Intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment

Abstract

The present application relates to the technical field of fertilizer production equipment, and specifically discloses an intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment, which uses artificial intelligence technology based on deep learning to perform real-time monitoring and time series analysis on the stirring speed during the mixing reaction of needle-shaped fertilizer raw materials, so as to dig out the dynamic change pattern of the stirring speed in the global time domain, and perform feature correction on its global time series characteristics based on the local time series fluctuation characteristics of the stirring speed, so as to comprehensively utilize the multi-scale time series information of the stirring speed, and adaptively adjust the spray pressure of the adhesive during the stirring and mixing reaction of the raw materials. In this way, the intelligent control of the adhesive spray pressure can be achieved to adapt it to the dynamic changes of the stirring speed, thereby reducing the risk of human intervention and significantly improving the automation and intelligence of the needle-shaped fertilizer extrusion granulation process.

Description

Technical Field

[0001] The present application relates to the technical field of fertilizer production equipment, and more specifically, to an intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment. Background Art

[0002] In the field of modern agriculture and fertilizer manufacturing, needle-shaped fertilizer, as a highly efficient and slow-release fertilizer form, has received extensive attention and application due to its unique physical form and slow-release characteristics. In the production process of needle-shaped fertilizer, the traditional extrusion granulation method is usually dry granulation. After the materials are mixed, they are agglomerated and formed by mechanical pressure. The fertilizer produced in this way usually has problems such as easy pulverization and low product granule strength.

[0003] In order to improve the defects of dry granulation, in the existing agricultural fertilizer production technology, adhesives are sprayed during the mixing and stirring reaction of various materials to enhance the bonding force between materials and improve the mechanical strength of the particles. However, in this process, the control of spray pressure is a key factor affecting the granulation effect. Too high pressure may cause the material structure to be destroyed, while too low pressure may not form the ideal particle shape. Existing spray pressure control methods usually rely on manual experience to set fixed parameters. This method is not only inefficient, but also has large errors. It cannot adapt to changes in conditions during the production process in real time, which easily leads to unstable product quality.

[0004] Therefore, an intelligent and multifunctional needle-shaped fertilizer extrusion granulation equipment is expected. Summary of the invention

[0005] The present application provides an intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment, which can realize intelligent control of the adhesive spraying pressure, so that it can adapt to the dynamic changes of the stirring speed, thereby reducing the risk of human intervention and significantly improving the automation and intelligence of the needle-shaped fertilizer extrusion granulation process.

[0006] In the first aspect, an intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment is provided, including: a stirring speed monitoring module, used to obtain the time series of the stirring speed collected by a speed sensor; a stirring speed timing encoding module, used to perform time encoding on the time series of the stirring speed to obtain a sequence of local timing implicitly associated feature vectors of the stirring speed; a timing propagation aggregation module, used to perform feature timing propagation aggregation on the sequence of local timing implicitly associated feature vectors of the stirring speed to obtain a stirring speed timing aggregate representation vector; a timing fluctuation correction module, used to correct the stirring speed timing aggregate representation vector based on the timing fluctuation characteristics of the sequence of local timing implicitly associated feature vectors of the stirring speed to obtain a stirring speed timing aggregate fluctuation correction representation vector; a spray pressure decoding recommendation module, used to determine a recommended spray pressure decoding value based on the stirring speed timing aggregate fluctuation correction representation vector.

[0007] In a possible implementation, the stirring speed time series encoding module is used to: input the time series of the stirring speed into a sequence encoder based on a Bi-LSTM model to obtain a sequence of implicitly associated feature vectors of the local time series of the stirring speed.

[0008] In a possible implementation, the time series propagation aggregation module is used to: input the sequence of the stirring speed local time series implicitly associated feature vectors into a node feature propagation network based on a node energy decay mechanism to obtain the stirring speed time series aggregation representation vector.

[0009] In a possible implementation, the time series propagation aggregation module includes: an energy significant descriptor calculation unit, which is used to calculate the energy significant descriptors of each stirring speed local time series implicitly associated feature vector in the sequence of the stirring speed local time series implicitly associated feature vector based on the mean and variance of each stirring speed local time series implicitly associated feature vector to obtain a time series of energy significant descriptors; a timestamp marking unit, which is used to extract the timestamps of each stirring speed local time series implicitly associated feature vector in the sequence of the stirring speed local time series implicitly associated feature vector to obtain a time series of timestamps; and a historical node feature division unit, which is used to use the stirring speed local time series implicitly associated feature vector corresponding to the current time point in the sequence of the stirring speed local time series implicitly associated feature vector as the current stirring speed local time series implicitly associated feature vector, and use the stirring speed local time series implicitly associated feature vectors corresponding to other time points as the historical stirring speed local time series implicitly associated feature vectors to obtain the current time series of the stirring speed local time series implicitly associated feature vector. A sequence of previous stirring speed local time series implicitly associated feature vectors and historical stirring speed local time series implicitly associated feature vectors; a decay factor calculation unit, used to calculate the energy time decay factor of each historical stirring speed local time series implicitly associated feature vector in the sequence of historical stirring speed local time series implicitly associated feature vectors relative to the current stirring speed local time series implicitly associated feature vector to obtain a time series of energy time decay factors; a historical node propagation aggregation unit, used to use the time series of the energy significant descriptor as a positive adjustment factor and the time series of the energy time decay factor as a negative adjustment factor to perform time series propagation aggregation on the sequence of historical stirring speed local time series implicitly associated feature vectors to obtain a historical stirring speed local time series propagation aggregation representation feature vector; a global time series aggregation unit, used to fuse the historical stirring speed local time series propagation aggregation representation feature vector and the current stirring speed local time series implicitly associated feature vector to obtain the stirring speed time series aggregation representation vector.

[0010] In a possible implementation, the energy-significant descriptor calculation unit is used to: calculate the expected value of the fourth power of the difference between each eigenvalue in the local time series implicit association feature vector of the stirring speed and its eigenvalue mean, and divide the expected value by the square of the eigenvariance of the local time series implicit association feature vector of the stirring speed to obtain the energy-significant descriptor.

[0011] In a possible implementation, the attenuation factor calculation unit is used to: calculate the timestamp difference between the historical stirring speed local time series implicitly associated feature vector and the current stirring speed local time series implicitly associated feature vector; round down the timestamp difference, divide it by a preset time attenuation period, and then square it, and multiply the square result by a preset attenuation rate to obtain an initial energy attenuation coefficient; calculate an exponential function value with a natural constant as the base and the initial energy attenuation coefficient as the exponent to obtain the energy time attenuation factor.

[0012] In a possible implementation, the timing fluctuation correction module includes: a timing fluctuation coefficient calculation unit, which is used to calculate the stirring speed timing fluctuation coefficient between each two adjacent stirring speed local timing implicitly associated feature vectors in the sequence of the stirring speed local timing implicitly associated feature vectors to obtain the stirring speed timing fluctuation representation vector; a correction unit, which is used to use the stirring speed timing fluctuation representation vector as a weight vector to correct the stirring speed timing aggregation representation vector to obtain the stirring speed timing aggregation fluctuation correction representation vector.

[0013] In a possible implementation, the correction unit is used to: perform probabilistic processing on the stirring speed time series fluctuation representation vector based on the Sigmoid function to obtain a stirring speed time series fluctuation weight representation vector; calculate the point-by-point multiplication between the stirring speed time series aggregation representation vector and the stirring speed time series fluctuation weight representation vector to obtain the stirring speed time series aggregation fluctuation correction representation vector.

[0014] In a possible implementation, the spray pressure decoding recommendation module is used to: input the stirring speed time series aggregation fluctuation correction representation vector into a decoder-based spray pressure adaptive regulator to obtain the recommended spray pressure decoding value.

[0015] The present application provides an intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment, which uses artificial intelligence technology based on deep learning to perform real-time monitoring and time series analysis on the stirring speed during the mixing reaction of needle-shaped fertilizer raw materials, so as to dig out the dynamic change pattern of the stirring speed in the global time domain, and perform feature correction on its global time series characteristics based on the local time series fluctuation characteristics of the stirring speed, so as to comprehensively utilize the multi-scale time series information of the stirring speed, and adaptively adjust the spray pressure of the adhesive during the stirring and mixing reaction of the raw materials. In this way, the intelligent control of the adhesive spray pressure can be achieved, so that it can adapt to the dynamic changes of the stirring speed, thereby reducing the risk of human intervention and significantly improving the automation and intelligence of the needle-shaped fertilizer extrusion granulation process. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1This is a schematic block diagram of the intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment according to an embodiment of the present application.

[0017] Figure 2 This is a data flow diagram of the intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment according to an embodiment of the present application.

[0018] Figure 3 It is a schematic block diagram of a timing propagation aggregation module in the intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment according to an embodiment of the present application.

[0019] Figure 4 It is a schematic block diagram of a timing fluctuation correction module in the intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment according to an embodiment of the present application.

[0020] Figure 5 This is a schematic flow chart of a control method for the intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment according to an embodiment of the present application. DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0022] Needle-shaped fertilizers play an important role in modern agricultural production due to their excellent slow-release effect and efficient nutrient supply. However, their production process still faces challenges. Traditionally, needle-shaped fertilizers are produced through dry granulation technology. Although this method is simple, it often results in fragile fertilizer particles and insufficient strength. To solve this problem, modern processes introduce adhesives during the material mixing stage to enhance the bonding force of the particles. The key lies in the control of the spray pressure of the adhesive, which is crucial to the final form of the particles. Too high a pressure may destroy the material structure, while too low a pressure may not form ideal particles. At present, the regulation of spray pressure mostly depends on the experience of the operator. This method is not only inefficient, but also difficult to adapt to real-time changes in the production process due to its lack of flexibility, thus affecting the stability of product quality.

[0023] In view of the above technical problems, the technical concept of this application is to use artificial intelligence technology based on deep learning to monitor and analyze the stirring speed in the mixing reaction process of needle-shaped fertilizer raw materials in real time, so as to dig out the dynamic change pattern of the stirring speed in the global time domain, and perform feature correction on its global time characteristics based on the local time series fluctuation characteristics of the stirring speed, so as to comprehensively utilize the multi-scale time series information of the stirring speed and adaptively adjust the spray pressure of the adhesive during the stirring and mixing reaction of the raw materials. In this way, the intelligent control of the adhesive spray pressure can be achieved to adapt it to the dynamic changes of the stirring speed, thereby reducing the risk of human intervention and significantly improving the automation and intelligence of the needle-shaped fertilizer extrusion granulation process.

[0024] Figure 1 Schematic diagram of the intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment according to the embodiment of the present application. Figure 1 As shown, the intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment 100 includes: a stirring speed monitoring module 110, which is used to obtain the time series of the stirring speed collected by the speed sensor; a stirring speed timing encoding module 120, which is used to perform timing encoding on the time series of the stirring speed to obtain a sequence of local timing implicitly associated feature vectors of the stirring speed; a timing propagation aggregation module 130, which is used to perform feature timing propagation aggregation on the sequence of local timing implicitly associated feature vectors of the stirring speed to obtain a stirring speed timing aggregation representation vector; a timing fluctuation correction module 140, which is used to correct the stirring speed timing aggregation representation vector based on the timing fluctuation characteristics of the sequence of local timing implicitly associated feature vectors of the stirring speed to obtain a stirring speed timing aggregation fluctuation correction representation vector; a spray pressure decoding recommendation module 150, which is used to determine a recommended spray pressure decoding value based on the stirring speed timing aggregation fluctuation correction representation vector.

[0025] In the above-mentioned intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment, the stirring speed monitoring module 110 is used to obtain the time series of the stirring speed collected by the speed sensor. It should be understood that the spray pressure of the adhesive plays a decisive role in the penetration and distribution of the adhesive in the material. When the spray pressure is sufficient, it can ensure that the adhesive penetrates into the pores of the material, enhances the bonding force inside the particles, and thus improves the strength and durability of the particles. At the same time, in the granulation process, the stirring speed determines the uniformity of the material mixing, and then affects the contact area and bonding effect between the adhesive and the material. Therefore, in order to optimize the particle quality, in actual production, the stirring speed and the spray pressure of the adhesive need to be coordinated. Based on this, in the technical solution of the present application, the time series of the stirring speed collected by the speed sensor is first obtained, and according to the timing fluctuation characteristics of the stirring speed, the spray pressure is dynamically adjusted to optimize the bonding effect.

[0026] Optionally, in one embodiment of the present application, obtaining a time series of stirring speed collected by a speed sensor includes: installing a speed sensor at a key position of the stirring device, such as near a stirring shaft or a stirring paddle. Using a data acquisition system (DAS) to record the stirring speed detected by the sensor in real time. Filtering and digitizing the raw signal collected by the sensor to ensure the accuracy and availability of the data. Ensure that the data collection of all sensors is synchronized with a unified time reference to facilitate subsequent time series analysis. Store the collected data in a database or data recording system for subsequent access and analysis.

[0027] In the above-mentioned intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment, the stirring speed time series encoding module 120 is used to perform time series encoding on the time series of the stirring speed to obtain a sequence of implicitly associated feature vectors of the local time series of the stirring speed. It should be understood that since the stirring speed data has complex time series dependence and nonlinear dynamic characteristics, in order to effectively capture the time series variation pattern of the stirring speed, the present application adopts the Bi-LSTM model as a sequence encoder to perform time series modeling on the time series of the stirring speed. The Bi-LSTM model can effectively capture the before and after correlation of the stirring speed characteristics at each time point through the bidirectional propagation of the hidden state in the time dimension, and output each hidden layer in the Bi-LSTM model as the time series of implicitly associated feature vectors of the local time series of the stirring speed, so as to reveal the dynamic change of the stirring speed in the time dimension. That is, the stirring speed time series encoding module 120 is used to: input the time series of the stirring speed into the sequence encoder based on the Bi-LSTM model to obtain a sequence of implicitly associated feature vectors of the local time series of the stirring speed.

[0028] In the above-mentioned intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment, the time series propagation aggregation module 130 is used to perform feature time series propagation aggregation on the sequence of the local time series implicitly associated feature vectors of the stirring speed to obtain the stirring speed time series aggregation representation vector. It should be understood that in order to integrate the local time series features of the stirring speed at each time step to obtain the time series feature aggregation representation of the stirring speed in the global time domain, the time series of the local time series implicitly associated feature vectors of the stirring speed is further feature aggregated. At the same time, considering that in the feature aggregation process, the stirring speed features of the early time steps may gradually weaken the influence on the current stirring speed over time, therefore, the present application further introduces a node feature propagation network based on a node energy decay mechanism, based on the feature significance of the local time series implicitly associated feature vectors of the stirring speed at each time node and the time series decay characteristics of the information to aggregate features, thereby improving the expression ability of features. That is, the time series propagation aggregation module is used to: input the sequence of the local time series implicitly associated feature vectors of the stirring speed into the node feature propagation network based on the node energy decay mechanism to obtain the stirring speed time series aggregation representation vector. Specifically, the node feature propagation network first calculates the energy significant descriptor of each stirring speed local time series implicit association feature vector according to its mean and variance to quantify its feature importance. Subsequently, the stirring speed local time series implicit association feature vector of each time node is given time information, and the time series decay mechanism is used to generate the corresponding energy time decay factor based on the time span between the stirring speed local time series implicit association feature vector of each historical time node and the current (i.e., the end of the time series) stirring speed local time series implicit association feature vector, so as to perform feature weighted aggregation according to the feature significance and information time series decay characteristics of each time node, avoid excessive accumulation of early features, and generate a stirring speed time series aggregate representation vector.

[0029] Optionally, in one embodiment of the present application, Figure 3 Schematic block diagram of the timing propagation aggregation module in the intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment of the embodiment of the present application. Figure 3As shown, the time series propagation aggregation module 130 includes: an energy significant descriptor calculation unit 210, which is used to calculate the energy significant descriptors of each stirring speed local time series implicitly associated feature vector in the sequence of the stirring speed local time series implicitly associated feature vector based on the mean and variance of each stirring speed local time series implicitly associated feature vector in the sequence of the stirring speed local time series implicitly associated feature vector to obtain a time series of energy significant descriptors; a timestamp marking unit 220, which is used to extract the timestamps of each stirring speed local time series implicitly associated feature vector in the sequence of the stirring speed local time series implicitly associated feature vector to obtain a time series of timestamps; a historical node feature division unit 230, which is used to use the stirring speed local time series implicitly associated feature vector corresponding to the current time point in the sequence of the stirring speed local time series implicitly associated feature vector as the current stirring speed local time series implicitly associated feature vector, and use the stirring speed local time series implicitly associated feature vectors corresponding to other time points as the historical stirring speed local time series implicitly associated feature vectors to obtain the current stirring speed local time series implicitly associated feature vector. A sequence of local time series implicitly associated feature vectors of speed and local time series implicitly associated feature vectors of historical stirring speed; a decay factor calculation unit 240, used to calculate the energy time decay factor of each local time series implicitly associated feature vector of historical stirring speed in the sequence of local time series implicitly associated feature vectors of historical stirring speed relative to the local time series implicitly associated feature vector of current stirring speed to obtain a time series of energy time decay factors; a historical node propagation aggregation unit 250, used to perform time series propagation aggregation on the sequence of local time series implicitly associated feature vectors of historical stirring speed using the time series of the energy significant descriptor as a positive adjustment factor and the time series of the energy time decay factor as a negative adjustment factor to obtain a local time series propagation aggregation representation feature vector of historical stirring speed; a global time series aggregation unit 260, used to fuse the local time series propagation aggregation representation feature vector of historical stirring speed and the local time series implicitly associated feature vector of current stirring speed to obtain the time series aggregation representation vector of stirring speed.

[0030] Optionally, in a sub-embodiment of the present application, the energy-significant descriptor calculation unit is used to: calculate the expected value of the fourth power of the difference between each eigenvalue in the local time series implicit correlation feature vector of the stirring speed and its eigenvalue mean, and divide the expected value by the square of the eigenvariance of the local time series implicit correlation feature vector of the stirring speed to obtain the energy-significant descriptor.

[0031] Optionally, in a sub-embodiment of the present application, the attenuation factor calculation unit is used to: calculate the timestamp difference between the historical stirring speed local time series implicitly associated feature vector and the current stirring speed local time series implicitly associated feature vector; round down the timestamp difference, divide it by a preset time attenuation period and then square it, and multiply the square result by a preset attenuation rate to obtain an initial energy attenuation coefficient; calculate an exponential function value with a natural constant as the base and the initial energy attenuation coefficient as the exponent to obtain the energy time attenuation factor.

[0032] More specifically, in one embodiment of the present application, the time series propagation aggregation module 120 is used to process the sequence of the local time series implicitly associated feature vectors of the stirring speed using the following time series propagation fusion formula to obtain the stirring speed time series aggregation representation vector, wherein the time series propagation fusion formula is:

[0033] X={x1,x2,...x k-1 ,x k}

[0034]

[0035] Wherein, X represents the sequence of the implicitly associated feature vectors of the local time series of the stirring speed, the value of k is the number of feature vectors in the sequence of the implicitly associated feature vectors of the local time series of the stirring speed, and x1, x2, x i 、x k-1 and x k respectively represent the first, second, ith, k-1th and current local time series implicit correlation feature vectors of stirring speed in the sequence of the local time series implicit correlation feature vectors of stirring speed, represents the jth eigenvalue of the local time series implicitly associated eigenvector of the i-th stirring speed, the value of L is the length of the local time series implicitly associated eigenvector of the i-th stirring speed, μ i is the characteristic mean of the implicit correlation characteristic vector of the local time series of the i-th stirring speed, σ 2 i is the characteristic variance of the implicit correlation eigenvector of the local time series of the i-th stirring speed, represents the fourth-order central moment of the i-th stirring speed local time series implicit correlation eigenvector, that is, the expected value of the fourth power of the difference between each eigenvalue in the i-th stirring speed local time series implicit correlation eigenvector and its eigenmean, is the energy significant descriptor of the implicit correlation feature vector of the local time series of the i-th stirring speed, is the energy significant descriptor of the implicit correlation feature vector of the local time series of the current stirring speed, t kThe timestamp representing the local time series implicitly associated feature vector of the current stirring speed, t i The timestamp representing the implicitly associated feature vector of the local time series of the i-th stirring speed, represents rounding down, β represents the preset decay rate, L represents the preset time decay period, exp(·) represents the exponential function operation, α is the weight hyperparameter, x k+1 A vector representing the time series aggregation of the stirring speed.

[0036] In the above-mentioned intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment, the timing fluctuation correction module 140 is used to correct the stirring speed time series aggregation representation vector based on the timing fluctuation characteristics of the sequence of the local timing implicitly associated feature vector of the stirring speed to obtain the stirring speed time series aggregation fluctuation correction representation vector. It should be understood that considering that the above-mentioned feature aggregation process mainly focuses on the global timing information of the stirring speed, it may cause the sensitivity to the local fluctuation of the stirring speed to be reduced. Therefore, in order to further capture the local small fluctuations of the stirring speed, in the technical scheme of the present application, the stirring speed timing fluctuation coefficient between each two adjacent local timing implicitly associated feature vectors of the stirring speed in the sequence of the local timing implicitly associated feature vector of the stirring speed is calculated to obtain the stirring speed timing fluctuation representation vector composed of multiple stirring speed timing fluctuation coefficients. In this way, the amplitude of the change of the stirring speed between adjacent time steps is quantified and described, and the fluctuation information of the stirring speed in a short time is obtained, so as to enhance the sensitivity of the model to the local change of the stirring speed. In an embodiment of the present application, the difference measurement methods such as Euclidean distance and Manhattan distance can be used to calculate the stirring speed timing fluctuation coefficient.

[0037] Optionally, in one embodiment of the present application, Figure 4 Schematic block diagram of the timing fluctuation correction module in the intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment of the embodiment of the present application. Figure 4 As shown, the timing fluctuation correction module 140 includes: a timing fluctuation coefficient calculation unit 310, which is used to calculate the stirring speed timing fluctuation coefficient between each two adjacent stirring speed local timing implicitly associated feature vectors in the sequence of the stirring speed local timing implicitly associated feature vectors to obtain the stirring speed timing fluctuation representation vector; a correction unit 320, which is used to use the stirring speed timing fluctuation representation vector as a weight vector to correct the stirring speed timing aggregation representation vector to obtain the stirring speed timing aggregation fluctuation correction representation vector.

[0038] Then, in order to comprehensively consider the global time series aggregation characteristics and local time series fluctuation characteristics of the stirring speed, the present application further uses each stirring speed time series fluctuation coefficient in the stirring speed time series fluctuation representation vector as a correction factor to perform weighted correction on the stirring speed time series aggregation representation vector. Specifically, first, in order to reduce the influence of extreme values ​​in the stirring speed time series aggregation representation vector on the correction result, the stirring speed time series fluctuation representation vector is further probabilistically processed based on the Sigmoid function. That is, each value in the stirring speed time series aggregation representation vector is mapped to the [0,1] interval through the Sigmoid function to obtain a smooth weight distribution, and generate a stirring speed time series fluctuation weight representation vector.

[0039] Then, the stirring speed time series fluctuation weight representation vector and the stirring speed time series aggregation representation vector are fused element by element by position point multiplication to achieve weighted adjustment of the stirring speed time series aggregation representation vector and obtain the stirring speed time series aggregation fluctuation correction representation vector. In this way, it can ensure that while capturing the global change trend of the stirring speed, its local fluctuation is fully considered, so as to more accurately reflect the dynamic change of the stirring speed and provide a more accurate basis for the subsequent adaptive adjustment of the spray pressure.

[0040] Optionally, in one embodiment of the present application, the correction unit 320 is used to: perform probabilistic processing on the stirring speed time series fluctuation representation vector based on the Sigmoid function to obtain a stirring speed time series fluctuation weight representation vector; calculate the point-by-point multiplication between the stirring speed time series aggregation representation vector and the stirring speed time series fluctuation weight representation vector to obtain the stirring speed time series aggregation fluctuation correction representation vector.

[0041] In the above-mentioned intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment, the spray pressure decoding recommendation module 150 is used to determine the recommended spray pressure decoding value based on the stirring speed time series aggregation fluctuation correction representation vector. It should be understood that in the technical solution of the present application, the stirring speed time series aggregation fluctuation correction representation vector is input into the spray pressure adaptive regulator based on the decoder to obtain the recommended spray pressure decoding value. That is, the spray pressure decoding recommendation module is used to: input the stirring speed time series aggregation fluctuation correction representation vector into the spray pressure adaptive regulator based on the decoder to obtain the recommended spray pressure decoding value. In the technical solution of the present application, the decoder can dynamically predict the recommended spray pressure decoding value under the current stirring speed conditions according to the characteristic information in the stirring speed time series aggregation fluctuation correction representation vector by learning the nonlinear collaborative control relationship between the stirring speed and the spray pressure, thereby achieving instant response to changes in the stirring speed and reasonable adjustment of the spray pressure.

[0042] Optionally, in one embodiment of the present application, the decoder is used to decode and regress the stirring speed time series aggregate fluctuation correction representation vector using the following formula to obtain a recommended spray pressure decoding value, wherein the formula is: Wherein, X is the stirring speed time series aggregate fluctuation correction representation vector, Y is the recommended spray pressure decoding value, W is the weight matrix, Represents matrix multiplication.

[0043] In the technical solution of the present application, the stirring speed time series fluctuation representation vector is obtained by calculating the stirring speed time series fluctuation coefficient between each adjacent two stirring speed local time series implicitly associated feature vectors in the sequence of the stirring speed local time series implicitly associated feature vectors. That is, at the feature receptive field level, the receptive field of the eigenvalues ​​at each position in the stirring speed time series fluctuation representation vector is at the local time domain level. The stirring speed time series aggregation representation vector is obtained by inputting the sequence of the stirring speed local time series implicitly associated feature vectors into a node feature propagation network based on the node energy decay mechanism. Therefore, there is a feature receptive field scale correspondence deviation between the stirring speed time series aggregation representation vector and the stirring speed time series fluctuation weight representation vector at the feature value fine-grained level, which will cause the obtained stirring speed time series aggregation fluctuation correction representation vector to cause an outlier decoding regression inference mapping deviation due to the feature receptive field correspondence deviation, thereby affecting the accuracy of the recommended spray pressure decoding value obtained by the stirring speed time series aggregation fluctuation correction representation vector through the decoder-based spray pressure adaptive regulator.

[0044] Therefore, in a preferred example, the stirring speed time series aggregate fluctuation correction representation vector is passed through a decoder-based spray pressure adaptive regulator to obtain a recommended spray pressure decoding value, which includes:

[0045] Calculating the characteristic mean of the stirring speed time series aggregate fluctuation correction representation vector, and dividing the characteristic mean by the difference between the maximum eigenvalue and the minimum eigenvalue of the stirring speed time series aggregate fluctuation correction representation vector to obtain the stirring speed time series aggregate fluctuation correction representation distribution characterization value;

[0046] Subtract the stirring speed time series aggregate fluctuation correction representation distribution characterization value from one and divide it by the stirring speed time series aggregate fluctuation correction representation distribution characterization value to obtain the stirring speed time series aggregate fluctuation correction representation distribution modulation value; activate the stirring speed time series aggregate fluctuation correction representation vector through a probabilistic function to obtain a probabilistic stirring speed time series aggregate fluctuation correction representation vector;

[0047] Subtract the probabilistic stirring speed time series aggregate fluctuation correction representation vector from the stirring speed time series aggregate fluctuation correction representation distribution modulation value, take the absolute value and calculate the negative of the logarithmic value with base 2 to obtain the probabilistic stirring speed time series aggregate fluctuation correction representation distribution modulation information vector;

[0048] After dividing the stirring speed time series aggregate fluctuation correction representation distribution characterization value by one minus the difference of each eigenvalue of the probabilistic stirring speed time series aggregate fluctuation correction representation vector, summing all eigenvalues ​​of the probabilistic stirring speed time series aggregate fluctuation correction representation vector and dividing by the length of the stirring speed time series aggregate fluctuation correction representation vector to obtain the probabilistic stirring speed time series aggregate fluctuation correction representation distribution modulation bias value;

[0049] The optimized stirring speed time series aggregate fluctuation correction representation vector is obtained by multiplying the probabilistic stirring speed time series aggregate fluctuation correction representation distribution modulation information vector with the probabilistic stirring speed time series aggregate fluctuation correction representation distribution modulation bias value and the weight as a hyperparameter;

[0050] The optimized stirring speed time series aggregation fluctuation correction representation vector is passed through a decoder-based spray pressure adaptive regulator to obtain a recommended spray pressure decoding value.

[0051] The optimized stirring speed time series polymerization fluctuation correction representation vector is expressed as:

[0052]

[0053] And among them:

[0054]

[0055] in, represents the characteristic mean of the agitation speed time series aggregate fluctuation correction representation vector, v max and v min represents the maximum eigenvalue and the minimum eigenvalue in the stirring speed time series aggregate fluctuation correction representation vector respectively, p represents the stirring speed time series aggregate fluctuation correction representation distribution characterization value, V represents the probabilistic stirring speed time series aggregate fluctuation correction representation vector obtained after the stirring speed time series aggregate fluctuation correction representation vector is activated by the probabilistic function, v i represents the i-th eigenvalue of the probabilistic stirring speed time series aggregate fluctuation correction representation vector, ε is the weight as a hyperparameter, the value of L is the length of the stirring speed time series aggregate fluctuation correction representation vector, and V′ represents the optimized stirring speed time series aggregate fluctuation correction representation vector.

[0056] That is, the eigenvalue-based probability information distribution planning of the stirring speed time series aggregate fluctuation correction representation vector is carried out through the Bernoulli probability modulation distribution of the stirring speed time series aggregate fluctuation correction representation vector relative to the eigenvalue distribution, and the probability reverse mapping of the probability characteristics of the stirring speed time series aggregate fluctuation correction representation vector is used as the extended coverage of the set mapping space of the stirring speed time series aggregate fluctuation correction representation vector to independently understand the interaction path between the intuitive probability information distribution and the abstract probability space mapping of the stirring speed time series aggregate fluctuation correction representation vector, so as to improve the accuracy of the recommended spray pressure decoding value obtained by the optimized stirring speed time series aggregate fluctuation correction representation vector through the decoder-based spray pressure adaptive regulator by avoiding the counterfactual reasoning mapping of the outlier feature distribution of the stirring speed time series aggregate fluctuation correction representation vector to the class probability.

[0057] It is worth mentioning that in other specific examples of the present application, the recommended spray pressure decoding value can also be determined by other means, for example: installing a speed sensor on the stirring equipment to monitor the stirring speed in real time. Record the data of the stirring speed changing over time to form a time series. Remove outliers and noise in the data. Apply methods such as Fourier transform or wavelet transform to analyze the periodicity and trend of the stirring speed. Extract key features from the time series, such as velocity peak, fluctuation amplitude, energy distribution, etc. Select features that are significantly correlated with the spray pressure. Select a machine learning model suitable for time series prediction, such as a long short-term memory network (LSTM) or a convolutional neural network (CNN). Use historical data to train the model so that it can predict the spray pressure based on the characteristics of the stirring speed. Develop an algorithm to correct the fluctuation of the stirring speed to reduce the prediction error. Continuously optimize the model parameters through methods such as cross-validation. : Input the stirring speed monitored in real time into the model to predict the current recommended spray pressure. Integrate the model prediction results into the control system to achieve automatic adjustment of the spray pressure. Collect feedback data based on the actual effect in the production process. Adjust the model based on the feedback data to improve the prediction accuracy.

[0058] In summary, the intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment according to the embodiment of the present application is explained, which uses artificial intelligence technology based on deep learning to perform real-time monitoring and time series analysis on the stirring speed during the mixing reaction of the needle-shaped fertilizer raw materials, so as to dig out the dynamic change pattern of the stirring speed in the global time domain, and perform feature correction on its global time series characteristics based on the local time series fluctuation characteristics of the stirring speed, so as to comprehensively utilize the multi-scale time series information of the stirring speed, and adaptively adjust the spray pressure of the adhesive during the stirring and mixing reaction of the raw materials. In this way, the intelligent control of the adhesive spray pressure can be achieved to adapt it to the dynamic changes of the stirring speed, thereby reducing the risk of human intervention and significantly improving the automation and intelligence of the needle-shaped fertilizer extrusion granulation process.

[0059] Figure 5 Schematic flow chart of the control method of the intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment of the embodiment of the present application. Figure 5 As shown, the control method of the intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment includes: S110, acquiring the time series of the stirring speed collected by the speed sensor; S120, performing time series encoding on the time series of the stirring speed to obtain a sequence of local time series implicitly associated feature vectors of the stirring speed; S130, performing feature time series propagation aggregation on the sequence of local time series implicitly associated feature vectors of the stirring speed to obtain a stirring speed time series aggregate representation vector; S140, based on the time series fluctuation characteristics of the sequence of local time series implicitly associated feature vectors of the stirring speed, correcting the stirring speed time series aggregate representation vector to obtain a stirring speed time series aggregate fluctuation correction representation vector; S150, determining a recommended spray pressure decoding value based on the stirring speed time series aggregate fluctuation correction representation vector.

[0060] Here, those skilled in the art will appreciate that the specific operations of each step in the control method of the above-mentioned intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment have been described in the above reference. Figures 1 to 4 The intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment has been introduced in detail, and therefore, its repeated description will be omitted.

[0061] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0062] In the above embodiments, the description of each embodiment has its own emphasis. For parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the module division is only a logical function division, and there may be other division methods in actual implementation. The modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0063] In addition, each functional module 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. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of hardware plus software functional modules.

[0064] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0065] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units stated in the device claim can also be implemented by one unit through software or hardware.

[0066] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. An intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment, characterized in that: include: A stirring speed monitoring module, used to obtain the time series of the stirring speed collected by the speed sensor; A stirring speed time series encoding module, used for performing time series encoding on the time series of the stirring speed to obtain a sequence of implicitly associated feature vectors of the local time series of the stirring speed; A time series propagation aggregation module is used to perform feature time series propagation aggregation on the sequence of implicitly associated feature vectors of the stirring speed local time series to obtain a stirring speed time series aggregation representation vector; A timing fluctuation correction module, used for correcting the stirring speed timing aggregate representation vector based on the timing fluctuation characteristics of the sequence of implicitly associated feature vectors of the stirring speed local timing to obtain a stirring speed timing aggregate fluctuation correction representation vector; A spray pressure decoding recommendation module, for determining a recommended spray pressure decoding value based on the agitation speed time series aggregate fluctuation correction representation vector; The timing propagation aggregation module comprises: An energy significant descriptor calculation unit is used to calculate the energy significant descriptors of each stirring speed local time series implicit association feature vector in the sequence of stirring speed local time series implicit association feature vectors based on the mean and variance of each stirring speed local time series implicit association feature vector to obtain a time series of energy significant descriptors; A timestamp marking unit, used for extracting the timestamp of each stirring speed local time series implicitly associated feature vector in the sequence of stirring speed local time series implicitly associated feature vectors to obtain a time series of timestamps; A historical node feature division unit, used for taking the local time series implicitly associated feature vector of stirring speed corresponding to the current time point in the sequence of the local time series implicitly associated feature vector of stirring speed as the current local time series implicitly associated feature vector of stirring speed, and taking the local time series implicitly associated feature vector of stirring speed corresponding to other time points as the historical local time series implicitly associated feature vector of stirring speed, so as to obtain a sequence of the current local time series implicitly associated feature vector of stirring speed and the historical local time series implicitly associated feature vector of stirring speed; The attenuation factor calculation unit is used to calculate the energy time attenuation factor of each historical stirring speed local time series implicitly associated feature vector in the sequence of the historical stirring speed local time series implicitly associated feature vector relative to the current stirring speed local time series implicitly associated feature vector to obtain a time series of energy time attenuation factors.

2. The intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment according to claim 1 is characterized in that: The stirring speed timing coding module is used for: The time series of the stirring speed is input into a sequence encoder based on a Bi-LSTM model to obtain a sequence of implicitly associated feature vectors of the local time series of the stirring speed.

3. The intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment according to claim 2 is characterized in that: The timing propagation aggregation module is used to: The sequence of the local time series implicitly associated feature vectors of the stirring speed is input into a node feature propagation network based on a node energy decay mechanism to obtain the stirring speed time series aggregate representation vector.

4. The intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment according to claim 3 is characterized in that: The timing propagation aggregation module also includes: A historical node propagation aggregation unit is used to perform time series propagation aggregation on the sequence of the local time series implicitly associated feature vectors of the historical stirring speed using the time series of the energy significant descriptor as a positive adjustment factor and the time series of the energy time decay factor as a negative adjustment factor to obtain a local time series propagation aggregation representation feature vector of the historical stirring speed; The global time series aggregation unit is used to fuse the historical stirring speed local time series propagation aggregation representation feature vector and the current stirring speed local time series implicit association feature vector to obtain the stirring speed time series aggregation representation vector.

5. The intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment according to claim 4 is characterized in that: The energy significant descriptor calculation unit is used to: The expected value of the fourth power of the difference between each eigenvalue in the stirring speed local time series implicit association eigenvector and its eigenmean is calculated, and the expected value is divided by the square of the eigenvariance of the stirring speed local time series implicit association eigenvector to obtain the energy significant descriptor.

6. The intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment according to claim 5 is characterized in that: The attenuation factor calculation unit is used to: Calculating the timestamp difference between the local time series implicit correlation feature vector of the historical stirring speed and the local time series implicit correlation feature vector of the current stirring speed; The timestamp difference is rounded down and divided by a preset time decay period and then squared, and the square result is multiplied by a preset decay rate to obtain an initial energy decay coefficient; An exponential function value with a natural constant as the base and the initial energy decay coefficient as the exponent is calculated to obtain the energy time decay factor.

7. The intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment according to claim 6 is characterized in that: The timing fluctuation correction module comprises: A time series fluctuation coefficient calculation unit, used for calculating the stirring speed time series fluctuation coefficient between each two adjacent stirring speed local time series implicitly associated characteristic vectors in the sequence of the stirring speed local time series implicitly associated characteristic vectors to obtain a stirring speed time series fluctuation representation vector; The correction unit is used to correct the stirring speed time series aggregation representation vector by using the stirring speed time series fluctuation representation vector as a weight vector to obtain the stirring speed time series aggregation fluctuation correction representation vector.

8. The intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment according to claim 7 is characterized in that: The correction unit is used for: The stirring speed time series fluctuation representation vector is subjected to probabilistic processing based on a Sigmoid function to obtain a stirring speed time series fluctuation weight representation vector; The stirring speed time series aggregated fluctuation correction representation vector is obtained by calculating the point-by-point multiplication between the stirring speed time series aggregated representation vector and the stirring speed time series fluctuation weight representation vector.

9. The intelligent multifunctional needle-shaped fertilizer extrusion granulation equipment according to claim 8 is characterized in that: The spray pressure decoding recommendation module is used to: The stirring speed time series aggregate fluctuation correction representation vector is input into the decoder-based spray pressure adaptive regulator to obtain the recommended spray pressure decoding value.

Images