Prediction method and early warning method of material flow velocity, medium and system

Abstract

The invention relates to a material flow velocity prediction method, an early warning method, a medium and a system, and the method comprises the steps: obtaining material position data at different time points and material form data related to the material flow velocity, and constructing a training data set; according to the training data set, a prediction model is constructed and trained, and the prediction model is constructed based on a regression model and a state space model; the regression model is used for extracting a material flowing speed change value according to the material position data at different time points and the material form data related to the material flowing speed; the state space model is used for outputting a predicted material flowing speed according to the material position data at different time points, the material form data related to the material flowing speed and the material flowing speed change value; and acquiring currently acquired material position data at different time points and material form data related to the material flow velocity, inputting the data into the trained prediction model, and outputting the predicted material flow velocity.

Description

Technical Field

[0001] The present invention relates to the field of flow velocity detection, and in particular to a prediction method, an early warning method, a medium and a system for material flow velocity. Background Art

[0002] Monitoring changes in material flow speed is extremely important in the fields of industrial production, warehousing and logistics. The stability of material flow speed directly affects production efficiency, resource utilization and operating cost control. In addition, in certain circumstances, such as a sudden increase or decrease in material flow, it may cause equipment failure, material waste and even safety issues. Therefore, real-time monitoring of changes in material flow speed is particularly critical.

[0003] In modern industry and logistics, monitoring the flow rate of materials is crucial to ensuring production efficiency, optimizing inventory management, and maintaining production safety. In the past, traditional material flow monitoring methods mainly relied on manual visual inspection or the use of simple sensors (such as photoelectric switches, weight sensors, etc.). Although these methods are low-cost, they have obvious limitations: Accuracy limitations: Manual visual inspections are easily affected by factors such as fatigue and distraction, and the monitoring accuracy of simple sensors is usually limited to single-dimensional data (such as height, position, etc.). Lack of information: Traditional methods cannot provide information on the detailed shape and movement state of materials, making the understanding of material flow conditions too rough. Slow response: When abnormal material flow occurs, it is difficult for traditional methods to detect and respond immediately, resulting in production interruptions or increased safety hazards. Poor adaptability: Faced with complex production environments and diverse material types, traditional monitoring methods are difficult to respond flexibly.

[0004] Therefore, how to provide a smarter and more accurate material flow velocity prediction method is a technical problem that needs to be solved urgently in this field. Summary of the invention

[0005] In order to solve at least one of the above technical problems, the present invention provides a method for predicting material flow rate, comprising:

[0006] S1: Obtain material position data at different time points and material shape data related to material flow speed to construct a training data set;

[0007] S2: Based on the training data set, a prediction model is constructed and trained that takes the material position data at different time points and the material form data related to the material flow speed as input and predicts the material flow speed as output;

[0008] The prediction model is constructed based on a regression model and a state space model; the regression model is used to extract the change value of the material flow velocity according to the material position data at different time points and the material form data related to the material flow velocity; the state space model is used to output the predicted material flow velocity according to the material position data at different time points, the material form data related to the material flow velocity, and the change value of the material flow velocity.

[0009] S3: Obtain the material position data at different time points and the material form data related to the material flow velocity collected currently, input them into the trained prediction model, and output the predicted material flow velocity.

[0010] Further, step S1 includes:

[0011] S11: Collect the continuous point cloud data of the material;

[0012] S12: Perform three-dimensional reconstruction according to the point cloud data to simulate the spatial distribution and motion state of the material on the conveying system;

[0013] S13: Extract the material position data and the material form data related to the material flow data from the three-dimensional model.

[0014] Further, the prediction model includes:

[0015] An input module for inputting the material position data at different time points and the material form data related to the material flow velocity;

[0016] A regression model connected to the input module, which is used to extract the change value of the material flow velocity according to the material position data at different time points and the material form data related to the material flow velocity;

[0017] A state space model connected to the outputs of the input module and the regression model, which is used to obtain the material flow velocity according to the material position data at different time points, the material form data related to the material flow velocity, and the change value of the material flow velocity;

[0018] An output module connected to the output of the state space model for outputting the material flow velocity.

[0019] Further, the input module includes:

[0020] The first channel is used to input the position sequence of the material at different time points;

[0021] The second channel is used to input the volume sequence of the material at different time points;

[0022] The third channel is used to input the density sequence of the material at different time points;

[0023] The fourth channel is used to input the shape sequences of the material at different time points.

[0024] Furthermore, the prediction model further includes:

[0025] An identification model: connected to the input module, used to identify the current working condition as a low-flow condition, a medium-flow condition, or a high-flow condition according to the material position data at different time points and the material shape data related to the material flow velocity;

[0026] A regression module, including: a ridge regression module and a Lasso regression module, connected to the identification model;

[0027] The ridge regression module is used to extract the change value of the material flow velocity according to the material position data at different time points and the material shape data related to the material flow velocity in the low-flow condition;

[0028] The Lasso regression module is used to extract the change value of the material flow velocity according to the material position data at different time points and the material shape data related to the material flow velocity in the high-flow condition;

[0029] The ridge regression module and the Lasso regression module are used to jointly extract the change value of the material flow velocity according to the material position data at different time points and the material shape data related to the material flow velocity in the medium-flow condition.

[0030] Furthermore, the prediction model further includes a fusion module; the prediction in the medium-flow condition includes:

[0031] The ridge regression module extracts the first change value of the material flow velocity according to the material position data at different time points and the material shape data related to the material flow velocity;

[0032] The Lasso regression module extracts the second change value of the material flow velocity according to the material position data at different time points and the material shape data related to the material flow velocity;

[0033] The fusion module determines the final change value of the material flow velocity according to the first change value of the material flow velocity and the second change value of the material flow velocity by weights.

[0034] On the other hand, the present invention also provides a warning method for the material flow velocity, which uses any of the above prediction methods to output the predicted material flow velocity; when the predicted material flow velocity exceeds the warning threshold, a warning is issued.

[0035] On the other hand, the present invention also provides a computer-readable storage medium, on which a computer program for executing any of the above methods is stored.

[0036] On the other hand, the present invention further provides a computer system, including the computer-readable storage medium described in any one of the above and one or more processors;

[0037] The processor is configured to run the computer program. Description of the Drawings

[0038] Figure 1 It is a flowchart of an embodiment of the method for predicting the material flow rate of the present invention;

[0039] Figure 2 It is a schematic structural diagram of an embodiment of the prediction model. Detailed Embodiments

[0040] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0041] It should be noted that if there are directional indications in the embodiments of the present invention, such as up, down, left, right, front, back..., then the directional indications are only used to explain the relative positional relationship and movement conditions between components in a specific posture. If the specific posture changes, the directional indications will also change accordingly. In addition, if there are descriptions such as "first, second", "S1, S2", "step one, step two" in the embodiments of the present invention, then such descriptions are only for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the quantity of the indicated technical features or indicating the execution order of the method, etc. Those skilled in the art can understand that all those that do not violate the invention points under the inventive concept of the invention should be included in the protection scope of the present invention.

[0042] As Figure 1 shown, the present invention provides a method for predicting the material flow rate, including:

[0043] S1: Obtain the material position data at different time points and the material form data related to the material flow rate, and construct a training data set;

[0044] S2: According to the training data set, construct and train a prediction model with the material position data at different time points and the material form data related to the material flow rate as the input and the predicted material flow rate as the output;

[0045] The prediction model is constructed based on a regression model and a state space model. The regression model is used to extract the change value of the material flow velocity according to the material position data at different time points and the material form data related to the material flow velocity. The state space model is used to output the predicted material flow velocity according to the material position data at different time points, the material form data related to the material flow velocity, and the change value of the material flow velocity.

[0046] S3: Obtain the material position data at different time points and the material form data related to the material flow velocity collected currently, input them into the trained prediction model, and output the predicted material flow velocity.

[0047] In this embodiment, a method for predicting the material flow velocity of the present invention is given. The key lies in the prediction model of the material flow velocity, which is constructed based on a regression model and a state space model. Beneficial effects: The flow velocity of the material is closely related to the material position data at different time points and the material form data related to the material flow velocity. These data contain both static relationships and time-dynamic characteristics. Therefore, through the regression model, first extract the dynamic feature of the change value of the material flow velocity from the material position data at different time points and the material form data related to the material flow velocity. Then, combine the initial input with the extracted change value of the material flow velocity, and based on the state space model for processing time series, finally output the predicted material flow velocity, capturing the dynamic changes and potential laws in the material flow process, and improving the prediction accuracy of the prediction model.

[0048] Generally speaking, through time series analysis of these point cloud data of the present invention, the flow velocity of the material within a specific time period can be calculated. Using machine learning or deep learning techniques, the extracted features are used as input variables to train a model that can predict the material flow velocity. This model combines a regression model and a state space model based on time series, specifically depending on the characteristics and requirements of the data.

[0049] In a preferred embodiment, in step S1, it is optional but not limited to installing image acquisition devices such as radars, cameras, and cameras at the observation positions of the material, such as the top, etc., to collect continuous point cloud data to record the position data and form data of the material at different time points. The position data can reflect the distribution of the material in space, such as the change in the position of the material conveyor belt, etc. The form data, which is related to the material flow velocity, such as the bulk density, particle size, humidity, etc. of the material, and these form characteristics will affect the flow performance of the material.

[0050] In a preferred embodiment, step S1 includes:

[0051] S11: Collect continuous point cloud data of the material. Specifically, it is optional to collect continuous point cloud data from the radar and camera devices installed on the material. This continuous point cloud data can be collected regularly or irregularly. Specifically, it is also optional to preprocess the obtained point cloud data to remove outliers and noise data to ensure the accuracy and reliability of the data, such as denoising, filtering, downsampling, etc. operations to improve data quality and processing efficiency. Specifically, it is also optional to store the collected point cloud data in a standard format, such as PLY, PCD, etc. These formats can be conveniently read and processed by subsequent processing software.

[0052] S12: Based on the point cloud data, perform 3D reconstruction to simulate the spatial distribution and motion state of the material on the conveying system. Specifically, use the obtained image data and combine computer vision algorithms for 3D reconstruction to accurately simulate the spatial distribution and motion state of the material on the conveying system. Specifically, it is optional to first use the point cloud data for 3D reconstruction to generate a 3D model of the material on the conveying system. Specifically, methods such as voxel grids and surface reconstruction can be used. Then, based on the continuous point cloud data, simulate the motion state of the material on the conveying system. The optical flow method or other motion estimation methods can be used.

[0053] S13: Extract the material position data and the material morphology data related to the material flow data from the 3D model. Specifically, it is optional to extract the features related to the material flow velocity from the reconstructed 3D model, such as the volume, density, shape, etc. of the material, and their changes over time. Specifically, it is optional to use the center points of the voxel grid as the position data and use Open3D to extract the morphology data.

[0054] In a preferred embodiment, in step S2, as Figure 2 shown, the prediction model may optionally but not limited to include:

[0055] An input module for inputting the material position data and the material morphology data related to the material flow velocity at different time points. Specifically, it is optional to integrate these data into a structured data set, and each sample contains the material position data and morphology data corresponding to the time point as the input features of the training data set.

[0056] A regression model, connected to the input module, is used to extract the change value of the material flow velocity according to the material position data at different time points and the material form data related to the material flow velocity. Specifically, methods such as linear regression, polynomial regression, or non - linear regression can be selected. By establishing a mathematical function relationship, the input features are mapped to the change value of the material flow velocity. For example, for a linear regression model, it can be optionally expressed as: y = β0 + β1x1 + β2x2 + … + βnxn + ∈; where y is the change value of the material flow velocity; x1, x2, …, xn are input features, β0, β1, …, βn are model parameters, and ∈ is the error term.

[0057] A state - space model, connected to the output of the input module and the regression model, is used to obtain the material flow velocity according to the material position data at different time points, the material form data related to the material flow velocity, and the change value of the material flow velocity. Specifically, the model can optionally capture the dynamic changes and potential laws in the material flow process by introducing hidden state variables; algorithms such as Kalman filtering can be optionally used to estimate and update the state - space model to improve the prediction accuracy.

[0058] An output module, connected to the output of the state - space model, is used to output the material flow velocity.

[0059] In this embodiment, a preferred embodiment of the prediction model is given. The constructed regression model and state - space model are combined as a feature extraction module to form a comprehensive prediction model. In the model training stage, the training data set is used to train the model. Through optimization algorithms such as the gradient descent method, Adam optimizer, etc., the model parameters are adjusted to enable the model to better fit the data and reduce the prediction error. In the model prediction stage, the collected data is input into the trained prediction model, and the model outputs the predicted material flow velocity according to the input data and the previously learned rules. This prediction result can provide a decision - making basis for production scheduling, material management, etc., helping to optimize the material flow process and improve production efficiency.

[0060] In a preferred embodiment, the input module may optionally include:

[0061] A first channel for inputting the position sequence of the material at different time points;

[0062] A second channel for inputting the volume sequence of the material at different time points;

[0063] A third channel for inputting the density sequence of the material at different time points;

[0064] A fourth channel for inputting the shape sequence of the material at different time points.

[0065] In this embodiment, a preferred embodiment of the input channel is given. The material form data related to the material flow velocity, including the volume, density, and shape of the material, combined with the position, can be the data directly collected using these parameters, or the data reflecting the dynamic changes of the position and flow characteristics by statistically analyzing the changes of these parameters over time. Optionally, the time series data can be converted into a vector with a fixed length and then subsequent processing can be performed.

[0066] In a preferred embodiment, the prediction model may further include: an identification model;

[0067] The identification model is connected to the input module and is used to identify the current working condition as a low-flow working condition, a medium-flow working condition, or a high-flow working condition according to the material position data at different time points and the material form data related to the material flow velocity.

[0068] Specifically, it is optional to label the historical data of the material position data and the material form data related to the material flow velocity, and set a threshold range for subsequent identification of which type of working condition the current working condition is. Specifically, it is optional to determine the current material flow velocity according to the position data and form data at different time points, set a low threshold for the material flow velocity, such as 2.0, and judge whether the current material flow velocity is lower than the set threshold. If it is lower than the set threshold, it is considered that the flow rate in the space such as the pipeline or equipment is small, and its working state is in the low-flow working condition, such as when the equipment starts, stops, or the flow rate is slow. Subsequently, the ridge regression module is selected for prediction output; a high threshold for the material flow velocity is also set, such as 3.0, and it is judged whether the current material flow velocity is higher than the set threshold. If it is higher than the set threshold, it is considered that the flow rate in the space such as the pipeline or equipment is large, and its working state is in the high-flow working condition, such as in the flow peak, high-load operation and other working conditions. Subsequently, the Lasso regression module is selected for prediction output; if the current material flow velocity is between the low threshold and the high threshold, it is identified as a medium-flow working condition with a moderate flow rate in the space. Subsequently, the ridge regression module and the Lasso regression module are jointly used for prediction output.

[0069] The regression module includes a ridge regression module and a Lasso regression module and is connected to the identification model; the ridge regression module is used to extract the change value of the material flow velocity according to the material position data at different time points and the material form data related to the material flow velocity in the low-flow working condition; the Lasso regression module is used to extract the change value of the material flow velocity according to the material position data at different time points and the material form data related to the material flow velocity in the high-flow working condition; the ridge regression module and the Lasso regression module are used to jointly extract the change value of the material flow velocity according to the material position data at different time points and the material form data related to the material flow velocity in the medium-flow working condition.

[0070] More preferably, the prediction module further includes a fusion module. The prediction of the medium flow rate condition includes:

[0071] The ridge regression module extracts the first material flow velocity change value according to the material position data at different time points and the material form data related to the material flow velocity.

[0072] The Lasso regression module extracts the second material flow velocity change value according to the material position data at different time points and the material form data related to the material flow velocity.

[0073] The fusion module determines the final material flow velocity change value according to the first material flow velocity change value and the second material flow velocity change value according to the weight.

[0074] In this embodiment, a preferred embodiment of the regression model is given, including an identification model, two selective regression modules and a fusion module, mainly because: in the practice process, it is found that the laws of different working conditions are different. Comparing the large material working condition with the small material working condition, the changes in speed, form and position are relatively small. Therefore, in this preferred regression model, an identification model is first used to distinguish the working conditions involved in the current input data through threshold judgment, which are small flow rate conditions, medium flow rate conditions and large flow rate conditions; and appropriate regression modules are selected for different working conditions. Specifically, for the small flow rate condition, the ridge regression module is used, which is relatively insensitive to the scaling of features and generally does not require specific preprocessing of features to simplify the steps and shorten the training and prediction time; and in order to capture the tiny change features of the large flow rate condition, Lasso regression is selected; and for the medium flow rate condition, the two are jointly predicted and output. In addition, both ridge regression and Lasso regression are regularization techniques that can solve the problem of multicollinearity. Ridge regression adds an L2 regularization term to the loss function, while Lasso regression adds an L1 regularization term, and the latter is also beneficial to feature selection.

[0075] In a preferred embodiment, the state space model (State-Space Models) decomposes the system into a measurement equation and a state equation, which can not only capture the internal dynamic structure of the time series, but also introduce exogenous variables to explain the changes in the measured values. First, the regression model outputs the material flow velocity change value ΔV, plus the input - the position x and form s of the material at different time points. Based on the state space model of the time series, the material flow velocity is finally output.

[0076] In a preferred embodiment, the output module is a fully connected layer. More preferably, the loss function of the model can optionally adopt the mean square error loss function, Huber Loss, etc.

[0077] The beneficial effects of the present invention:

[0078] 1. Real-time performance: Through high-speed data acquisition and real-time processing, it can quickly identify changes in the material flow rate and respond in a timely manner.

[0079] 2. Adaptability: Whether it is continuous flow or intermittent flow, whether it is solid, liquid or gas, it can be effectively monitored and analyzed to a certain extent through 3D modeling, promoting the development of industrial production towards higher efficiency, lower cost and higher safety.

[0080] 3. Real-time monitoring and early prediction: Input the newly collected real-time data into the trained model to calculate and monitor the changes in the material flow rate in real time. Once the algorithm detects a significant decrease or complete stop in the material flow rate, an abnormal alarm will be triggered. Such a change in speed may mean that the material encounters obstacles during transportation, such as blockage, jamming or equipment failure.

[0081] 4. Feedback optimization: According to the monitoring results, continuously adjust and optimize the model to improve its accuracy and adaptability. At the same time, the material conveying system can also be optimized and improved to improve the overall efficiency and safety.

[0082] The advantage of this method is that it can not only monitor the material flow in real time, but also provide predictions for future material flow through the learning and analysis of historical data, helping to prevent potential problems and optimize the production process.

[0083] Advantages of applying 3D modeling:

[0084] 1. Precise prediction and optimization: By establishing a 3D model of the material, the flow behavior of the material under different conditions can be accurately simulated, thereby predicting the flow rate changes under different operating conditions. This helps to optimize the production process, such as adjusting the speed of the conveyor belt or the design of the silo, to achieve the best flow efficiency.

[0085] 2. Reducing downtime: Real-time monitoring of changes in the material flow rate can help detect potential blockage or leakage problems in advance, and take timely measures to avoid the interruption of the production line due to failures, significantly reducing downtime and maintenance costs.

[0086] 3. Improving product quality: Through the fine control of the material flow, problems such as uneven mixing or uneven heating during the production process can be reduced. This is particularly important for products that require strict control of processing conditions (such as food and pharmaceuticals), helping to improve the quality consistency of the final product.

[0087] 4. Enhancing safety: In some high-risk industries, such as mining and chemical industries, material flow may cause explosions or fires. 3D modeling technology can pre-identify these risk points and guide the design of safer operating procedures to effectively prevent accidents.

[0088] 5. Energy conservation and emission reduction: Optimizing material flow can reduce energy consumption as equipment can operate more efficiently. Additionally, reducing material waste is also an important part of achieving green production.

[0089] 6. Quick response to market changes: With the help of data analysis capabilities, enterprises can adjust production strategies more quickly according to market demands. For example, in the ever-changing consumer electronics market, they can flexibly adjust the material supply speed of the assembly line to meet the rapidly changing order requirements.

[0090] In summary, applying 3D modeling to monitor changes in material flow speed can not only improve the intelligent level of industrial production but also bring significant improvements in economic benefits, safety, environmental protection, etc. It is one of the important directions for the development of modern manufacturing. However, it should be noted that the implementation of these technologies often requires a relatively high initial investment and continuous data maintenance. Therefore, cost-benefit analysis and long-term development plans need to be comprehensively considered in practical applications.

[0091] On the other hand, the present invention also provides a computer storage medium storing executable program code; the executable program code is used to execute any of the above methods.

[0092] On the other hand, the present invention also provides a terminal device including a memory and a processor; the memory stores program code executable by the processor; the program code is used to execute any of the above methods.

[0093] Exemplarily, the program code can be divided into one or more modules / units, and the one or more modules / units are stored in the memory and executed by the processor to complete the present invention. The one or more modules / units can be a series of computer program instruction segments capable of performing specific functions, and these instruction segments are used to describe the execution process of the program code in the terminal device.

[0094] The terminal device can be a computing device such as a desktop computer, a notebook, a palm computer, and a cloud server. The terminal device may include, but is not limited to, a processor and a memory. Those skilled in the art can understand that the terminal device may further include input / output devices, network access devices, a bus, etc.

[0095] The processor may be a Central Processing Unit (CPU), or may also be other general-purpose processors, Digital Signal Processors (DSPs), Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0096] The memory may be an internal storage unit of the terminal device, such as a hard disk or memory. The memory may also be an external storage device of the terminal device, such as a plug-in hard disk equipped on the terminal device, a SmartMedia Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. Further, the memory may also include both an internal storage unit and an external storage device of the terminal device. The memory is used to store the program code and other programs and data required by the terminal device. The memory may also be used to temporarily store data that has been output or will be output.

[0097] The above warning method, computer storage medium, and terminal device are created based on the above prediction method. Their technical effects and beneficial effects are not elaborated here. The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity of description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0098] The above embodiments only represent several implementation manners of the present invention. The description is relatively specific and detailed, but it should not be construed as a limitation on the scope of the invention patent. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention.

Claims

1. A method for predicting material flow rate, characterized in that: include: S1: Obtain material position data at different time points and material shape data related to material flow speed to build a training data set; S2: Based on the training data set, a prediction model is constructed and trained that takes the material position data at different time points and the material form data related to the material flow speed as input and predicts the material flow speed as output; The prediction model is constructed based on a regression model and a state space model; the regression model is used to extract the material flow velocity change value according to the material position data at different time points and the material form data related to the material flow velocity; the state space model is used to output the predicted material flow velocity according to the material position data at different time points, the material form data related to the material flow velocity and the material flow velocity change value; S3: Obtain the material position data collected at different time points and the material shape data related to the material flow speed, input them into the trained prediction model, and output the predicted material flow speed.

2. The prediction method according to claim 1, characterized in that: Step S1 includes: S11: Collect continuous point cloud data of materials; S12: Perform 3D reconstruction based on point cloud data to simulate the spatial distribution and movement state of materials on the conveying system; S13: extracting material position data and material shape data related to material flow data from the three-dimensional model.

3. The prediction method according to claim 1, characterized in that: Predictive models, including: An input module, used to input material position data at different time points and material form data related to material flow speed; A regression model, connected to the input module, is used to extract the change value of the material flow rate according to the material position data at different time points and the material form data related to the material flow rate; The state space model is connected with the input module and the output of the regression model, and is used to obtain the material flow speed according to the material position data at different time points, the material form data related to the material flow speed, and the material flow speed change value; The output module is connected to the output of the state space model and is used to output the material flow rate.

4. The prediction method according to claim 1, characterized in that: Input modules, including: The first channel is used to input the position sequence of materials at different time points; The second channel is used to input the volume sequence of materials at different time points; The third channel is used to input the density sequence of materials at different time points; The fourth channel is used to input the shape sequence of the material at different time points.

5. The prediction method according to claim 1, characterized in that: Predictive models also include: Identification model: connected to the input module, used to identify the current working condition as a small flow condition, a medium flow condition, or a large flow condition based on the material position data at different time points and the material form data related to the material flow speed; Regression module, including: Ridge regression module and Lasso regression module, connected with the recognition model; Ridge regression module, used to extract the change value of material flow velocity under low flow conditions based on the material position data at different time points and the material form data related to the material flow velocity; Lasso regression module, used to extract the change value of material flow velocity according to the material position data at different time points and the material form data related to the material flow velocity under large flow conditions; The ridge regression module and the lasso regression module are used to jointly extract the change value of the material flow velocity under medium flow conditions based on the material position data at different time points and the material form data related to the material flow velocity.

6. The prediction method according to claim 5, characterized in that: The prediction model also includes a fusion module; the prediction of medium flow conditions includes: A ridge regression module extracts a first material flow velocity change value based on material position data at different time points and material form data related to the material flow velocity; Lasso regression module, extracting the change value of the flow speed of the second material according to the material position data at different time points and the material form data related to the material flow speed; The fusion module determines a final material flow speed change value according to the first material flow speed change value and the second material flow speed change value based on the weight.

7. A material flow velocity early warning method, characterized in that: The prediction method shown in any one of claims 1 to 6 is used to output the predicted material flow speed; when the predicted material flow speed exceeds the warning threshold, an early warning is triggered.

8. A computer-readable storage medium, characterized in that: A computer program for executing the method according to any one of claims 1 to 7 is stored thereon.

9. A computer system, characterized in that: comprising the computer-readable storage medium of claim 8 and one or more processors; The processor is configured to run the computer program.

Images