Deep cone concentration machine modeling method and system based on state space

Abstract

The invention provides a deep-cone thickener modeling method and system based on a state space, and relates to the technical field of artificial intelligence, and the method comprises the steps: obtaining historical operation data of a deep-cone thickener; preprocessing the historical operation data; under a state space model framework, a deep-cone thickener prediction model based on a neural network is constructed, and the deep-cone thickener prediction model comprises a fragmentation operation module, a Mama module and a prediction unit module; inputting the preprocessed historical operation data as a training set into a deep-cone thickener prediction model so as to train the deep-cone thickener prediction model until the loss function value of the deep-cone thickener prediction model is smaller than a preset loss function value; the trained deep-cone thickener prediction model is output, and modeling of the deep-cone thickener is completed. According to the method, the generalization ability of the model under different working conditions can be ensured, the underflow concentration change of the deep-cone thickener can be accurately predicted, and the efficiency and stability of the paste filling process are improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a deep cone concentration machine modeling method and system based on state space. Background Art

[0002] A deep cone thickener is a device used for solid-liquid separation and is widely used in mining, metallurgy, chemical industry and other industries. Its main function is to concentrate a suspension containing solid particles into a high-concentration underflow, while reducing the liquid in the underflow as much as possible. At the same time, a deep cone thickener is also a device for preparing pastes. A deep cone thickener modeling method based on state space is a technical means for mathematical modeling and prediction of deep cone thickeners. It combines the theory of state space models to describe and control the dynamic behavior of deep cone thickeners.

[0003] With the development of the times, the scale of underground mining of minerals by human beings has also expanded, resulting in an increasing number of underground cavities. At the same time, a large amount of tailings left after mining remain on the surface, causing serious environmental pollution. In order to solve these problems, paste backfilling technology is adopted, and the deep cone thickener is the equipment for preparing paste. Since the deep cone thickener plays a vital role in the paste filling process in the mining field, its performance is directly related to the efficiency and stability of the entire filling system. Therefore, modeling of the deep cone thickener is of great significance to improving the stability, efficiency and reliability of the thickener, industrial production and environmental protection.

[0004] In the early stage, people tried to use manual methods to predict the underflow concentration, but manual predictions were highly subjective and uncertain, resulting in large differences and uncertainties in the prediction results. With the development of simulation technology, researchers have used simulation technology to control and optimize process flows. However, due to the partial unobservability of the thickener system and the extremely complex internal flow, which involves multiphase flow and numerous physical and chemical reactions, it is actually very challenging to rely purely on mathematical models to simulate and predict parameter changes under actual operating conditions, although some scholars have adopted a completely data-driven neural network method to predict the key operating parameters of the thickener. However, most of the methods based on neural network learning have difficulty in capturing long-term dependencies in industrial systems. Some transformer-like models can capture long-term dependencies, but their self-attention mechanisms often cause great time complexity when facing long time series of industrial systems, resulting in low prediction accuracy, long prediction time, and low prediction efficiency, resulting in a waste of resources. Summary of the invention

[0005] In order to solve the technical problem that most of the existing neural network learning-based methods are difficult to capture long-term dependencies in industrial systems, some transformer-like models can capture long-term dependencies, but their self-attention mechanisms often cause great time complexity when facing long time series of industrial systems, resulting in low prediction accuracy, long prediction time and low prediction efficiency, resulting in waste of resources, the present invention provides a deep cone concentration machine modeling method and system based on state space.

[0006] The technical solution provided by the embodiment of the present invention is as follows:

[0007] First aspect

[0008] An embodiment of the present invention provides a deep cone concentration machine modeling method based on state space, comprising:

[0009] S1: Obtain historical operation data of deep cone thickener;

[0010] S2: preprocessing historical operation data;

[0011] S3: Under the state space model framework, a deep cone thickener prediction model based on neural network is constructed, wherein the deep cone thickener prediction model includes a slicing operation module, a Mamba module and a prediction unit module;

[0012] S4: inputting the preprocessed historical operation data as a training set into the deep cone thickener prediction model to train the deep cone thickener prediction model until the loss function value of the deep cone thickener prediction model is less than a preset loss function value;

[0013] S5: Output the trained deep cone thickener prediction model to complete the modeling of the deep cone thickener.

[0014] Second aspect

[0015] An embodiment of the present invention provides a deep cone concentration machine modeling system based on state space, comprising:

[0016] processor;

[0017] A memory having computer-readable instructions stored therein, wherein when the computer-readable instructions are executed by the processor, the state-space-based deep-cone concentration machine modeling method according to the first aspect is implemented.

[0018] The third aspect

[0019] An embodiment of the present invention provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the state-space-based deep-cone concentration machine modeling method according to the first aspect is implemented.

[0020] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0021] In the present invention, by obtaining historical operation data and preprocessing the historical operation data, a deep cone thickener prediction model based on a neural network is constructed under the framework of a state space model. By inputting the preprocessed historical operation data into the model and using a loss function for training, the model can gradually adjust its parameters, reduce the error between the predicted value and the true value, and finally obtain a trained model, thereby completing the modeling process of the deep cone thickener and realizing the efficiency and adaptability of the deep cone thickener modeling method based on the state space. The method uses slicing technology to process long sequence data, which helps to improve the model's ability to process long sequences. At the same time, the Mamba module uses the linear characteristics of the state space model to capture long-term dependencies in the sequence data through the selective state space, while the feedforward network layer can process more direct relationships between variables. By combining the Mamba module and the feedforward network layer, the algorithm can effectively handle variable correlation and time correlation problems, improve the accuracy of prediction, ensure the generalization ability of the model under different working conditions, optimize the operation speed, and improve the efficiency and stability of the paste filling process. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0023] Figure 1 A schematic flow chart of a deep cone concentration machine modeling method based on state space provided in an embodiment of the present invention;

[0024] Figure 2 A schematic diagram of the structure of a deep cone concentrator provided by an embodiment of the present invention;

[0025] Figure 3 A schematic diagram of the principle of a state space equation provided by an embodiment of the present invention;

[0026] Figure 4 A schematic diagram of the structure of a deep cone thickener prediction model provided by an embodiment of the present invention;

[0027] Figure 5 A schematic structural diagram of a deep cone concentration machine modeling system based on state space provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0028] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0029] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "example" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "example" is intended to present the concept in a specific way. In addition, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0030] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0031] Reference Manual Attached Figure 1 , showing a flow chart of a deep cone concentration machine modeling method based on state space provided in an embodiment of the present invention.

[0032] The embodiment of the present invention provides a state-space-based deep cone concentration machine modeling method, which can be implemented by a state-space-based deep cone concentration machine modeling device, which can be a terminal or a server. The processing flow of the state-space-based deep cone concentration machine modeling method can include the following steps:

[0033] S1: Get the historical operation data of the deep cone thickener.

[0034] Among them, the historical operation data of the deep cone thickener refers to the collection of various important operating parameters of the deep cone thickener under different working conditions and time periods. By obtaining the historical operation data of the thickener, detailed operating conditions of the equipment under different working conditions can be provided, laying a data foundation for model construction and performance optimization.

[0035] In a possible implementation, the historical operation data includes underflow flow, feed flow, underflow concentration, feed concentration, mud layer pressure, overflow flow, and rake torque.

[0036] It should be noted that by collecting multi-dimensional parameter data, we can have a more comprehensive understanding of the dynamic response of the equipment and enhance the accuracy and adaptability of the prediction model.

[0037] Specifically, let the length of the historical sequence segment be T h , the length of the prediction segment sequence is T p , then for a given time t0, the offline prediction problem of the thickener system state can be expressed as:

[0038]

[0039] in, represents the underflow concentration of the predicted system state, represents the control input set {u2,u3,u4,u5,u6,u7,u8,u9,u 10 All parameters within the time range (t0, t0-T h ) sequences, X(t0,-T h ) represents the historical time range (t0, t0-T h ), represents the control input set {u2,u3,u4,u5,u6,u7,u8,u9,u 10 All parameters within the time range (t0, t0+T p ) sequences.

[0040] Table 1 Input parameters of deep cone thickener prediction model

[0041] symbol meaning dimension <![CDATA[u1]]> Underflow <![CDATA[m 3 / h]]> <![CDATA[u2]]> Feed flow <![CDATA[m 3 / h]]> <![CDATA[u3]]> Underflow concentration ％ <![CDATA[u4]]> Feed concentration ％ <![CDATA[u5]]> Mud layer pressure MPA <![CDATA[u6]]> Overflow flow <![CDATA[m 3 / h]]> <![CDATA[u7]]> Rake 1 torque N·m <![CDATA[u8]]> Rake 2 torque N·m <![CDATA[u9]]> Rake 3 torque N·m <![CDATA[u 10 ]]> Rake 4 Torque N·m <![CDATA[u 11 ]]> Timestamp s

[0042] As shown in Table 1, for the deep cone thickener prediction model established, the system state is the underflow flow u1, and the control input is the feed flow set value u2, underflow concentration u3, feed concentration u4, mud layer pressure u5, overflow flow u6, rake torque u i ,i=7,…,10.

[0043] Reference Manual Attached Figure 2 , showing a schematic structural diagram of a deep cone concentrator provided in an embodiment of the present invention.

[0044] Figure 2 Among them, the deep cone thickener is a device used for solid-liquid separation. Its body is a cylindrical upper part and a conical lower part. The cone is deeper. This device works based on the principles of centrifugal force and gravity sedimentation. It can concentrate the slurry with a low solid content into an underflow slurry with a high solid content through flocculation, gravity sedimentation and other processes.

[0045] The main components of the deep cone thickener include agitator, hopper, deep cone barrel, bearing, transmission device, feed water pipe, discharge port, compressed air pipeline, etc. Among them, the deep cone barrel is the core component of the equipment, which consists of a cone, a cylinder and a cone head. During the working process, the slurry to be treated is introduced into the thickening area of ​​the deep cone thickener through the feed pipe, and the forced stirring device is turned on to stir the feed suspension evenly. Due to the centrifugal force, the solid particles begin to settle to the bottom of the thickening barrel. The higher the slurry concentration, the greater the centrifugal force, and the faster the solid particles settle. By adjusting parameters such as the stirring speed and the height of the overflow pipe, the concentration and concentration effect of the slurry can be controlled. When the concentration effect meets the requirements, the stirring device and the overflow pipe are closed, the feeding is stopped, and the solid-liquid separation is achieved. The concentrated slurry is discharged through the overflow pipe, and the cleaned material is discharged through the bottom discharge device. At the same time, the stability of the underflow concentration is a basic indicator for judging its process level. There are multiple parameters in the production process that affect the stability of the underflow concentration. The instability of the volume and concentration of the feed flow will cause the underflow concentration to fluctuate. Other parameters such as feed flow rate, underflow flow rate, etc. will also affect the underflow concentration.

[0046] S2: Preprocess the historical operation data.

[0047] It should be noted that by preprocessing historical data, the data quality can be improved and its consistency can be ensured, making model training more stable and reliable.

[0048] In a possible implementation, the preprocessing specifically includes abnormal segment removal preprocessing, missing value interpolation preprocessing, and z-score standardization preprocessing.

[0049] Among them, abnormal segment removal preprocessing refers to identifying and removing data segments in historical data that do not conform to normal operating conditions, such as abnormal situations such as sensor failure or unexpected shutdown. Missing value interpolation preprocessing refers to filling missing data points with corresponding interpolation methods to ensure the continuity and integrity of the data. Z-score standardization refers to normalizing the data according to the standard score, eliminating the dimensional differences between different data by removing the mean and scaling to unit standard deviation, so that they conform to the same scale and are convenient for model processing.

[0050] It should be noted that the removal of abnormal segments can eliminate biased data that is unfavorable to the model and avoid interfering with the model's learning process. Missing value interpolation can reasonably fill in the missing data and retain the continuity of key information. The z-score standardization balances the weights of different parameters, allowing them to play a reasonable role in training.

[0051] In a possible implementation, the missing value interpolation preprocessing specifically includes:

[0052] Determine the moment i in the historical time series where the historical operation data is missing.

[0053] It should be noted that in the industrial field, especially in the operation of deep cone thickeners, different sensors will collect data at different frequencies in response to their specific working cycles and monitoring tasks. For underflow flow u1 and feed flow u2, since they are directly affected by upstream and downstream processes and are more sensitive to changes, electromagnetic flow meters are used to collect data every 3 seconds in order to capture small changes in flow in a timely manner. Although underflow concentration u3 and feed concentration u4 are also affected by upstream and downstream processes, they change more slowly, so nuclear concentration meters are used to measure every 5 seconds. This frequency can not only ensure the real-time nature of the data, but also effectively control the measurement cost. As for mud layer pressure u5, overflow water flow u6 and rake frame torque u7, u8, u9, u 10 ,Since their changes are relatively slow, they do not need to be measured too frequently, so it is chosen to collect data every 15 seconds.

[0054] Specifically, in order to process these data uniformly, they need to be adjusted to the same time scale. Therefore, the lowest common multiple of the acquisition frequency of all variables, that is, 30 seconds, is selected as a unified time interval. However, due to the lack of synchronization benchmarks between sensors, the collected data are not aligned within the unified time interval, resulting in missing data for some variables in each 30-second interval. This data misalignment needs to be properly handled in subsequent data preprocessing and analysis to ensure that the model can accurately reflect the operating status and trends of the deep cone thickener. Missing historical operating data refers to the failure to record data at certain moments during the past operation of the thickener under the unified time interval standard due to the lack of synchronization between sensors. These data points are missing in the time series. The historical time series represents a set of thickener operating data arranged in chronological order, including various operating parameters at multiple time points, which are used to describe the operating change trend of the thickener.

[0055] Identify the nonmissing data surrounding the missing historical run data:

[0056] x i-2 ,x i-1 ,x i+1

[0057] Among them, use x i-2 ,x i-1 ,x i+1 As the interpolation point, x i-2 represents the non-missing data at time i-2, x i-1 represents the non-missing data at time i-1, x i+1 represents the non-missing data at time i+1, let y be the value of each interpolation point x, use Newton interpolation method, and use the missing point x i Nearby data points, let the interpolation point and its value satisfy the mapping yi =f(x i ), construct a second-order polynomial P(x) to estimate the missing values:

[0058] P(x)=f(x i-2 )+f[x i-2 ,x i-1 ](xx i-2 )+f[x i-2 ,x i-1 ,x i+1 ](xx i-2 )(xx i-1 )

[0059] The first-order difference quotient is:

[0060]

[0061] The second-order difference quotient is:

[0062]

[0063] Set x=x i Substitute into the polynomial p(x) and calculate p(x i ) as missing value y i The estimate,

[0064] For variables with small variations, use x i-1 ,x i+1 As interpolation points, linear interpolation is used, y is set as the value of each interpolation point x, and the following polynomial is constructed:

[0065]

[0066] Set x=x i Substitute into the polynomial p(x) and calculate p(x i ) as missing value y i Estimates.

[0067] Among them, non-missing historical operation data refers to the operation data points that are successfully recorded in the time series and have no gaps, that is, valid data. These data can be used as a reference to estimate missing values. Missing historical operation data values ​​refer to data points that are not recorded due to sensor failure or other reasons, that is, missing values ​​in the time series, which need to be estimated and completed to ensure data integrity.

[0068] The specific z-score standardization preprocessing is:

[0069]

[0070] in, represents the non-missing historical running data at time i after standardization, xi represents the non-missing historical running data at time i, μ i represents the mean of the non-missing historical running data at time i, σ i represents the standard deviation of non-missing historical running data at time i, and N represents the total number of historical running data.

[0071] It should be noted that the missing data points are replaced by the average value of the surrounding recorded data to fill the missing parts of the data and ensure the continuity of the data in the time series, while the z-score normalization balances the weights of different parameters so that it plays a reasonable role in training. The robustness and generalization ability of the model are enhanced through data preprocessing, which helps to improve the accuracy and stability of the prediction.

[0072] Reference Manual Attached Figure 3 , showing a schematic diagram of the principle of a state space equation provided by an embodiment of the present invention.

[0073] Figure 3 In the above figure, x represents the input historical time series of the state space model, B represents the input matrix, h represents the state variables of the historical operation data, A represents the state transfer matrix, C represents the output matrix, and y represents the predicted output of the state space model.

[0074] After the input historical time series data enters the model, the time series data is transformed or weighted through the input matrix, and the processed time series data is passed to the state variable. The state variable receives the time series data from the input matrix and the state transfer matrix and calculates the state change. The current state is fed back to the state variable through the state transfer matrix to further control the change and stability of the time series data state. The updated state is output as a vector to realize the control of the deep cone thickener.

[0075] It should be noted that when the state-space model is applied to a computer, the computer contacts and processes discrete data, so the zero-order hold technique is used to convert the discretized data into continuous data. First, each time a discrete signal is received, its value is retained until a new discrete signal is received. The result of this operation is the creation of a continuous signal that can be used by the state-space model. The time to maintain the value is represented by a new learnable parameter, called the step size, which represents the staged retention of the input. With continuous input data, continuous output can be generated, and the values ​​are sampled only according to the input time step. These sampled values ​​are discrete outputs.

[0076] In practical applications, basic state-space models perform poorly, especially when dealing with long-term dependencies. This is because the solutions to linear first-order ordinary differential equations are usually exponential functions, which may cause the gradient to grow exponentially in the sequence length, causing the problem of gradient vanishing or exploding. Therefore, the HIPPO (High-order Polynomial Projection Operators) theory is used. The HiPPO theory specifies a special class of matrices that, when incorporated into the SSM equations, enable the state to remember the historical information of the input. These special matrices are called HiPPO matrices, which have a specific mathematical form that can effectively capture long-term dependencies and replace the state transfer matrix with the HiPPO matrix. It attempts to compress all currently seen input signals into a coefficient vector and produce the optimal solution of the state matrix through function approximation.

[0077] Reference Manual Attached Figure 4 , showing a schematic structural diagram of a deep cone thickener prediction model provided by an embodiment of the present invention.

[0078] Figure 4 In the equation, u1…u i and x i Represents the historical operating data of the deep cone thickener, Represents the prediction result, t represents the time axis of the historical sequence and the prediction sequence, x represents the historical operation data sequence obtained, the historical sequence represents the historical operation data used to train the deep cone thickener prediction model, and the prediction sequence represents the future data to be predicted by the model. In order to deal with missing or insufficient data, the figure shows that the method of filling with 0 is used. The slicing processing means that the slicing operation module slices the input historical time series data according to a fixed length. This operation is similar to cutting a long time series into several shorter segments. The purpose of slicing is to facilitate subsequent feature extraction and help the model understand the data change characteristics within a local time period. Feature extraction is used to align the sliced ​​data into a two-dimensional matrix, one dimension of which is time, d modelIndicates that another dimension is the feature dimension. In this step, the original data is mapped to a high-dimensional space through feature projection, so that the data of each time segment can be represented in the high-dimensional feature space. Then, the high-dimensional feature data after feature alignment and projection is input into the Mamba module. The Mamba module is based on the framework of the state space model and can efficiently process long time series data. It can filter and extract key information through specific mechanisms (such as RNN-like gating mechanisms). Finally, the data processed by the Mamba module is sent to the prediction unit module. The prediction head represents the prediction unit module. The function of the prediction unit module is to generate the prediction results of the model. The prediction unit module can be a linear layer or other deep learning structure to generate future prediction values. The final prediction result It will be generated based on the historical data sequence, which means that the model uses past data to infer future trends and predict the operating status of the thickener.

[0079] S3: Under the state space model framework, a neural network-based deep cone thickener prediction model is constructed, wherein the deep cone thickener prediction model includes a slicing operation module, a Mamba module and a prediction unit module.

[0080] Among them, the state space model refers to a mathematical model used to represent the dynamic behavior of the system evolving over time. It describes the internal state of the system with state variables and establishes the relationship between the state and time. It is suitable for the prediction of dynamic systems. The neural network refers to a machine learning model that processes complex data by imitating the structure of neurons in the human brain. The neural network can automatically extract features and patterns from the data for prediction and classification. The deep cone thickener prediction model is an intelligent prediction model for the operating status of the deep cone thickener. The slicing operation module refers to the module that cuts long-term historical data into multiple sub-slices so that the neural network model can process short time series one by one to improve the computational efficiency and stability of the model. The Mamba module refers to a special structure in the model that is used to extract key features and correlations between variables from historical time series, so that the model can more effectively capture the dynamic behavior of the thickener. The prediction unit module refers to the module that receives the key features processed by the Mamba module and generates the final prediction results to help identify and predict future operating conditions.

[0081] It should be noted that the introduction of neural networks and their dedicated modules under the state-space model framework enables the model to flexibly cope with the complex dynamic characteristics of the deep cone thickener. The slicing operation module improves the data processing efficiency and facilitates the model to focus on local features. The Mamba module further enhances the accuracy of key feature extraction and helps to capture deep correlations in the data. The prediction unit module converts the extracted features into prediction outputs, thereby improving the accuracy of the prediction. Through this modular design, the model has stronger adaptability and robustness, which helps to achieve accurate prediction of the operating status of the thickener.

[0082] In a possible implementation, the sharding operation module in S3 is used to cut a long time series into multiple sub-shards, that is, to divide the entire time series into multiple time period shards, specifically including:

[0083] S301: Convert the preprocessed historical operation data into a historical time series, and input the historical time series into a deep cone thickener prediction model.

[0084] Among them, the historical time series refers to the operating data of the deep cone thickener arranged in chronological order. These data record the operating status of the equipment at different time points, such as flow rate, concentration, pressure, etc., and can reflect the dynamic change trend of the equipment.

[0085] Specifically, let x = [x1, x2, ..., x n ], x represents the historical time series, x n It represents the time series at the nth moment, including 11 variables (feed flow u1, underflow flow u2, underflow concentration u3, feed concentration u4, mud layer pressure u5, overflow flow u6, rake torque u7, rake torque u8, rake torque u9, rake torque u 10 , timestamp u 11 )’s measured value.

[0086] S302: Slice the input historical time series into fixed length segments:

[0087] x patch =x[:,:,stride:P+stride]

[0088] Among them, x patch represents the shards obtained by sharding processing, x represents the historical time series, P represents the shard length, and stride represents the step size.

[0089] S303: Perform value embedding and position embedding operations on each sub-shard:

[0090] x embed =ValueEmbedding(x patch)+PositionEmbedding(x patch )

[0091] Among them, ValueEmbedding represents the value embedding layer used to map each slice to a fixed-dimensional vector space. This embedding layer fuses the input data with time information along the slice length, and increases the slice length from 16 to 128. PositionEmbeddingPositionEmbedding represents the position embedding layer used to add position information to each slice. embed Represents the embedded historical time series.

[0092] Table 2 Results of hyperparameter tuning of deep cone thickener prediction model

[0093]

[0094] As shown in Table 2, the number of Mamba modules indicates the number of Mamba modules used in the model. The embedding layer dimension indicates the embedding layer dimension of the model, that is, the dimension size of the input feature. When the embedding dimension is 128, higher accuracy can be obtained. If a higher embedding dimension is selected, the computational complexity will be too large. The prediction length refers to the length of the time series generated by the model in the prediction task. MSE indicates the average of the squared error between the prediction result and the true value, which is used to measure the accuracy of the model prediction. The smaller the value, the better. MAE indicates the average of the absolute error between the prediction result and the true value, which is also an indicator to measure the prediction accuracy. The smaller the value, the better.

[0095] It should be noted that through the conversion of historical time series, the model can directly use the sorted operating data to facilitate subsequent operations. Then the sharding processing cuts the long time series into subsequences of appropriate length, helping the model focus on the data features within a short period of time, effectively reducing the computational burden and enhancing the analysis of local dynamics. The value embedding and position embedding operations enable each shard to have a fixed-dimensional representation in the vector space, while carrying time and position information, ensuring that the model can make full use of the numerical and temporal characteristics of the data, thereby improving the processing effect of time series data.

[0096] In one possible implementation, the Mamba module in S3 is used to extract key features and variable correlations in historical time series, specifically including:

[0097] S304: Processing historical time series through the selection mechanism in the Mamba module:

[0098] g t =σ(W g x t +b g) where gt is the selection gating vector at time step t, which determines the importance of each feature or variable at the current time step, and x t represents the data at time t in the historical time series, W g and b g represents a learnable parameter, and σ represents a sigmoid function, which is used to convert the linear output into a value between 0 and 1, indicating the importance of each feature.

[0099] It should be noted that by processing historical time series through the selection mechanism in the Mamba module, the processing of information can be adjusted according to the relevance and importance of the input data at each time step. This mechanism enables the model to identify and emphasize those variables that have a greater impact on paste concentration, such as feed flow and underflow flow, while reducing the weights of those variables with less impact.

[0100] S305: Update the output vector according to the state space model:

[0101]

[0102] in, represents the randomly initialized learnable parameters, h k represents the state vector at the current time k, h k-1 represents the state vector at time k-1, ⊙ represents element-by-element multiplication, which is used to transform the gate vector g t With the original data x t Multiplication is essentially weighting the data after selecting the importance of each variable. The model will weight the state h at the previous moment. k-1 and the current input x k Combined, calculate the current state h k .

[0103] S306: Generate a prediction output based on the updated output vector:

[0104] y k =Ch k

[0105] Among them, y k represents the predicted output at time k, and C represents the state vector h at the current time k k The output matrix converted to output.

[0106] It should be noted that this model can not only process the current input data, but also take into account the impact of past data on the current state, which is crucial for predicting the future state of the deep cone thickener. This capability enables the Mamba module to accurately simulate the dynamic interaction between variables and provide accurate concentration prediction and operation guidance for the deep cone thickener, thereby improving the model's prediction accuracy and enhancing its adaptability and stability in complex industrial environments.

[0107] S307: Extract key features from the prediction output through the feed-forward network layer in the Mamba module:

[0108] y=SiLU(W2SiLU(W1x+b1)+b2)

[0109] Among them, y represents the output result after the feedforward network layer, SiLU represents the SiLU activation function, and x represents the predicted output y k , W1 and W2 both represent the weights of the feedforward network layer, and b1 and b2 both represent the bias of the feedforward network layer.

[0110] It should be noted that the selection mechanism in the Mamba module effectively filters and selects key information of historical data, allowing the model to focus on the most important features for prediction, thereby improving the pertinence and efficiency of data processing. By updating the state vector, it ensures that the model can reflect the dynamic changes of the system and provide accurate state transition information, thereby enhancing the understanding of system behavior. The updated output vector is used to generate accurate prediction results, which combines the state of the previous moment and the current input to improve the consistency and reliability of the prediction. Finally, the key features in the prediction output are extracted through the feedforward network layer to further refine the information and enhance the model's ability to capture complex patterns.

[0111] In a possible implementation, the prediction unit module is used to receive the key features processed by the Mamba module and output the prediction results:

[0112] y pred =W out FFN(x)+b out

[0113] Among them, W out represents the weight of the prediction head output layer, b out represents the bias of the prediction head output layer, y pred Represents the prediction result, and FFN represents the feed-forward network layer.

[0114] It should be noted that the prediction unit module effectively receives and processes the key features extracted by the Mamba module to achieve high-quality prediction results, so that the model can achieve more accurate predictions, reduce errors, and improve the prediction quality of the deep cone thickener operating status.

[0115] S4: inputting the preprocessed historical operation data as a training set into the deep cone thickener prediction model to train the deep cone thickener prediction model until the loss function value of the deep cone thickener prediction model is less than a preset loss function value.

[0116] Among them, the preprocessed historical operating data refers to the cleaned data set obtained after preprocessing such as outlier removal, interpolation, and standardization of the historical operating data of the deep cone thickener; the training set refers to the preprocessed data set used to train the prediction model of the deep cone thickener to help the model learn the relationship between input data and output predictions; the loss function refers to an evaluation indicator used to measure the prediction effect of the model, which represents the deviation between the model prediction value and the actual value. By optimizing the loss function, the model continuously adjusts its parameters to improve the prediction accuracy; the preset loss function value refers to the threshold used to evaluate the end of model training. When the loss function is less than this value, the model is considered to have achieved the expected accuracy and the training can be stopped.

[0117] It should be noted that using preprocessed data as the training set ensures the quality of the model's input data, thereby improving the model's learning effect. By introducing a loss function evaluation mechanism, the model can detect its prediction error in real time, gradually optimize the model parameters, and ultimately achieve high-precision predictions. Setting a preset loss threshold also avoids the risk of over-training, saves computing resources and accelerates the training process.

[0118] It should be noted that those skilled in the art can set the size of the preset loss function value according to actual needs, and the present invention is not limited here.

[0119] In one possible implementation, the loss function of the deep cone thickener prediction model is a mean square error loss function;

[0120] The mean square error loss function is specifically:

[0121]

[0122] Among them, MSE represents the mean square error loss function, n represents the total number of samples in the historical running data, and y true Represents the actual value of each sample in the historical running data, y pred Represents the predicted value of each sample in the historical running data.

[0123] It should be noted that the MSE loss function provides a reliable optimization direction for the model, helping it to quickly converge to a low error state and ensure that the prediction results of the deep cone thickener are more accurate and stable.

[0124] S5: Output the trained deep cone thickener prediction model to complete the modeling of the deep cone thickener.

[0125] Among them, the trained deep cone thickener prediction model refers to the model that has been optimized, learned and achieved the target loss value through the training process. It has good prediction capabilities and can effectively predict the status of the deep cone thickener in actual operation.

[0126] It should be noted that the optimized deep cone thickener prediction model is output to ensure that the model has achieved the expected prediction accuracy and can be directly applied to the production process. While completing the modeling, a verified model is provided, which not only saves time for repeated adjustments, but also ensures that the model has good generalization ability in practical applications.

[0127] Table 3 Comparison of performance indicators of different models in predicting the underflow concentration of deep cone thickener

[0128]

[0129] As shown in Table 3, Patch_Mamba is the name of the model used in the present invention. AutoTransformer, Dlinear and Non-Stationary Transformer are all Transformer models for processing time series data. The model of the present invention is compared with the three, and the MSE and MAE of their output results are compared with the current mainstream models to obtain the lowest error. The comparison indicators are MAE, MSE, RMSE, RMSE, and MSPE, respectively. MAE represents mean absolute error, MSE represents mean square error, RMSE represents root mean square error, MAPE represents mean absolute percentage error, and MSPE represents mean square percentage error.

[0130] In the present invention, the IP address, port range, domain name, URL and / or TCP message that needs to be blocked is determined in an automated manner according to the abnormal category, so that the network abnormality can be responded to quickly and timely. No manual intervention is required, which can reduce the response time and improve the real-time performance and efficiency of the system. The target that needs to be blocked can be accurately determined according to the abnormal category, thereby preventing the spread of malicious attacks. The source of abnormal data and related information can be blocked in a timely manner to curb the spread of attack behaviors and protect network security. Corresponding bypass blocking based on the abnormal category of network transmission data can automatically respond to network abnormalities, prevent the spread of malicious attacks, protect system and data security, and improve network reliability. It also has flexibility and accuracy to effectively respond to various network security threats.

[0131] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0132] In the present invention, by obtaining historical operation data and preprocessing the historical operation data, a deep cone thickener prediction model based on a neural network is constructed under the framework of a state space model. By inputting the preprocessed historical operation data into the model and using a loss function for training, the model can gradually adjust its parameters, reduce the error between the predicted value and the true value, and finally obtain a trained model, thereby completing the modeling process of the deep cone thickener and realizing the efficiency and adaptability of the deep cone thickener modeling method based on the state space. The method uses slicing technology to process long sequence data, which helps to improve the model's ability to process long sequences. At the same time, the Mamba module uses the linear characteristics of the state space model to capture long-term dependencies in the sequence data through the selective state space, while the feedforward network layer can process more direct relationships between variables. By combining the Mamba module and the feedforward network layer, the algorithm can effectively handle variable correlation and time correlation problems, improve the accuracy of prediction, ensure the generalization ability of the model under different working conditions, optimize the operation speed, and improve the efficiency and stability of the paste filling process.

[0133] Reference Manual Attached Figure 5 , showing a structural schematic diagram of a deep cone concentration machine modeling system based on state space provided by the present invention.

[0134] The present invention further provides a state-space-based deep-cone concentration machine modeling system 20, which is applied to the state-space-based deep-cone concentration machine modeling method, comprising:

[0135] Processor 201.

[0136] The memory 202 stores computer-readable instructions. When the computer-readable instructions are executed by the processor 201, a state-space-based deep cone concentration machine modeling method as described in the method embodiment is implemented.

[0137] The state-space-based deep-cone concentration machine modeling system 20 provided in the present invention can execute the above-mentioned state-space-based deep-cone concentration machine modeling method and achieve the same or similar technical effects. To avoid repetition, the present invention will not go into details.

[0138] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0139] In the present invention, by obtaining historical operation data and preprocessing the historical operation data, a deep cone thickener prediction model based on a neural network is constructed under the framework of a state space model. By inputting the preprocessed historical operation data into the model and using a loss function for training, the model can gradually adjust its parameters, reduce the error between the predicted value and the true value, and finally obtain a trained model, thereby completing the modeling process of the deep cone thickener and realizing the efficiency and adaptability of the deep cone thickener modeling method based on the state space. The method uses slicing technology to process long sequence data, which helps to improve the model's ability to process long sequences. At the same time, the Mamba module uses the linear characteristics of the state space model to capture long-term dependencies in the sequence data through the selective state space, while the feedforward network layer can process more direct relationships between variables. By combining the Mamba module and the feedforward network layer, the algorithm can effectively handle variable correlation and time correlation problems, improve the accuracy of prediction, ensure the generalization ability of the model under different working conditions, optimize the operation speed, and improve the efficiency and stability of the paste filling process.

[0140] It should be understood that the processor in the embodiment of the present invention may be a central processing unit (CPU), and the processor may also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0141] It should also be understood that the memory in the embodiments of the present invention may be a volatile memory or a non-volatile memory, or may include both volatile and non-volatile memories. Among them, the non-volatile memory may be a read-only memory (ROM), a programmable read-only memory (PROM), an erasable programmable read-only memory (EPROM), an electrically erasable programmable read-only memory (EEPROM), or a flash memory. The volatile memory may be a random access memory (RAM), which is used as an external cache. By way of example and not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic random access memory (DRAM), synchronous DRAM (SDRAM), double data rate synchronous dynamic random access memory (DDR SDRAM), enhanced synchronous dynamic random access memory (ESDRAM), synchronous link DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0142] The above embodiments can be implemented in whole or in part by software, hardware (such as circuits), firmware or any other combination. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When a computer instruction or computer program is loaded or executed on a computer, a process or function according to an embodiment of the present invention is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. Computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center by wired (such as infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a tape), an optical medium (for example, a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0143] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.

[0144] In the present invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can be represented by: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or multiple.

[0145] It should be understood that in various embodiments of the present invention, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0146] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of the present invention.

[0147] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment, devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0148] In the several embodiments provided by the present invention, it should be understood that the disclosed devices, apparatuses and methods can be implemented in other ways. For example, the device embodiments described above are only schematic, for example, the division of units is only a logical function division, and there may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another device, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0149] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0150] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0151] If the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods of various embodiments of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.

[0152] An embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the program is executed by a processor, a state-space-based deep cone concentration machine modeling method as described in the method embodiment is implemented.

[0153] A computer-readable storage medium provided by the present invention can implement the steps and effects of the deep cone concentration machine modeling method based on state space in the above method embodiment. To avoid repetition, the present invention will not go into details.

[0154] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0155] In the present invention, by obtaining historical operation data and preprocessing the historical operation data, a deep cone thickener prediction model based on a neural network is constructed under the framework of a state space model. By inputting the preprocessed historical operation data into the model and using a loss function for training, the model can gradually adjust its parameters, reduce the error between the predicted value and the true value, and finally obtain a trained model, thereby completing the modeling process of the deep cone thickener and realizing the efficiency and adaptability of the deep cone thickener modeling method based on the state space. The method uses slicing technology to process long sequence data, which helps to improve the model's ability to process long sequences. At the same time, the Mamba module uses the linear characteristics of the state space model to capture long-term dependencies in the sequence data through the selective state space, while the feedforward network layer can process more direct relationships between variables. By combining the Mamba module and the feedforward network layer, the algorithm can effectively handle variable correlation and time correlation problems, improve the accuracy of prediction, ensure the generalization ability of the model under different working conditions, optimize the operation speed, and improve the efficiency and stability of the paste filling process.

[0156] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art who is familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

[0157] There are a few points to note:

[0158] (1) The drawings of the embodiments of the present invention only relate to the structures related to the embodiments of the present invention, and other structures may refer to the general design.

[0159] (2) For the sake of clarity, in the drawings used to describe the embodiments of the present invention, the thickness of the layers or regions is exaggerated or reduced, that is, these drawings are not drawn according to the actual scale. It is understood that when an element such as a layer, film, region or substrate is referred to as being "on" or "under" another element, the element may be "directly" "on" or "under" the other element or there may be intermediate elements.

[0160] (3) In the absence of conflict, the embodiments of the present invention and the features therein may be combined with each other to obtain new embodiments.

[0161] The above are only specific embodiments of the present invention, but the protection scope of the present invention is not limited thereto. The protection scope of the present invention shall be based on the protection scope of the claims.

Claims

1. A deep cone thickener modeling method based on state space, characterized in that: include: S1: Obtain historical operation data of deep cone thickener; S2: preprocessing the historical operation data; S3: Under the state space model framework, a deep cone thickener prediction model based on a neural network is constructed, wherein the deep cone thickener prediction model includes a slicing operation module, a Mamba module and a prediction unit module; S4: inputting the preprocessed historical operation data as a training set into the deep cone thickener prediction model to train the deep cone thickener prediction model until the loss function value of the deep cone thickener prediction model is less than a preset loss function value; S5: Output the trained deep cone thickener prediction model to complete the modeling of the deep cone thickener.

2. The state-space based deep cone thickener modeling method according to claim 1, characterized in that: The historical operation data include underflow flow, feed flow, underflow concentration, feed concentration, mud layer pressure, overflow flow and rake torque.

3. The state-space based deep cone thickener modeling method according to claim 1, characterized in that: The preprocessing specifically includes abnormal segment removal preprocessing, missing segment interpolation preprocessing and z-score standardization preprocessing.

4. The state-space based deep cone thickener modeling method according to claim 3, characterized in that: The missing value interpolation preprocessing specifically includes: Determine the moment i in the historical time series at which the historical operation data is missing; Determine the non-missing data around the missing historical run data: x i-2 ,x i-1 ,x i+1 Among them, use x i-2 ,x i-1 ,x i+1 As the interpolation point, x i-2 represents the non-missing data at time i-2, x i-1 represents the non-missing data at time i-1, x i+1 represents the non-missing data at time i+1, let y be the value of each interpolation point x, use Newton interpolation method, and use the missing point x i Nearby data points, let the interpolation point and its value satisfy the mapping y i =f(x i ), construct a second-order polynomial P(x) to estimate the missing values: P(x)=f(x i-2 )+f[x i-2 ,x i-1 ](x-x i-2 )+f[x i-2 ,x i-1 ,x i+1 ](x-x i-2 )(x-x i-1 ) The first-order difference quotient is: The second-order difference quotient is: x = x i Substitute into the polynomial p(x) and calculate p(x i ) as missing value y i The estimate, For variables with small changes, use x i-1 ,x i+1 As interpolation points, linear interpolation is used, y is set as the value of each interpolation point x, and the following polynomial is constructed: x = x i Substitute into the polynomial p(x) and calculate p(x i ) as missing value y i The z-score standardization preprocessing is specifically as follows: in, represents the non-missing historical running data at time i after standardization, x i represents the non-missing historical running data at time i, μ i represents the mean of the non-missing historical running data at time i, σ i represents the standard deviation of non-missing historical running data at time i, and N represents the total number of historical running data.

5. The deep cone thickener modeling method based on state space according to claim 1 is characterized in that: The sharding operation module in S3 is used to cut a long time series into multiple sub-shards, that is, to divide the entire time series into multiple time period shards, specifically including: S301: converting the pre-processed historical operation data into a historical time series, and inputting the historical time series into the deep cone thickener prediction model; S302: Slice the input historical time series into fixed length segments: x patch =x[:,:,stride:P+stride] Among them, x patch represents the shards obtained by sharding, x represents the historical time series, P represents the shard length, and stride represents the step length; S303: Perform value embedding and position embedding operations on each sub-shard: x embed =ValueEmbedding(x patch )+PositionEmbedding(x patch ) Among them, ValueEmbedding represents the value embedding layer used to map each shard to a vector space of fixed dimension, PositionEmbeddingPositionEmbedding represents the position embedding layer used to add position information to each shard, and x embed Represents the embedded historical time series.

6. The state-space based deep cone thickener modeling method according to claim 5, characterized in that: The Mamba module in S3 is used to extract key features and variable correlations in historical time series, including: S304: Processing the historical time series through the selection mechanism in the Mamba module: g t =σ(W g x t +b g ) Among them, gt is the selection gating vector at time step t, which determines the importance of each feature or variable at the current time step, and x t represents the data at time t in the historical time series, W g and b g represents a learnable parameter, σ represents the sigmoid function, which is used to convert the linear output into a value between 0 and 1, indicating the importance of each feature; S305: Update the output vector according to the state space model: in, represents the randomly initialized learnable parameters, h k represents the state vector at the current time k, h k-1 represents the state vector at time k-1, ⊙ represents element-by-element multiplication, which is used to transform the gate vector g t With the original data x t Multiplication is essentially weighting the data after selecting the importance of each variable. The model will weight the state h at the previous moment. k-1 and the current input x k Combined, calculate the current state h k ; S306: Generate a prediction output based on the updated output vector: the k =Ch k Among them, y k represents the predicted output at time k, and C represents the state vector h at the current time k k Transformed into the output matrix of the output; S307: Extract key features in the predicted output through the feed-forward network layer in the Mamba module: y=SiLU(W2SiLU(W1x+b1)+b2) Among them, y represents the output result after the feedforward network layer, SiLU represents the SiLU activation function, and x represents the predicted output y k , W1 and W2 both represent the weights of the feedforward network layer, and b1 and b2 both represent the bias of the feedforward network layer.

7. The state-space based deep cone thickener modeling method according to claim 6, characterized in that: The prediction unit module is used to receive the key features processed by the Mamba module and output the prediction results: the pred =W out FFN(x)+b out Among them, W out represents the weight of the prediction head output layer, b out represents the bias of the prediction head output layer, y pred Represents the prediction result, and FFN represents the feed-forward network layer.

8. The state-space based deep cone thickener modeling method according to claim 1, characterized in that: The loss function of the deep cone thickener prediction model is a mean square error loss function; The mean square error loss function is specifically: Among them, MSE represents the mean square error loss function, n represents the total number of samples in the historical running data, and y true Represents the actual value of each sample in the historical running data, y pred Represents the predicted value of each sample in the historical running data.

9. A deep cone thickener modeling system based on state space, characterized in that: include: processor; A memory having computer-readable instructions stored thereon, wherein when the computer-readable instructions are executed by the processor, the state-space-based deep-cone thickener modeling method according to any one of claims 1 to 8 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the state-space-based deep-cone thickener modeling method according to any one of claims 1 to 8 is implemented.

Images