Minimum p norm random configuration network-based ore pulp density intelligent detection method

Abstract

The invention discloses an ore pulp density intelligent detection method based on a minimum p-norm random configuration network, and the method comprises the steps: carrying out the mechanism analysis of the flowing process of ore pulp in a pipeline, and building an ore pulp density detection mixed model which comprises a linear part and a nonlinear part; identifying a linear part of the ore pulp density detection mixed model by adopting a least square method, establishing an ore pulp density detection linear model, and outputting to obtain an estimated value of the ore pulp density linear model; a preset minimum p norm random configuration network is adopted to identify the nonlinear part of the ore pulp density detection mixed model, an ore pulp density detection nonlinear model is established, and a compensation value of the ore pulp density nonlinear model is obtained through output; and fusing the estimated value of the ore pulp density linear model and the compensation value of the ore pulp density nonlinear model to obtain a final ore pulp density predicted value, and outputting a prediction result of the ore pulp density. The method can effectively suppress interference and improve the stability and measurement precision of the model.

Description

Technical Field

[0001] The present invention relates to the technical field of key operating indicator detection in a grinding and classification process, and in particular to an intelligent detection method for slurry density based on a minimum p-norm random configuration network. Background Art

[0002] Slurry density is a critical parameter in the mineral processing process, and accurate and real-time slurry density measurement is crucial for production guidance. Whether slurry density is within a reasonable range directly impacts process quality and equipment performance upstream and downstream of the mineral processing process. Traditional methods for accurately measuring slurry density in flowing pipelines include the density pot method and the densitometer method. The density pot method involves filling a density pot with slurry liquid, measuring the weight using a spring scale or platform scale, subtracting the weight of the density pot, and manually calculating the slurry density using a formula. The density pot method is simple to operate and offers high accuracy for slurry samples. However, manual sampling and testing is slow, and test results exhibit time lags. Furthermore, manual sampling cycles are generally long, making real-time density measurement impossible, making it difficult to meet the real-time parameter monitoring requirements of industrial production environments. This method can improve the accuracy of real-time testing and optimize production efficiency and cost-effectiveness.

[0003] In related technologies, densitometers measure density indirectly through the interaction of a physical field with the slurry. Ultrasonic densitometers calculate density based on the attenuation of gamma rays as they penetrate the slurry. Nuclear densitometers correlate the propagation speed or attenuation of sound waves in the slurry with density, inferring density from the transmitted and received signals. Differential pressure densitometers measure the static pressure difference between two points in a pipeline and infer density based on flow velocity and pipeline height differences. However, these methods have issues such as high cost, limited application, and low accuracy. Summary of the Invention

[0004] The present invention aims to address, at least to some extent, one of the technical problems in the related art. To this end, a first object of the present invention is to propose an intelligent slurry density detection method based on a minimum p-norm randomly configured network. This method effectively suppresses unknown noise interference in industrial processes, improving model stability and measurement accuracy.

[0005] To achieve the above objectives, a first embodiment of the present invention proposes an intelligent detection method for slurry density based on a minimum p-norm random configuration network, the method comprising:

[0006] S1, by analyzing the mechanism of the slurry flow process in the pipeline, establishing a slurry density detection hybrid model, wherein the slurry density detection hybrid model includes a linear part and a nonlinear part;

[0007] S2, using the least squares method to identify the linear part of the slurry density detection mixed model, establish a slurry density detection linear model, and output an estimated value of the slurry density linear model;

[0008] S3, using a preset minimum p-norm random configuration network to identify the nonlinear part of the slurry density detection hybrid model, establish a nonlinear model for slurry density detection, and output a compensation value of the slurry density nonlinear model;

[0009] S4, fusing the estimated value of the slurry density linear model and the compensation value of the slurry density nonlinear model to obtain the final slurry density prediction value, and outputting the prediction result of the slurry density.

[0010] In addition, the method for intelligent detection of slurry density based on a minimum p-norm random configuration network according to the above embodiment of the present invention may also have the following additional technical features:

[0011] According to one embodiment of the present invention, the slurry density detection hybrid model is expressed by the following formula:

[0012] ρ(t)=ρ0(t)+Δρ(t)

[0013] ρ0(t)=k1p H (t)+k2p L (t)+k3

[0014] Δρ(t)=l(p H (t),p L (t),f(t),i(t))

[0015] Wherein, ρ0(t) is the linear part of the hybrid model for slurry density detection, k1, k2, and k3 are unknown parameters; Δρ(t) is the nonlinear part of the hybrid model for slurry density detection, and l(·) represents the unknown nonlinear term including instrument measurement error and slurry flow process.

[0016] According to one embodiment of the present invention, step S2 includes:

[0017] Based on the data collected on site, define X(k-1)=[x T (k-1),...,x T (kN)] T is N-dimensional input data, where P H (k) represents the high voltage signal at the kth moment, P L (k) represents the low-voltage signal at the kth moment; define Y(k-1) = [y(k-1),…,y(kN)] as the N-dimensional output data; θ(k-1) = [k1, k2, k3] T;Y(k-1)=X(k-1)θ(k-1)+ε, where ε is the residual;

[0018] The least squares objective function is:

[0019] J(θ)=||Y-Xθ|| 2 =(Y-Xθ) T (Y-Xθ)=Y T Y-2θ T X T Y+θ T X T Xθ

[0020] Derivative the above formula with respect to the parameter θ yields:

[0021]

[0022] Finally, let the above formula be 0 and solve it:

[0023] X T Xθ=X T Y,

[0024] θ=(X T X) -1 X T Y

[0025] in,

[0026] Thus, the parameters k1, k2, k3 of the slurry density linear model can be obtained, the linear part can be identified, and the linear model of slurry density detection can be established.

[0027] The high-voltage signal P of the slurry flow pipeline H , low voltage signal P L As the input of the linear model, the output can be the estimated value of the linear model of slurry density

[0028] According to one embodiment of the present invention, step 3 includes:

[0029] Based on the collected process parameters, including the high-pressure signal P of the slurry flow pipeline H , low voltage signal P L , slurry pump current I and slurry pump frequency f, and the corresponding artificial test slurry density value ρ are used as data for training the model;

[0030] The artificial test value of slurry density ρ and the linear model estimate The difference As the label of the nonlinear model, a nonlinear model for slurry density detection is established through the minimum p-norm random configuration network;

[0031] The high-voltage signal P of the slurry flow pipeline H , low voltage signal P L , slurry pump current I and slurry pump frequency f are used as inputs of the nonlinear model, and the output is the compensation value of the slurry density nonlinear model

[0032] According to one embodiment of the present invention, the minimum p-norm random configuration network model construction includes:

[0033] S31, X=[P H ,P L ,f,I] as the input of the random configuration network with the minimum p norm, and initialize the maximum number of hidden layer nodes L max , maximum number of configuration times T max , critical tolerance error ε, random parameter range [ - λ max ,λ max ] and the change step size Δλ;

[0034] S32, assuming that the number of nodes in the current hidden layer is L-1, calculate the network output residual e L-1 , if L≤L max ,||e L-1 || 2 >∈, then the Lth hidden layer node is added, and the supervision mechanism for selecting the hidden layer node is given by the following formula:

[0035]

[0036] in, Given 1-∈<r<1, μ L =(1-r) / (L+1);

[0037] S33, random configuration T max Each configuration will randomly select the input weight ω of the Lth hidden layer node in the supervision mechanism. L and bias b L , and calculate ξ L,q ; If minξ L,1 ,ξ L,2 ≥0, then ω L 、b L ,ξ L,q Store; if all L,q If none of them meet the conditions, a larger r value will be selected for reconfiguration; after the random configuration is completed, the largest The corresponding ω L 、b L As the input weight and bias of the Lth node;

[0038] S34, after determining the model structure, calculate the output weight β:

[0039] According to the definition of p-norm, the p-norm of a given vector x can be expressed as:

[0040]

[0041] The objective function is constructed by using the p-power of the error vector p-norm and combining it with the regularization term The objective function is obtained as:

[0042]

[0043] in,

[0044] At this time, an implicit equation about β is obtained, and then the fixed point iteration method is used to solve the implicit equation, let

[0045]

[0046] Then the core iterative process of fixed point iteration can be expressed as

[0047] β(t)=f(β(t-1)),

[0048] Where β(t) is the solution of the output weight obtained at the t-th iteration;

[0049] S35, when L≤L max or ||e L-1 || 2 When >∈, stop adding nodes; finally, the output compensation value of the data-driven model is obtained, which is expressed as:

[0050]

[0051] In this way, the identification of the nonlinear part is completed and a nonlinear model for slurry density detection is established.

[0052] According to one embodiment of the present invention, step S4 includes:

[0053] The estimated value of the slurry density linear model and the compensation value of the slurry density nonlinear model are added together to output the final slurry density prediction value.

[0054] The intelligent detection method for slurry density based on a minimum p-norm random configuration network according to an embodiment of the present invention has the following beneficial effects:

[0055] By analyzing the mechanics of slurry density during the beneficiation process and modeling the known components, a linear estimation model for slurry density detection is constructed. Data modeling is performed on the unknown, strongly nonlinear components to construct a nonlinear compensation model for slurry density. Finally, the outputs of the two models are added together to obtain the final slurry density prediction. This method can meet the requirements for accurate and stable slurry density detection in complex industrial processes.

[0056] Additional aspects and advantages of the present invention will be set forth in part in the description which follows and, in part, will be obvious from the description which follows, or may be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 Flowchart of a method for intelligent detection of slurry density based on a minimum p-norm random configuration network according to an embodiment of the present invention;

[0058] Figure 2 The figure is a flow chart of a method for intelligent detection of slurry density based on a minimum p-norm random configuration network according to an embodiment of the present invention. DETAILED DESCRIPTION

[0059] The following describes embodiments of the present invention in detail, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present invention, and are not to be construed as limiting the present invention.

[0060] The following describes an intelligent detection method for slurry density based on a minimum p-norm random configuration network according to an embodiment of the present invention with reference to the accompanying drawings.

[0061] like Figure 1 As shown, the slurry density intelligent detection method based on the minimum p-norm random configuration network according to the embodiment of the present invention may include the following steps:

[0062] S1, by analyzing the mechanism of the slurry flow process in the pipeline, a slurry density detection hybrid model is established. The slurry density detection hybrid model includes a linear part and a nonlinear part.

[0063] According to one embodiment of the present invention, the slurry density detection hybrid model is expressed by the following formula:

[0064] ρ(t)=ρ0(t)+Δρ(t)

[0065] ρ0(t)=k1p H (t)+k2p L (t)+k3

[0066] Δρ(t)=l(pH (t),p L (t),f(t),i(t))

[0067] Where ρ0(t) is the linear part of the hybrid model for slurry density detection, k1, k2, and k3 are unknown parameters, Δρ(t) is the nonlinear part of the hybrid model for slurry density detection, and l(·) represents the unknown nonlinear term including instrument measurement error and the slurry flow process.

[0068] Specifically, under ideal conditions, if there is no resistance loss, the pressure difference is:

[0069] Δp=ρgΔH

[0070] Where ρ is the slurry density, g is the acceleration of gravity, and ΔH is the height difference of the liquid surface. During the slurry flow, there will be resistance loss along the flow, which is calculated as follows:

[0071]

[0072] Where γ is the resistance coefficient along the way, L is the pipe length, D is the pipe diameter, V is the average flow velocity in the section, and g is the acceleration due to gravity. In actual industrial processes, the total pressure difference Δp = ρgΔH-H f , we can get:

[0073]

[0074] Directly using the pressure difference signal as the input of the density measurement model may affect the measurement results to some extent. In addition, when using pressure instruments to measure pressure, systematic errors and random errors will occur. Therefore, it is necessary to measure the absolute pressure p on the high-pressure side. H and low pressure side absolute pressure p L Make the corresponding correction, and the pressure difference measured at time t is Δp(t)=p H (t)-p L (t) is rewritten as:

[0075] Δp(t)=Ap H (t)-Bp L (t)+C+l1(p H (t),p L (t))

[0076] Where A and B are the correction coefficients for high and low pressure, C is the offset term, and l1(·) represents the unknown nonlinear error in measuring pressure. The unknown nonlinear expressions for the average flow velocity V and the slurry pump current i and frequency f are as follows:

[0077] V(t)=l2(f(t),i(t))

[0078] Arranged:

[0079] ρ(t)=ρ0(t)+Δρ(t)

[0080] in,

[0081] ρ0(t)=k1p H (t)+k2p L (t)+k3

[0082] Δρ(t)=l(p H (t),p L (t),f(t),i(t))

[0083]

[0084] In the formula, l(·) represents the unknown nonlinear term including the instrument measurement error and the slurry flow process. From the above formula, we can see that the linear part is related to P H 、P L The nonlinear part is related to P H 、P L , I and f are related.

[0085] S2, using the least squares method to identify the linear part of the slurry density detection mixed model, establish the slurry density detection linear model, and output the estimated value of the slurry density linear model.

[0086] Specifically, the least square method is used to identify the linear part of the mixed model. Based on the data collected on site, define X(k-1)=[x T (k-1),...,x T (kN)] T is N-dimensional input data, where P H (k), P L (k) represents the high and low voltage signals at the kth moment; Y(k-1) = [y(k-1), ..., y(kN)] is the N-dimensional output data; θ(k-1) = [k1, k2, k3] T Ideally, there is However, since the data may be contaminated by interference and measurement noise, it is considered to introduce the residual ε1, and in this case Y(k-1)=X(k-1)θ(k-1)+ε1.

[0087] The least squares objective function is:

[0088] J(θ)=||Y-Xθ|| 2 =(Y-Xθ) T (Y-Xθ)

[0089] =Y T Y-2θT X T Y+θ T X T Xθ

[0090] Derivative the above formula with respect to parameter θ yields:

[0091]

[0092] Finally, let the above formula be 0 and solve it:

[0093] X T Xθ=X T Y,

[0094] θ=(X T X) -1 X T Y

[0095] in, The parameters k1, k2, k3 of the linear model of slurry density can be obtained, the identification of the linear part can be completed, and the linear model of slurry density detection can be established. H , low voltage signal P L As the input of the linear model, the output can be the estimated value of the linear model of slurry density

[0096] S3, using the preset minimum p-norm random configuration network to identify the nonlinear part of the slurry density detection hybrid model, establish a nonlinear model for slurry density detection, and output the compensation value of the slurry density nonlinear model. Figure 2 As shown, the nonlinear model is the LP-SCN nonlinear model.

[0097] According to one embodiment of the present invention, step 3 includes:

[0098] Based on the collected process parameters, including the high-pressure signal P of the slurry flow pipeline H , low voltage signal P L , slurry pump current I and slurry pump frequency f, and the corresponding artificial test slurry density value ρ are used as data for training the model;

[0099] The artificial test value of slurry density ρ and the linear model estimate The difference As the label of the nonlinear model, a nonlinear model for slurry density detection is established through the minimum p-norm random configuration network;

[0100] The high-voltage signal P of the slurry flow pipeline H , low voltage signal P L, slurry pump current I and slurry pump frequency f are used as inputs of the nonlinear model, and the output is the compensation value of the slurry density nonlinear model

[0101] According to one embodiment of the present invention, the minimum p-norm random configuration network model construction includes:

[0102] S31, X=[P H ,P L ,f,I] as the input of the random configuration network with the minimum p norm, and initialize the maximum number of hidden layer nodes L max , maximum number of configuration times T max , critical tolerance error ε, expected accuracy ∈, random parameter range [ - λ max ,λ max ] and the change step size Δλ;

[0103] S32, assuming that the number of nodes in the current hidden layer is L-1, calculate the network output residual e L-1 , if L≤L max ,||e L-1 || 2 >∈, then the Lth hidden layer node is added, and the supervision mechanism for selecting the hidden layer node is given by the following formula:

[0104]

[0105] in, Given 1-∈<r<1, μ L =(1-r) / (L+1);

[0106] S33, random configuration T max Each configuration will randomly select the input weight ω of the Lth hidden layer node in the supervision mechanism. L and bias b L , and calculate ξ L,q ; If minξ L,1 ,ξ L,2 ≥0, then ω L 、b L ,ξ L,q Store; if all L,q If none of them meet the conditions, a larger r value will be selected for reconfiguration; after the random configuration is completed, the largest The corresponding ω L 、b L As the input weight and bias of the Lth node;

[0107] S34, after determining the model structure, calculate the output weight β:

[0108] According to the definition of p-norm, the p-norm of a given vector x can be expressed as:

[0109]

[0110] The objective function is constructed by using the p-power of the error vector p-norm and combining it with the regularization term The objective function is obtained as:

[0111]

[0112] in

[0113] At this time, an implicit equation about β is obtained, and then the fixed point iteration method is used to solve the implicit equation, let

[0114]

[0115] Then the core iterative process of fixed point iteration can be expressed as

[0116] β(t)=f(β(t-1)),

[0117] Where β(t) is the solution of the output weight obtained at the t-th iteration;

[0118] S35, when L≤L max or ||e L-1 || 2 When >∈, stop adding nodes; finally, the output compensation value of the data-driven model is obtained, which is expressed as:

[0119]

[0120] In this way, the identification of the nonlinear part is completed and a nonlinear model for slurry density detection is established.

[0121] Specifically, a random configuration network based on the minimum p norm is used as an algorithm to identify nonlinear terms. Based on the collected historical process parameters, including the high-pressure signal P of the slurry flow pipeline, H , low voltage signal P L , slurry pump current I and slurry pump frequency f, and the corresponding artificial test slurry density value ρ are used as the data for training the model. The artificial test value ρ of slurry density and the linear model estimate are used as the training data. The difference As the label of the nonlinear model, a nonlinear model for slurry density detection is established through the minimum p-norm random configuration network.

[0122] The specific steps for constructing the minimum p-norm random configuration network model are as follows:

[0123] (1) X = [P H,P L ,f,I] as the input of the random configuration network with the minimum p norm, and initialize the maximum number of hidden layer nodes L max , maximum number of configuration times T max , critical tolerance error ε, random parameter range [ - λ max ,λ max ] and the change step size Δλ;

[0124] (2) Assuming that the number of nodes in the current hidden layer is L-1, calculate the network output residual e L-1 , if L≤L max ,||e L-1 || 2 >∈, then the Lth hidden layer node is added, and the supervision mechanism for selecting the hidden layer node is given by the following formula:

[0125]

[0126] in, Given 1-∈<r<1, μ L =(1-r) / (L+1).

[0127] (3) Randomly configure T max Each configuration will randomly select the input weight ω of the Lth hidden layer node in the supervision mechanism. L and bias b L , and calculate ξ L,q ; If minξ L,1 ,ξ L,2 ≥0, then ω L 、b L ,ξ L,q Store; if all L,q If none of them meet the conditions, a larger r value will be selected for reconfiguration; after the random configuration is completed, the largest The corresponding ω L 、b L As the input weight and bias of the Lth node;

[0128] (4) After determining the model structure, calculate the output weight β:

[0129] According to the definition of p-norm, the p-norm of a given vector x can be expressed as:

[0130]

[0131] The objective function is constructed using the p-power of the error vector p-norm and combined with the regularization term The objective function is obtained as:

[0132]

[0133] By taking the derivative of the objective function with respect to β, we can obtain:

[0134]

[0135] in is a diagonal matrix consisting of errors.

[0136] Let the derivative be 0, we can get:

[0137]

[0138] in At this point, we get an implicit equation about β, and then use the fixed point iteration method to solve the implicit equation, let

[0139]

[0140] Then the core iterative process of fixed point iteration can be expressed as

[0141] β(t)=f(β(t-1))

[0142] Where β(t) is the solution of the output weight obtained at the t-th iteration. In actual operation, an initialized weight vector can be randomly given and a maximum number of iterations can be set to start and end the iteration process.

[0143] (5) When L≤L max or ||e L-1 || 2 When >∈, stop adding nodes. Finally, the output compensation value of the data-driven model is obtained, which is expressed as:

[0144]

[0145] The nonlinear part is thus identified and the nonlinear model of slurry density detection is established. H , low voltage signal P L , slurry pump current I and slurry pump frequency f are used as inputs of the nonlinear model, and the output is the compensation value of the slurry density nonlinear model

[0146] S4, fusing the estimated value of the slurry density linear model and the compensation value of the slurry density nonlinear model to obtain the final slurry density prediction value, and outputting the prediction result of the slurry density.

[0147] According to one embodiment of the present invention, step S4 includes: adding the estimated value of the slurry density linear model and the compensation value of the slurry density nonlinear model to output a final slurry density prediction value.

[0148] Specifically, the slurry density linear model estimate obtained in step S2 is and the slurry density nonlinear model compensation value obtained in step S3 Add and output the final predicted value of slurry density

[0149] After on-site collection and screening, the data set is divided into 3:1, the first 3 / 4 of the data set is used as the model training set, and the last 1 / 4 of the data set is used as the model test set. At the same time, the data set needs to be processed with Min-Max Normalization, so that X = [P H ,P L ,f,I], the normalization process is obtained by the following formula:

[0150]

[0151] The algorithm designed by the present invention selects the expected accuracy ∈, the maximum number of random configurations T through the hyperparameter tuning technology of cross validation and grid search max , regularization coefficient C and other parameters, and the mean absolute error (MAE), root mean square error (RMSE) and determination coefficient (R-square, R2) are used as model evaluation indicators:

[0152]

[0153] Where n is the number of samples, y i is the i-th true value, is the i-th predicted value, is the average of all true values.

[0154] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0155] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "plurality" means at least two, such as two, three, etc., unless otherwise specifically defined.

[0156] In the present invention, unless otherwise specified or limited, the terms "installed," "connected," "connect," "fixed," etc. should be understood in a broad sense. For example, they can refer to fixed connection, detachable connection, or integration; mechanical connection, electrical connection; direct connection, or indirect connection through an intermediate medium; internal communication between two components, or interaction between two components, unless otherwise specified. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0157] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. An intelligent detection method for slurry density based on a minimum p-norm random configuration network, characterized in that: The method comprises: S1, by analyzing the mechanism of the slurry flow process in the pipeline, establishing a slurry density detection hybrid model, wherein the slurry density detection hybrid model includes a linear part and a nonlinear part; S2, using the least squares method to identify the linear part of the slurry density detection mixed model, establish a slurry density detection linear model, and output an estimated value of the slurry density linear model; S3, using a preset minimum p-norm random configuration network to identify the nonlinear part of the slurry density detection hybrid model, establish a nonlinear model for slurry density detection, and output a compensation value of the slurry density nonlinear model; S4, fusing the estimated value of the slurry density linear model and the compensation value of the slurry density nonlinear model to obtain the final slurry density prediction value, and outputting the prediction result of the slurry density.

2. The method for intelligent detection of slurry density based on minimum p-norm random configuration network according to claim 1 is characterized in that: The slurry density detection hybrid model is expressed by the following formula: ρ(t)=ρ0(t)+Δρ(t) ρ0(t)=k1p H (t)+k2p L (t)+k3 Δρ(t)=l(p H (t),p L (t),f(t),i(t)) Wherein, ρ0(t) is the linear part of the hybrid model for slurry density detection, k1, k2, and k3 are unknown parameters; Δρ(t) is the nonlinear part of the hybrid model for slurry density detection, and l(·) represents the unknown nonlinear term including instrument measurement error and slurry flow process.

3. The method for intelligent detection of slurry density based on minimum p-norm random configuration network according to claim 2, characterized in that: Step S2 includes: Based on the data collected on site, define X(k-1)=[x T (k-1),...,x T (kN)] T is N-dimensional input data, where P H (k) represents the high voltage signal at the kth moment, P L (k) represents the low-voltage signal at the kth moment; define Y(k-1) = [y(k-1), ..., y(kN)] as the N-dimensional output data; θ(k-1) = [k1, k2, k3] T ;Y(k-1)=X(k-1)θ(k-1)+ε, where ε is the residual; The least squares objective function is: J(θ)=||Y-Xθ|| 2 =(Y-Xθ) T (Y-Xθ)=Y T Y-2θ T X T Y+θ T X T Xth Derivative the above formula with respect to the parameter θ yields: Finally, let the above formula be 0 and solve it: X T Xθ=X T Y, θ=(X T X) -1 X T Y in, Thus, the parameters k1, k2, k3 of the slurry density linear model can be obtained, the linear part can be identified, and the linear model of slurry density detection can be established. The high-voltage signal P of the slurry flow pipeline H , low voltage signal P L As the input of the linear model, the output can be the estimated value of the linear model of slurry density 4. The method for intelligent detection of slurry density based on minimum p-norm random configuration network according to claim 3 is characterized in that: Step 3 includes: Based on the collected process parameters, including the high-pressure signal P of the slurry flow pipeline H , low voltage signal P L , slurry pump current I and slurry pump frequency f, and the corresponding artificial test slurry density value ρ are used as data for training the model; The artificial test value of slurry density ρ and the linear model estimate The difference As the label of the nonlinear model, a nonlinear model for slurry density detection is established through the minimum p-norm random configuration network; The high-voltage signal P of the slurry flow pipeline H , low voltage signal P L , slurry pump current I and slurry pump frequency f are used as inputs of the nonlinear model, and the output is the compensation value of the slurry density nonlinear model 5. The method for intelligent detection of slurry density based on minimum p-norm random configuration network according to claim 4 is characterized in that: The minimum p-norm random configuration network model construction includes: S31, X=[P H ,P L ,f,I] as the input of the random configuration network with the minimum p norm, and initialize the maximum number of hidden layer nodes L max , maximum number of configuration times T max , critical tolerance error ε, random parameter range [ - λ max ,λ max ] and the change step size Δλ; S32, assuming that the number of nodes in the current hidden layer is L-1, calculate the network output residual e L-1 , if L≤L max ,‖e L-1 ‖ 2 >∈, then the Lth hidden layer node is added, and the supervision mechanism for selecting the hidden layer node is given by the following formula: in, Given 1-∈<r<1, μ L =(1-r) / (L+1); S33, random configuration T max Each configuration will randomly select the input weight ω of the Lth hidden layer node in the supervision mechanism. L and bias b L , and calculate ξ L,q ; If minξ L,1 ,ξ L,2 ≥0, then ω L 、b L ,ξ L,q Store; if all L,q If none of them meet the conditions, a larger r value will be selected for reconfiguration; after the random configuration is completed, the largest The corresponding ω L 、b L As the input weight and bias of the Lth node; S34, after determining the model structure, calculate the output weight β: According to the definition of p-norm, the p-norm of a given vector x can be expressed as: The objective function is constructed by using the p-power of the error vector p-norm and combining it with the regularization term The objective function is obtained as: in, At this time, we get an implicit equation about β, and then use the fixed point iteration method to solve the implicit equation, let Then the core iterative process of fixed point iteration can be expressed as β(t)=f(β(t-1)), Among them, β(t) is the solution of the output weight obtained at the t-th iteration; S35, when L≤L max or ‖e L-1 ‖ 2 When >∈, stop adding nodes; finally, the output compensation value of the data-driven model is obtained, which is expressed as: In this way, the identification of the nonlinear part is completed and a nonlinear model for slurry density detection is established.

6. The method for intelligent detection of slurry density based on minimum p-norm random configuration network according to claim 5, characterized in that: Step S4: The estimated value of the slurry density linear model and the compensation value of the slurry density nonlinear model are added together to output the final slurry density prediction value.

Images