ leaching mine pumping and injection balance prediction module is the specific implementation of the in-situ leaching mine pumping and injection balance model, which mainly includes data preprocessing, model initialization, model training and prediction. Specifically, it is the in-situ leaching mine pumping and injection balance model prediction technology based on the Adaptive-ONN algorithm.

[0134] The above-mentioned Kafka stream processing technology, MongoDB database, Adaptive-ONN neural network, ECharts framework, BeeGo framework, Go language, HTML5 technology, JavaScript technology, WebGL framework, etc. are all existing technologies.

[0135] The above-mentioned mine basic information management subsystem, display subsystem, data preprocessing module, pumping balance prediction module, geological model unit, groundwater model unit, and three-dimensional drilling data unit can all be constructed using devices, systems and software modules in the existing technology and conventional technical means.

[0136] The embodiments described above are only used to illustrate the technical ideas and features of the present invention. Their purpose is to enable those skilled in the art to understand the contents of the present invention and implement them accordingly. The scope of the patent of the present invention cannot be limited by these embodiments alone. That is, any equivalent changes or modifications made to the spirit disclosed by the present invention still fall within the scope of the patent of the present invention.

Claims

1. A pumping balance prediction system for in-situ leaching mines, characterized by: The system includes a mine basic information management subsystem and an in-situ leaching mine pumping and injection balance prediction subsystem; the mine basic information management subsystem is used to collect and store mine geological data, hydrological data, and pumping and injection hole data; the in-situ leaching mine pumping and injection balance prediction subsystem includes a data preprocessing module and a pumping and injection balance prediction module; The data preprocessing module is used to process the data from the mine basic information management subsystem; the pumping and injection balance prediction module is built based on a neural network and is used to predict the pumping and injection balance point; the pumping and injection balance prediction module inputs the data from the data preprocessing module and outputs the pumping and injection balance point prediction signal; The pumping and injection balance prediction module is built based on the Adaptive-ONN neural network; Set the following prediction function F(x): h (1) =x (4) Use Adaptive algorithm to update the classifier weight α (l) : Online gradient descent algorithm updates the classifier parameters Θ (l) : Update weight W (l) : Where: L represents the number of classifiers in the model. There are L+1 classifiers in total. The last classifier in the model is a weighted combination of all the previous classifiers. l represents the sequence number of the classifier in the model. t represents the model iteration number; x represents the input data of the model; y t Represents the prediction result of the model's t-th iteration; α represents the weight of the classifier in the model, α (l) represents the weight of the lth classifier; represents the weight of the lth classifier in the tth iteration of the model; represents the weight of the lth classifier in the t+1th iteration of the model; h represents the prediction result of the feature classifier learned by the hidden layer in the model during training. (l) represents the prediction result of the lth feature classifier in the hidden layer; h (l-1) represents the prediction result of the l-1th feature classifier in the hidden layer; h (1) Represents the prediction result of the first feature classifier of the hidden layer; f represents the final prediction result of the feature classifier learned by the hidden layer in the model during the training process, f (l) Represents the final prediction result of the lth feature classifier in the hidden layer; Θ represents the parameters of the model hidden layer classifier, Θ (l) represents the parameters of the lth classifier in the hidden layer of the model, Represents the parameters of the lth classifier in the hidden layer of the model in the tth iteration of the model; Represents the parameters of the lth classifier in the hidden layer of the model in the t+1th iteration of the model; W represents the updated weight of the model, W (l) represents the weight corresponding to the lth classifier in the model, Represents the weight corresponding to the lth classifier in the tth iteration of the model; Represents the weight corresponding to the lth classifier in the t+1th iteration of the model; Softmax represents the activation function; σ represents the activation function; η is the learning rate of the model; β represents the discount factor, which is between 0 and 1; β L Represents the final discount factor of the model; L represents the cross entropy loss function.

2. The in-situ leaching mine pumping balance prediction system according to claim 1, characterized in that: It also includes a display subsystem, which inputs and displays data from the mine basic information management subsystem and the in-situ leaching mine pumping and injection balance prediction subsystem.

3. A method for predicting the balance of pumping and injection in an in-situ leaching mine, characterized in that: The method sets up an in-situ leaching mine pumping and injection balance prediction system, which sets up a mine basic information management subsystem and an in-situ leaching mine pumping and injection balance prediction subsystem; The mine basic information management subsystem is used to collect and store mine geological data, hydrological data, and pumping and injection hole data; the in-situ leaching mine pumping and injection balance prediction subsystem is equipped with a data preprocessing module and a pumping and injection balance prediction module; The data preprocessing module is used to perform stream processing on data from the mine basic information management subsystem; Constructing a pumping balance prediction module based on a neural network, using the pumping balance prediction module to predict the pumping balance point; inputting the data from the data preprocessing module into the pumping balance prediction module, and outputting a pumping balance point prediction signal; The pumping and injection balance prediction module is built based on the Adaptive-ONN neural network; Set the following prediction function F(x): h (1) =x (4) Use Adaptive algorithm to update the classifier weight α (l) : Online gradient descent algorithm updates the classifier parameters Θ (l) : Update weight W (l) : Where: L represents the number of classifiers in the model. There are L+1 classifiers in total. The last classifier in the model is a weighted combination of all the previous classifiers. l represents the sequence number of the classifier in the model. t represents the model iteration number; x represents the input data of the model; y t Represents the prediction result of the model's t-th iteration; α represents the weight of the classifier in the model, α (l) represents the weight of the lth classifier; represents the weight of the lth classifier in the tth iteration of the model; represents the weight of the lth classifier in the t+1th iteration of the model; h represents the prediction result of the feature classifier learned by the hidden layer in the model during training. (l) represents the prediction result of the lth feature classifier in the hidden layer; h (l-1) represents the prediction result of the l-1th feature classifier in the hidden layer; h (1) Represents the prediction result of the first feature classifier of the hidden layer; f represents the final prediction result of the feature classifier learned by the hidden layer in the model during the training process, f (l) Represents the final prediction result of the lth feature classifier in the hidden layer; Θ represents the parameters of the model hidden layer classifier, Θ (l) represents the parameters of the lth classifier in the hidden layer of the model, Represents the parameters of the lth classifier in the hidden layer of the model in the tth iteration of the model; Represents the parameters of the lth classifier in the hidden layer of the model in the t+1th iteration of the model; W represents the updated weight of the model, W (l) represents the weight corresponding to the lth classifier in the model, Represents the weight corresponding to the lth classifier in the tth iteration of the model; Represents the weight corresponding to the lth classifier in the t+1th iteration of the model; Softmax represents the activation function; σ represents the activation function; η is the learning rate of the model; β represents the discount factor, which is between 0 and 1; β L Represents the final discount factor of the model; L represents the cross entropy loss function.

4. The method for predicting the pumping balance of an in-situ leaching mine according to claim 3, characterized in that: The in-situ leaching mine pumping and injection balance prediction system is also provided with a display subsystem, so that the display subsystem inputs and displays data of the mine basic information management subsystem and the in-situ leaching mine pumping and injection balance prediction subsystem.

5. The method for predicting the pumping balance of an in-situ leaching mine according to claim 4, characterized in that: The in-situ leaching mine pumping balance prediction system is built based on the BeeGo framework, using the Go language as the back-end development language, using the front-end development software to implement interface display and two-dimensional graphics drawing, and building a three-dimensional model based on the WebGL framework.

6. The method for predicting the pumping balance of an in-situ leaching mine according to claim 5, characterized in that: The ECharts framework is used to render and display the prediction results.

7. The method for predicting the pumping balance of an in-situ leaching mine according to claim 5, characterized in that: Use HTML5 and JavaScript technology to realize interface display and two-dimensional graphics drawing.

8. The method for predicting the pumping balance of an in-situ leaching mine according to claim 3, wherein: The data preprocessing module normalizes the data and then inputs it into the pumping balance prediction module.

Images