Automatic slurry treatment system and method based on intelligent algorithm

Through the application of intelligent algorithms and convolutional neural networks, real-time monitoring and intelligent control of mud treatment systems are realized, solving the problems of unstable processing effects and high energy consumption in existing systems, and improving processing efficiency and safety.

CN118993483AInactive Publication Date: 2024-11-22CHINA RAILWAY NO 3 GRP CO LTD +2

Patent Information

Application Number
CN202411093871.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-09
Publication Date
2024-11-22
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing mud treatment system lacks real-time monitoring and intelligent control capabilities, resulting in unstable treatment effects, high energy consumption and large waste production, hindering the progress of the project and the realization of environmental protection goals.

Method used

An automated mud treatment system based on intelligent algorithms is adopted. The remote monitoring module collects mud status and device parameters in real time, the central control module cleanses and preprocesses data, and uses convolutional neural networks to perform prediction analysis to achieve automated adjustment and optimization control.

Benefits of technology

Real-time, comprehensive monitoring and intelligent control of the mud treatment process are achieved, manual intervention and error are reduced, processing efficiency and accuracy are improved, resource waste and energy consumption are reduced, equipment life is extended, and safety is improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118993483A_ABST
    Figure CN118993483A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of slurry dehydration treatment, and particularly relates to an automatic slurry treatment method and system based on an intelligent algorithm. Comprising the following steps: establishing a slurry treatment model comprising a remote monitoring module, a central control module and a data analysis and prediction module; slurry state parameters and device state parameters in the slurry treatment device are monitored through the remote monitoring module, and data are transmitted to the central control module; the central control module preprocesses the transmitted data; the central control module converts the preprocessed data into a JSON format and transmits the data to the data analysis and prediction module; the mud parameters are predicted and analyzed through the data analysis and prediction module; the data analysis and prediction module transmits a prediction and analysis result to the central control module, and the central control module adjusts operation parameters of all parts of the device and adjusts mud state parameters. The slurry treatment efficiency, precision and safety can be improved, human intervention is reduced, and automation of the device is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to mud processing technology, and in particular to an automated mud processing system and method based on an intelligent algorithm. Background Art

[0002] Mud treatment technology plays a vital role in infrastructure engineering fields such as construction. It is not only a key factor in ensuring the smooth progress of engineering projects, but also has a profound impact on environmental protection and resource conservation. Traditional mud treatment systems often encounter challenges such as large fluctuations in treatment effects, high energy consumption, and large amounts of waste due to the lack of real-time monitoring, intelligent control, and resource optimization capabilities. These problems not only reduce work efficiency, but also increase operating costs and cause environmental pollution.

[0003] Patent application number CN202211234479.6 is a mud processing system and processing method based on intelligent sensing technology. The mud state monitoring module obtains the mud physical property parameters; the central processing unit and the dosing control module cooperate to complete the preparation and addition of flocculation agents according to the measured mud physical property parameters; the flocculation state monitoring module senses the mud flocculation state in real time and feeds back data to the central processing unit, the central processing unit calculates and analyzes the mud flocculation state data and compares it with the user input limit, and then dynamically adjusts the flocculation agent mixing ratio and additive amount to achieve refined addition of mud flocculation agents; the user end can display the mud processing system operation data in real time to ensure that the mud dehydration process is efficient, controllable and stable. The above-mentioned device cannot achieve automatic regulation and control, and also requires human intervention. Therefore, it is urgent for those skilled in the art to solve the above-mentioned technical problems. Summary of the invention

[0004] The technical problem to be solved by the present invention is that the current mud treatment system generally faces the problems of incomplete real-time monitoring, insufficient intelligent control capability and limited resource optimization capability. These problems lead to unstable treatment effect, high energy consumption and large amount of waste, which seriously hinder the progress of the project and the achievement of environmental protection goals.

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0006] An automated mud processing system and method based on intelligent algorithm, comprising the following steps:

[0007] S1: Establish a mud processing model including a remote monitoring module, a central control module and a data analysis and prediction module;

[0008] S2: Monitor the mud state parameters and device state parameters in the mud processing device through the remote monitoring module, and transmit the data to the central control module;

[0009] S3: The central control module cleans the transmitted data, processes missing values ​​and outliers, and then pre-processes the data transmitted by the remote monitoring module through normalization;

[0010] S4: The central control module converts the pre-processed mud state parameters and device state parameters into JSON format and transmits them to the data analysis and prediction module;

[0011] S5: Predict and analyze mud parameters through the convolutional neural network in the data analysis and prediction module;

[0012] S6: The data analysis and prediction module transmits the prediction and analysis results to the central control module, and the central control module adjusts the operating parameters of each component of the device to adjust the mud state parameters.

[0013] Furthermore, in step S2, the remote monitoring module includes a mud density meter and a wireless sensor, the mud density meter transmits measured data to the central control module, and the wireless sensor transmits operating parameters of the mud processing device to the central control module.

[0014] The mud density meter can measure the density of the mud in real time, which is very important for maintaining the balance of drilling mud, because the density is directly related to whether the formation pressure can be effectively controlled to prevent blowouts or well leakage. Wireless sensors can monitor various operating parameters of the mud processing device in real time, such as temperature, pressure, flow, vibration, etc., so that the operator can quickly identify any deviation from the normal range and make timely adjustments. Automated data collection reduces the reliance on manual inspection and reduces labor costs. It also avoids human errors and improves the accuracy and consistency of data. Wireless sensors can be deployed in difficult-to-reach or dangerous locations to collect data that is difficult to obtain manually, thereby providing more Comprehensive information, wireless transmission allows the monitoring center to receive data without being on-site, supports remote operation and monitoring, especially in harsh or remote working environments, real-time monitoring helps to detect potential safety hazards early, such as equipment overheating, leakage or mechanical failure, so that timely measures can be taken to avoid accidents, and continuous data flow helps to implement predictive maintenance. By analyzing data patterns, it can be predicted when equipment needs maintenance, thereby reducing unplanned downtime and extending equipment life. The real-time and accuracy of the data provides high-quality input for subsequent data analysis and prediction modules, making data-based decisions more reliable and able to better optimize the mud treatment process.

[0015] Furthermore, in step S3, when the central module pre-processes the data, the following steps are included:

[0016] S31: Data cleaning to handle missing values ​​and outliers;

[0017] S32: transform the data into the same dimension by normalization;

[0018] In step S31, data records containing missing values ​​are deleted, missing values ​​are filled using regression prediction methods, and missing values ​​are filled using time series interpolation methods;

[0019] Use box plots or scatter plots to identify outliers in the data, set thresholds based on data distribution, and values ​​beyond the thresholds are considered outliers. Use other variables to predict outliers and replace the original outliers.

[0020] In step S32, the value of each feature is scaled to between [0, 1], as follows:

[0021]

[0022] Among them, Xnorm is the normalized value, Xmin and Xmax are the minimum and maximum values ​​of the feature respectively.

[0023] If the data set is large and the proportion of missing values ​​is small, deleting records with missing values ​​can avoid introducing bias, but it may cause information loss. It is suitable for situations where there are many missing values ​​or key features are missing. By building a regression model and predicting missing values ​​with other complete features, this method can retain more information, but the prediction accuracy is affected by the quality of the model. It is particularly effective for time series data, such as linear interpolation, spline interpolation, etc., which can maintain the temporal continuity of the data and are suitable for processing short-term missing values. Use box plots or scatter plots to identify outliers: intuitively display data distribution, easy to find outliers, set reasonable thresholds based on data distribution, replace outliers with predicted values ​​or neighboring values, and avoid the negative impact of outliers on model training. Through the formula Ensure that all features are at the same scale to avoid certain features dominating the model due to dimension or magnitude differences. The normalized data gradient is smoother, which helps the optimization algorithm find the optimal solution faster and reduces the error accumulation in numerical calculations, especially when using the gradient descent method, to avoid overflow or underflow problems caused by large values.

[0024] Further, in step S4, the pre-processed data is converted into JSON format by a computer program in the central control module, which specifically includes the following steps:

[0025] S41: Convert the preprocessed data into JSON format using a JSON conversion tool;

[0026] S42: formatting and parsing the converted JSON data to convert the text data into an easy-to-read format;

[0027] S43: Store the converted JSON data in a database.

[0028] JSON is a lightweight data exchange format that is easy for people to read and write, and easy for machines to parse and generate. Using the JSON format can ensure the consistency and readability of data when it is transmitted between different systems and platforms, and improve the standardization level of data. The JSON data structure is clear, similar to dictionaries and lists, which makes it easy to parse and process in various programming languages, which is conducive to further analysis and application of data. Compared with other data exchange formats such as XML, JSON usually takes up less bandwidth, so it can be faster during data transmission and reduce network latency, which is especially important for real-time monitoring and control applications. Storing JSON data in a database can use the indexing and query functions of the database to quickly locate and extract the required data, supporting efficient data management and analysis. Many modern Web APIs choose JSON as the data exchange format. Converting data to JSON format helps integrate with other systems and services, promote the construction of microservice architecture and the development of API economy. Data in JSON format is easy to be parsed by various data visualization tools, and charts, dashboards and reports can be quickly generated to help users intuitively understand the meaning and trends of data. During data transmission and storage, JSON data can be encrypted to protect sensitive information from unauthorized access.

[0029] Furthermore, in step S41, JSONLint is used to convert the preprocessed data into JSON format online.

[0030] JSONLint can check the syntactic correctness of JSON data to ensure that there are no syntax errors, such as mismatched brackets, comma errors, etc., which is the basis for successful data conversion. JSONLint can format the original, potentially difficult-to-read JSON string into a structure with indentation and line breaks, making it easier for humans to read and understand. If there are errors in the JSON data, JSONLint will point out the specific error location and type, making it convenient for developers to quickly locate and correct the problem. For online operations, JSONLint provides instant feedback, and you can view the conversion results and any potential problems without compiling or running additional code. Since it is an online tool, JSONLint does not require the installation of specific software or dependencies. It can be used as long as you can access the Internet. It is suitable for a variety of operating systems and devices.

[0031] Furthermore, in step S5, a convolutional neural network is used to predict and analyze the mud parameters. The convolutional neural network includes 5 convolutional layers and 2 fusion layers. The monitoring data is preliminarily processed through 3 consecutive convolutional layers, then fused through the first fusion layer, and then processed through 2 convolutional layers and 1 fusion layer to finally obtain the output result.

[0032] The convolutional layer is able to capture local features in the input data, which is very important for analyzing the time series or spatial distribution patterns of mud parameters. In addition, the convolution operation is translation invariant, which means that it can detect patterns regardless of their position in the input. Multiple convolutional layers can extract more and more abstract features layer by layer. The first layer may detect basic patterns such as edges or textures, while deeper layers can capture more complex combined features, which is particularly critical for predicting complex behaviors during mud processing. The filters in the convolutional neural network share weights across the entire input, which greatly reduces the number of parameters that need to be learned and reduces the risk of overfitting. At the same time, this structure only considers the local area in the input, reduces the redundancy between features, and improves the efficiency of the model. The activation function (such as ReLU) is applied after the convolution layer, which provides the network with a powerful nonlinear mapping capability, which is essential for dealing with nonlinear relationships and complex dynamic processes in mud processing. The fusion layer (or dense layer, fully connected layer) can integrate features from different convolutional layers to form a global representation, which is very helpful for understanding the mud state and predicting future trends from multiple perspectives. The deep network structure allows the model to learn very complex representations, which can provide more accurate predictions when dealing with mud processing data with high variability and uncertainty. CNN can automatically learn which features are most important for predicting changes in mud parameters, which reduces the need for manual feature engineering and improves the adaptability and robustness of the model.

[0033] Furthermore, the data analysis and prediction module also includes an optimization module, and the optimization module includes hidden feature training and recognition training;

[0034] For hidden feature training, common intermediate features are selected in the pre-trained model, and the model is trained using the quadruple loss function. The following formula is the quadruple loss function:

[0035]

[0036] in And V i , i=1,2...m is the i-th hidden feature extracted from the fixed pre-trained model, D(a,b) is the X1 distance between a and b, ω i is the weight coefficient, Gt is the positive sample image, T1 is the corresponding low-density mud, T2 is the corresponding high-density mud, and f = (I) is the output density feature;

[0037] For recognition training, the model is trained by introducing the gradient loss function:

[0038] The gradient loss function is introduced to make the density anchor point closer to the positive sample in the feature space and farther away from the simple negative sample. At this distance, it is farther away from the complex negative sample. The similarities and differences between different samples are compared, so that the model can better understand and process monitoring data of different densities.

[0039] Gradient loss function:

[0040]

[0041] Among them, f represents the output of the network, I represents the input data, Gt represents the standard density, represents the gradient operation;

[0042] The loss value is then calculated through the joint function, and the parameters of the model are updated through back propagation; Joint loss function: L = min||Gt-t(I)||+α*X1+β*X2

[0043] Among them, α and β are weight parameters, t represents the output of the network, I represents the input data, and Gt represents the standard density.

[0044] In addition, the present invention also includes an automated mud processing system based on an intelligent algorithm, a data processor, a central controller, a computer memory and the above-mentioned automated mud processing method based on an intelligent algorithm as the system operation logic, including a remote monitoring module, a central control module and a data analysis and prediction module, the remote monitoring module is used to monitor the mud density and the operating parameters of the mud processing device in the mud processing device, the central control module is used to convert the data into JOSN format data of the model scale, and adjust the operating parameters of the mud processing device according to the data analysis and prediction module, the data analysis and prediction module is used to analyze the change trend of the monitoring data and adjust the operating parameters of the mud processing device in advance through the central control module, the data analysis and prediction module also includes an optimization module, and the optimization module is used to enhance the model recognition capability according to hidden features.

[0045] The beneficial effects of the present invention are:

[0046] 1. The present invention realizes real-time and comprehensive monitoring of the mud treatment process through the remote monitoring module, the central control module and the data analysis and prediction module. The mud state and device operation parameters are collected through the remote monitoring module to ensure the real-time and accuracy of the data. The central control module can automatically process the data, including data cleaning, missing value processing, outlier detection and data normalization, which improves the data quality and the reliability of subsequent analysis. The data analysis and prediction module predicts and analyzes the mud parameters through intelligent algorithms such as convolutional neural networks, realizes intelligent control of the treatment process, can predict and adjust the mud state in advance, and avoids the lag and uncertainty of manual intervention;

[0047] 2. The automated data processing and control mechanism reduces the need for manual operation, reduces the error rate, and improves processing efficiency and accuracy. The closed-loop control mechanism automatically adjusts the control strategy through real-time monitoring and prediction, ensuring the optimization and stability of the mud treatment process, reducing resource waste and energy consumption. The intelligent algorithm optimizes the mud treatment process, which can reduce the use of chemicals, save energy, and reduce operating costs. The predictive maintenance function reduces unplanned downtime, extends equipment life, and further saves maintenance costs.

[0048] 3. Real-time monitoring helps to detect potential safety hazards at an early stage, such as equipment overheating, leakage or mechanical failure, so that timely measures can be taken to avoid accidents. The real-time and accuracy of data provides a solid foundation for decision support, helps prevent operational risks and improve operational safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a flow chart of the method of the present invention.

[0050] Figure 2 It is a schematic diagram of the data analysis and prediction module of the present invention.

[0051] Figure 3 It is a schematic diagram of the optimization module of the present invention. DETAILED DESCRIPTION

[0052] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific preferred embodiments.

[0053] In the description of the present invention, it should be understood that the terms "left side", "right side", "upper part", "lower part" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation. "First", "second" and the like do not indicate the importance of the components, and therefore cannot be understood as limiting the present invention. The specific dimensions used in this embodiment are only for illustrating the technical solution, and do not limit the scope of protection of the present invention.

[0054] Reference Figure 1 ,It can be known that an automated mud processing method based on ,intelligent algorithm includes the following steps, S1: establishing a mud processing model including a remote ,monitoring module, a central control module and a data analysis and ,prediction module;

[0055] S2: Monitor the mud state parameters and device state parameters in the mud processing device through the remote monitoring module, and transmit the data to the central control module;

[0056] S3: The central control module cleans the transmitted data, processes missing values ​​and outliers, and then pre-processes the data transmitted by the remote monitoring module through normalization;

[0057] S4: The central control module converts the pre-processed mud state parameters and device state parameters into JSON format and transmits them to the data analysis and prediction module;

[0058] S5: Predict and analyze mud parameters through the convolutional neural network in the data analysis and prediction module;

[0059] S6: The data analysis and prediction module transmits the prediction and analysis results to the central control module, and the central control module adjusts the operating parameters of each component of the device to adjust the mud state parameters;

[0060] The remote monitoring module includes a mud density meter and a wireless sensor, wherein the mud density meter transmits measured data to the central control module, and the wireless sensor transmits operating parameters of the mud processing device to the central control module;

[0061] In one embodiment, first, a mud processing model including a remote monitoring module, a central control module and a data analysis and prediction module is established. The remote monitoring module is responsible for data acquisition, the central control module processes data and performs control logic, and the data analysis and prediction module provides intelligent analysis. A mud density meter and a wireless sensor are installed in the mud processing device. The former measures mud density, and the latter monitors operating parameters such as temperature and pressure. The data collected by these sensors are transmitted to the central control module in real time via wireless. After receiving the data, the central control module first cleans the data, processes missing values ​​and outliers, deletes data records containing missing values, or uses regression prediction methods or time series interpolation methods to fill in missing values, uses box plots or scatter plots to identify outliers, sets thresholds, and values ​​exceeding the thresholds are considered outliers and replaced with predicted values. The data is converted to the same dimension, that is, scaled to the range of [0,1]. The formula is

[0062] The central control module uses JSONLint to convert the preprocessed data into JSON format. The converted JSON data is formatted and stored in the database for subsequent analysis and call. The data analysis and prediction module uses a convolutional neural network to predict and analyze mud parameters. The network contains 5 convolutional layers and 2 fusion layers to process and fuse data features. The prediction and analysis results are transmitted back to the central control module, which adjusts the operating parameters of the mud processing device accordingly to optimize the mud state. The online tool JSONLint is used to convert the preprocessed data into JSON format to ensure the grammatical correctness and readability of the data.

[0063] When the data set is large and has few missing values, delete the records with missing values. If there are many missing values, use regression prediction or time series interpolation to fill them. Outliers are identified and replaced by statistical methods to ensure that all features are on the same scale and to avoid certain features dominating the model due to dimension or magnitude differences. The convolutional neural network conducts in-depth analysis of mud parameters through multi-layer processing and fusion operations to improve prediction accuracy. Based on real-time feedback of data analysis results, equipment parameters are adjusted to achieve intelligent control and optimization of the mud treatment process.

[0064] When the remote monitoring module detects that the mud density is too high, the central control module controls the water injection of the mud treatment device or uses a vibrating screen to remove fine solid particles in the mud, thereby reducing the total solid content of the mud and indirectly reducing the density;

[0065] When the remote monitoring module detects that the mud density is too low, the central control module adjusts the mud density by controlling the water injection amount of the mud processing device or controlling the dosing device of the mud processing device to inject a weighting agent into the mud.

[0066] Reference Figure 2 and Figure 3 It can be seen that the data analysis and prediction module receives the pre-processed mud state parameters and device state parameters transmitted from the central control module. These parameters have been converted into JSON format to facilitate parsing and use by the data analysis and prediction module. The data analysis and prediction module first parses the JSON data and converts it into an internal data structure to facilitate algorithm processing. The data may include parameters such as mud density, temperature, pressure, flow, viscosity, and the operating status information of the device. According to the requirements of the prediction task, useful features are extracted from the raw data. Features may include time series features, statistical features (such as mean, variance), periodic features, etc., to capture the dynamic changes of mud states. The data analysis and prediction module loads a pre-trained convolutional neural network (CNN) model;

[0067] If the model has not been trained, the data analysis and prediction module needs to be trained using historical data to optimize model parameters to predict the changing trend of mud parameters, and use the trained CNN model to predict the current mud state parameters. CNN captures the spatial or temporal correlation of input data through its convolutional layer, and the fusion layer integrates these features to form a comprehensive understanding of the mud state. In the optimization module, the model is fine-tuned using the quadruple loss function to enhance the model's ability to identify hidden features. In this way, the model can learn to distinguish mud samples of different densities and improve the accuracy of predictions;

[0068] For hidden feature training, common intermediate features are selected in the pre-trained model, and the model is trained using the quadruple loss function. The following formula is the quadruple loss function:

[0069]

[0070] in And V i , i=1,2...m is the i-th hidden feature extracted from the fixed pre-trained model, D(a,b) is the X1 distance between a and b, ω i is the weight coefficient, Gt is the positive sample image, T1 is the corresponding low-density mud, T2 is the corresponding high-density mud, and f = (I) is the output density feature;

[0071] For recognition training, the model is trained by introducing the gradient loss function:

[0072] The gradient loss function is introduced to make the density anchor point closer to the positive sample in the feature space and farther away from the simple negative sample. At this distance, it is farther away from the complex negative sample. The similarities and differences between different samples are compared, so that the model can better understand and process monitoring data of different densities.

[0073] Gradient loss function:

[0074]

[0075] Among them, f represents the output of the network, I represents the input data, Gt represents the standard density, represents the gradient operation;

[0076] The loss value is then calculated through the joint function, and the parameters of the model are updated through back propagation; Joint loss function: L = min||Gt-t(I)||+α*X1+β*X2

[0077] Among them, α and β are weight parameters, t represents the output of the network, I represents the input data, and Gt represents the standard density;

[0078] The gradient loss function is introduced to optimize the recognition ability of the model. By comparing the similarities and differences between different samples, the model can better understand and process mud data of different densities. The joint loss function is used to combine the quadruple loss and gradient loss, and the model parameters are adjusted to achieve the best prediction performance. The model outputs predicted mud state parameters, such as future density changes, possible abnormal warnings, etc. These prediction results will be transmitted back to the central control module. The data analysis and prediction module will send the prediction results to the central control module. The central control module adjusts the operating parameters of the mud treatment device according to the prediction analysis results to achieve active regulation of the mud state. The module continuously collects new monitoring data for model retraining and optimization to ensure that the prediction model is continuously improved with changes in the mud treatment process, thereby improving long-term prediction accuracy and stability.

[0079] The preferred embodiments of the present invention are described in detail above; however, the present invention is not limited to the specific details in the above embodiments. Within the technical concept of the present invention, various equivalent transformations can be made to the technical solutions of the present invention, and these equivalent transformations all belong to the protection scope of the present invention.

Claims

1. An automated mud treatment method based on intelligent algorithm, characterized in that: The following steps are involved: S1: Establish a mud processing model including a remote monitoring module, a central control module and a data analysis and prediction module; S2: Monitor the mud state parameters and device state parameters in the mud processing device through the remote monitoring module, and transmit the data to the central control module; S3: The central control module cleans the transmitted data, processes missing values ​​and outliers, and then pre-processes the data transmitted by the remote monitoring module through normalization; S4: The central control module converts the pre-processed mud state parameters and device state parameters into JSON format and transmits them to the data analysis and prediction module; S5: Predict and analyze mud parameters through the convolutional neural network in the data analysis and prediction module; S6: The data analysis and prediction module transmits the prediction and analysis results to the central control module, and the central control module adjusts the operating parameters of each component of the device to adjust the mud state parameters.

2. The automated mud treatment method based on intelligent algorithm according to claim 1, characterized in that: In step S2, the remote monitoring module includes a mud density meter and a wireless sensor. The mud density meter transmits measured data to the central control module, and the wireless sensor transmits operating parameters of the mud processing device to the central control module.

3. The automated mud treatment method based on intelligent algorithm according to claim 1, characterized in that: In step S3, when the central module pre-processes the data, the following steps are included: S31: Data cleaning to handle missing values ​​and outliers; S32: Convert the data to the same dimension by normalization; In step S31, data records containing missing values ​​are deleted, missing values ​​are filled using regression prediction methods, and missing values ​​are filled using time series interpolation methods; Use box plots or scatter plots to identify outliers in the data, set thresholds based on data distribution, and values ​​beyond the thresholds are considered outliers. Use other variables to predict outliers and replace the original outliers. In step S32, the value of each feature is scaled to between [0, 1], as follows: Among them, Xnorm is the normalized value, Xmin and Xmax are the minimum and maximum values ​​of the feature respectively.

4. The automated mud treatment method based on intelligent algorithm according to claim 1, characterized in that: In step S4, the pre-processed data is converted into JSON format by a computer program in the central control module, which specifically includes the following steps: S41: Convert the preprocessed data into JSON format using a JSON conversion tool; S42: formatting and parsing the converted JSON data to convert the text data into an easy-to-read format; S43: Store the converted JSON data in a database.

5. The automated mud treatment method based on intelligent algorithm according to claim 4 is characterized in that: In step S41, JSONLint is used to convert the preprocessed data into JSON format online.

6. The automated mud treatment method based on intelligent algorithm according to claim 1, characterized in that: In step S5, a convolutional neural network is used to predict and analyze the mud parameters. The convolutional neural network includes 5 convolutional layers and 2 fusion layers. The monitoring data is preliminarily processed by 3 consecutive convolutional layers, then fused by the first fusion layer, and then processed by 2 convolutional layers and 1 fusion layer to finally obtain the output result.

7. The adaptive image defogging and enhancement method in a mine dust and fog environment according to claim 6 is characterized in that: The data analysis and prediction module also includes an optimization module, which includes hidden feature training and recognition training; For hidden feature training, common intermediate features are selected in the pre-trained model, and the model is trained using the quadruple loss function. The following formula is the quadruple loss function: in And V i , i=1,2...m is the i-th hidden feature extracted from the fixed pre-trained model, D(a,b) is the X1 distance between a and b, ω i is the weight coefficient, Gt is the positive sample image, T1 is the corresponding low-density mud, T2 is the corresponding high-density mud, and f = (I) is the output density feature; For recognition training, the model is trained by introducing the gradient loss function: The gradient loss function is introduced to make the density anchor point closer to the positive sample in the feature space and farther away from the simple negative sample. At this distance, it is farther away from the complex negative sample. The similarities and differences between different samples are compared, so that the model can better understand and process monitoring data of different densities. Gradient loss function: Among them, f represents the output of the network, I represents the input data, Gt represents the standard density, represents the gradient operation; The loss value is then calculated through the joint function, and the parameters of the model are updated through back propagation; Joint loss function: L = min||Gt-t(I)||+α*X1+β*X2 Among them, α and β are weight parameters, t represents the output of the network, I represents the input data, and Gt represents the standard density.

8. An automated mud processing system based on intelligent algorithms, a data processor, a central controller, a computer memory, and an automated mud processing method based on intelligent algorithms according to any one of claims 1 to 7, characterized in that: It includes a remote monitoring module, a central control module and a data analysis and prediction module. The remote monitoring module is used to monitor the mud density and operating parameters of the mud processing device. The central control module is used to convert the data into JOSN format data of the model scale, and adjust the operating parameters of the mud processing device according to the data analysis and prediction module. The data analysis and prediction module is used to analyze the change trend of the monitoring data and adjust the operating parameters of the mud processing device in advance through the central control module. The data analysis and prediction module also includes an optimization module, which is used to enhance the model recognition ability according to hidden features.

Citation Information

Patent Citations

  • Slurry treatment system based on intelligent sensing technology and treatment method thereof

    CN115925067A

  • Complete equipment for concentration, dehydration and drying

    CN117228922A

  • Vertical sludge dewatering system and dewatering method based on artificial neural network

    CN117602796A

  • Slurry online monitoring system

    CN118243897A

  • Floc state control device, sludge treatment facility, sludge flocculation facility, sludge thickening facility, floc state control method, and program

    JP2024083901A

Cited By

  • Industrial Internet of Things protocol conversion method and system based on intermediate grammar generation graph

    CN119788749A

  • Shield muck treatment site intelligent control system and control method

    CN120547195A