Water bloom monitoring data interpolation method and device based on second-order optimization hidden feature analysis

Abstract

The invention discloses a water bloom monitoring data interpolation method and device based on second-order optimization hidden feature analysis. The method specifically comprises the following steps: S1, collecting historical water bloom monitoring data from a server and storing the historical water bloom monitoring data; s2, constructing a water bloom monitoring data matrix V according to the stored historical water bloom monitoring data; and S3, constructing a target loss function according to historical water bloom monitoring data and the water bloom monitoring data matrix V, and outputting a water bloom monitoring data interpolation value with the highest precision. The method is specially used for water bloom monitoring data, and water bloom monitoring data interpolation which accords with a statistical rule and is high in accuracy can be carried out, so that interpolation of missing water bloom monitoring data is realized. The method can be widely applied to the fields of algal bloom monitoring, environment monitoring and the like.

Description

Technical Field

[0001] The invention relates to the technical field of environmental protection, and in particular to a method and a device for interpolating water bloom monitoring data based on second-order optimization latent feature analysis. Background Art

[0002] With the continuous changes in the water environment and the increasingly serious global climate problems, algal blooms have occurred frequently in various places, and the importance of algal bloom monitoring has become increasingly prominent. As an important indicator for measuring the health of water bodies and environmental changes, algal bloom monitoring data can reflect the degree of water pollution and the health of the aquatic ecosystem in the monitoring area. Environmental management agencies and scientific researchers can adjust water quality management plans and governance measures based on algal bloom monitoring data. However, due to the lack of monitoring data due to multi-source sensor failures or communication reasons, how to accurately analyze and interpolate algal bloom monitoring data has become a key challenge in water protection and environmental management.

[0003] At present, a feasible method for the problem of missing data interpolation in water bloom monitoring is to accurately interpolate missing data in water bloom monitoring through a representation learning method based on latent feature analysis. This method uses known historical water bloom monitoring data to infer missing water bloom monitoring data. When performing latent feature analysis, it is necessary to construct an incomplete water bloom monitoring data matrix between multi-source sensor monitoring data and time series, and based on the known data in the matrix, train two low-dimensional water bloom monitoring data latent feature embedding representation matrices, so that the product of the two low-dimensional latent feature embedding matrices is as close as possible to the known water bloom monitoring data matrix. However, due to the need to accurately fill in the incomplete water bloom monitoring data matrix, and the optimization objective of the training latent feature analysis representation learning method has a non-convex bilinear property, the conventional first-order optimization based latent feature analysis method only considers the first-order gradient information, and it is difficult to accurately interpolate the incomplete water bloom monitoring data matrix. Therefore, how to more accurately interpolate the incomplete water bloom monitoring data matrix has become a key problem that needs to be solved urgently. Summary of the invention

[0004] In view of the technical problem that water bloom monitoring data in the prior art is prone to missing, the present invention proposes a water bloom monitoring data interpolation method and device based on second-order optimization latent feature analysis.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] The algae bloom monitoring data interpolation method based on second-order optimization latent feature analysis specifically includes the following steps:

[0007] S1: Collect historical algal bloom monitoring data from the server and store it;

[0008] S2: construct an algae bloom monitoring data matrix V based on the stored historical algae bloom monitoring data;

[0009] S3: Construct a target loss function based on historical algal bloom monitoring data and the algal bloom monitoring data matrix V, and output the algal bloom monitoring data interpolation value with the highest accuracy.

[0010] Preferably, in S1, the historical algal bloom monitoring data is represented in a triplet form as Y=(t, s, v), where t represents the time of data collection, s represents the indicators monitored by various sensors, and v represents the values ​​of the indicators of type s monitored by various algal bloom monitoring sensors at time t.

[0011] Preferably, in S2, the construction method of the algal bloom monitoring data matrix V is:

[0012] All triples Y = (t, s, v) are constructed into an I × J algae bloom monitoring data matrix V, where I is the number of data collection times, J is the number of various sensor monitoring indicators, and the v in the matrix ij represents the specific value of s type indicator monitored by various types of algal bloom monitoring sensors at time t, 1≤i≤I, 1≤j≤J, and Γ is the set of known elements contained in the algal bloom monitoring data matrix V.

[0013] Preferably, S3 includes:

[0014] S3-1: Initialize the process parameters involved in the interpolation of algal bloom monitoring data;

[0015] S3-2: construct the target loss function ε;

[0016] S3-3: Iterate the training of the constructed target loss function ε;

[0017] S3-4: Determine whether the loss function ε converges during the training iteration on Γ. If so, terminate the iteration; otherwise, continue the iteration.

[0018] S3-5: Output the interpolated value of the algal bloom monitoring data and store it.

[0019] Preferably, in S3-1, the process parameters include latent feature matrices T and S; the first-order gradient term g ε Including g ε (t i ) and g ε (s j ), corresponding to the first-order gradient components of the two latent feature matrices T and S respectively; the update increment vector Δε includes Δε(t i ) and Δε(s j ), corresponding to the updated incremental component vector of each row of the two latent feature matrices T and S respectively; the dimension of the latent feature matrix D; the convergence termination threshold δ; and the damping term coefficient γ.

[0020] Preferably, in S3-2, the target loss function ε is:

[0021]

[0022] In formula (1), Γ is the set of known elements contained in the algal bloom monitoring data matrix V; v ij represents the monitoring value of the jth monitoring indicator of each type of algae bloom monitoring sensor in the algae bloom monitoring data matrix V at the i-th time; t i represents the i-th row vector in the temporal latent feature matrix T; s j It represents the j-th row vector in the latent feature matrix S of various sensor monitoring indicators; T represents the transposed matrix.

[0023] Preferably, in S3-3, the iteration formula is:

[0024]

[0025] In formula (2), It means that ε has a certain effect on t in τ-1 training iterations. i The first-order negative gradient information; t i Represents the i-th row vector in the temporal latent feature matrix T; represents the training iteration round s of ε in τ-1 j The first-order negative gradient information; s j represents the jth row vector in the latent feature matrix S of various sensor monitoring indicators; g ε Indicates that all and The first-order negative gradient information composed of the component vectors; Δ ε Indicates that g ε Corresponding to all and The component vectors are combined into an update increment vector; CGD represents the conjugate gradient second-order solver; γ is the damping term coefficient; ρ represents the convergence accuracy of the conjugate gradient second-order solver; M represents the maximum iteration round of the conjugate gradient second-order solver; Representation and Correspondingly, it is used to update the latent feature vector t i The update increment of Representation and Correspondingly, used to update the feature vector s j The update increment of t i τ Represents the value of the τth training iteration round of the i-th row vector in the temporal latent feature matrix T; Represents the value of the τ-1th training iteration round of the i-th row vector in the temporal latent feature matrix T; Represents the value of the j-th row vector in the latent feature matrix S of various sensor monitoring indicators in the τ-th training iteration round; It represents the value of the j-th row vector in the latent feature matrix S of various sensor monitoring indicators in the τ-1th training iteration round.

[0026] Preferably, in S3-5, the calculation formula for the interpolation value of the algal bloom monitoring data is:

[0027]

[0028] In formula (3), represents the interpolation value of algal bloom monitoring data; t i represents the i-th row vector in the temporal latent feature matrix T; s j Represents the j-th row vector in the latent feature matrix S of various sensor monitoring indicators.

[0029] The present invention also provides a water bloom monitoring data interpolation device based on second-order optimization latent feature analysis, comprising:

[0030] A data acquisition module, used for receiving historical algal bloom monitoring data from a server;

[0031] A storage module, used for storing received historical water bloom monitoring data and water bloom monitoring data interpolation values;

[0032] A matrix construction module, used to construct an algae bloom monitoring data matrix V according to the stored historical algae bloom monitoring data;

[0033] Hidden feature extraction module, used to construct target loss function based on historical algal bloom monitoring data and algal bloom monitoring data matrix, and extract hidden features;

[0034] The output module is used to output the interpolation value of the algal bloom monitoring data according to the extracted latent features.

[0035] Preferably, the latent feature extraction module includes an initialization unit, a target loss function construction unit, and a training unit;

[0036] An initialization unit, used to initialize process parameters involved in the interpolation process of algal bloom monitoring data;

[0037] A target loss function construction unit is used to construct a target loss function according to historical algal bloom monitoring data, algal bloom monitoring data matrix, and process parameters;

[0038] The training unit is used to train and optimize the target loss function and extract latent features.

[0039] In summary, due to the adoption of the above technical solution, compared with the prior art, the present invention has at least the following beneficial effects:

[0040] The present invention provides a water bloom monitoring data interpolation method and device based on second-order optimization latent feature analysis, which is specifically used for water bloom monitoring data and can interpolate water bloom monitoring data with high accuracy in accordance with statistical laws to achieve interpolation of missing water bloom monitoring data. It can be widely used in the fields of water bloom monitoring, environmental monitoring, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 Schematic diagram of a water bloom monitoring data interpolation method based on second-order optimized latent feature analysis according to an exemplary embodiment of the present invention.

[0042] Figure 2 Schematic diagram of a water bloom monitoring data interpolation device based on second-order optimized latent feature analysis according to an exemplary embodiment of the present invention.

[0043] Figure 3 FIG. 4 is a schematic diagram of a storage module according to an exemplary embodiment of the present invention.

[0044] Figure 4 Schematic diagram of a latent feature extraction module according to an exemplary embodiment of the present invention. DETAILED DESCRIPTION

[0045] The present invention is further described in detail below in conjunction with the examples and specific implementation methods. However, this should not be understood as the scope of the above subject matter of the present invention being limited to the following examples, and all technologies realized based on the content of the present invention belong to the scope of the present invention.

[0046] In the description of the present invention, it is necessary to understand that the terms "longitudinal", "lateral", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0047] like Figure 1 As shown, the present invention provides a water bloom monitoring data interpolation method based on second-order optimization latent feature analysis, which specifically includes the following steps:

[0048] S1: Receive the instruction of interpolating water bloom monitoring data sent by the server, collect historical water bloom monitoring data from the server, and store it in the form of triples.

[0049] The server may send instructions to the water bloom monitoring data receiving module by sending instructions to interpolate water bloom monitoring data to the water bloom monitoring data receiving module regularly, according to the notification of the device, or according to the notification of a server.

[0050] In this embodiment, the triple representation of historical algal bloom monitoring data is Y=(t, s, v), where t represents the time of data collection, s represents the specific indicators monitored by various sensors, including water temperature, water transparency, pH value, total phosphorus (TP) and total nitrogen (TN), etc.; v represents the specific value of s type indicator monitored by various algal bloom monitoring sensors at time t.

[0051] S2: construct an algae bloom monitoring data matrix V based on the stored historical algae bloom monitoring data.

[0052] In this embodiment, all triples Y = (t, s, v) are constructed into an I × J algae bloom monitoring data matrix V, where I is the number of data collection times, J is the number of various sensor monitoring indicators, and the v in the matrix ij represents the specific value of the s type indicator monitored by various algae bloom monitoring sensors at time t, 1≤i≤I, 1≤j≤J, and Γ is the set of known elements contained in the algae bloom monitoring data matrix V. Since various sensors cannot synchronously collect all indicators at all time points, and data loss may occur during data transmission, the constructed algae bloom monitoring data matrix V is a sparse incomplete matrix.

[0053] S3: Construct a target loss function based on historical algal bloom monitoring data and the algal bloom monitoring data matrix V, and output the algal bloom monitoring data interpolation value with the highest accuracy.

[0054] S3-1: Initialize the process parameters involved in the interpolation of algal bloom monitoring data.

[0055] In this embodiment, the process parameters include latent feature matrices T and S; the first-order gradient term g ε Including g ε (t i 0 and g ε (s j ), corresponding to the first-order gradient components of the two latent feature matrices T and S respectively; update the increment vector Δε, including Δε(t i ) and Δε(s j ), corresponding to the update increment component vector of each row of the two latent feature matrices T and S; the dimension of the latent feature matrix D; the convergence termination threshold δ; the damping term coefficient γ;

[0056] in,

[0057] The dimension D of the latent feature matrix determines the dimension of the latent feature space of the latent feature matrices T and S and is initialized to a positive integer;

[0058] The sizes of the latent feature matrices T and S are determined by the dimension values ​​of each order of the corresponding algae bloom monitoring data matrix V, that is, T is an I-row D-column latent feature matrix, and S is an J-row D-column latent feature matrix. The two latent feature matrices are initialized with small random positive numbers respectively.

[0059] The convergence termination threshold δ is a parameter used to determine whether the iterative process has converged, and is initialized with a very small positive number;

[0060] The damping term coefficient γ is a constant used to control the positive definiteness of the second-order curvature matrix and balance the first-order and second-order approximation degrees, and is initialized to a large positive number.

[0061] S3-2: For the known data set Γ of the algae bloom monitoring data matrix V, construct the target loss function ε to train and obtain the values ​​of the latent feature matrices T and S of the historical algae bloom monitoring data; the target loss function ε is expressed by the following formula:

[0062]

[0063] In formula (1), Γ is the set of known elements contained in the algal bloom monitoring data matrix V; v ij represents the monitoring value of the jth monitoring indicator of each type of algae bloom monitoring sensor in the algae bloom monitoring data matrix V at the i-th time; t i represents the i-th row vector in the temporal latent feature matrix T; s j It represents the j-th row vector in the latent feature matrix S of various sensor monitoring indicators; T represents the transposed matrix.

[0064] The objective loss function uses Euclidean distance as the optimization target.

[0065] S3-3: Calculate the first-order gradient information g of the optimization target loss function ε hidden feature matrix in the current training iteration during the training process ε ; Then, during the training process, the optimization objective loss function ε is calculated, the second-order curvature information of the latent feature matrix in the current training iteration round is calculated, and the first-order gradient information g obtained by the first-order optimization unit is ε Perform nonlinear scaling to obtain a more accurate update Δε of the optimization objective;

[0066] According to the loss function and related parameters involved in the interpolation process of the initialized algae bloom monitoring data, the values ​​of the latent feature matrices T and S of the historical algae bloom monitoring data are obtained through training;

[0067] In order to minimize the optimization target loss function ε, so that the value of the optimization target loss function ε can be as close to the second-order stable point as possible, the training iteration formula is as follows:

[0068]

[0069] In formula (2), It means that ε has a certain effect on t in τ-1 training iterations. i The first-order negative gradient information; t i Represents the i-th row vector in the temporal latent feature matrix T; represents the training iteration round s of ε in τ-1 j The first-order negative gradient information; s j represents the jth row vector in the latent feature matrix S of various sensor monitoring indicators; g ε Indicates that all and The first-order negative gradient information composed of the component vectors; Δ ε Indicates that g ε Corresponding to all and The component vectors are combined into an update increment vector; CGD represents the conjugate gradient second-order solver; γ is the damping term coefficient; ρ represents the convergence accuracy of the conjugate gradient second-order solver; M represents the maximum iteration round of the conjugate gradient second-order solver; Representation and Correspondingly, it is used to update the latent feature vector t i The update increment of Representation and Correspondingly, used to update the feature vector s j The update increment of Represents the value of the τth training iteration round of the i-th row vector in the temporal latent feature matrix T; Represents the value of the τ-1th training iteration round of the i-th row vector in the temporal latent feature matrix T; Represents the value of the j-th row vector in the latent feature matrix S of various sensor monitoring indicators in the τ-th training iteration round; It represents the value of the j-th row vector in the latent feature matrix S of various sensor monitoring indicators in the τ-1th training iteration round.

[0070] S3-4: Determine whether the loss function ε converges in the training iteration process on Γ. If so, terminate the iteration; otherwise, continue the iteration.

[0071] The convergence judgment condition is that the absolute value of the difference between the calculated value of the target loss function ε and the value of the target loss function ε in the current round of 10 rounds of training and the value of the target loss function ε in the previous round is less than the threshold δ.

[0072] S3-5: Output the interpolated value of the algal bloom monitoring data and store it.

[0073] In this embodiment, when the target loss function ε converges on the known data set Γ, the latent feature matrices T and S obtained by training when the target loss function ε reaches the minimum value are used to calculate the data collection time t with the highest accuracy.i Algal bloom monitoring sensor data indicators j Interpolation value of algae bloom monitoring data The calculated interpolation value of water bloom monitoring data is output to the water bloom monitoring data interpolation value storage unit for storage. The calculation formula is:

[0074]

[0075] In formula (3), represents the interpolation value of algal bloom monitoring data; t i represents the i-th row vector in the temporal latent feature matrix T; s j Represents the j-th row vector in the latent feature matrix S of various sensor monitoring indicators.

[0076] Based on the above-mentioned water bloom monitoring data interpolation method based on second-order optimization latent feature analysis, e.g. Figure 2 As shown, the present invention also provides a water bloom monitoring data interpolation device based on second-order optimization latent feature analysis, including a data acquisition module 1, a storage module 2, a matrix construction module 3, a latent feature extraction module 4 and an output module 5.

[0077] The output end of the data acquisition module 1 is connected to the first input end of the storage module 2, and the output ends of the storage module 2 are respectively connected to the input ends of the matrix construction module 3; the output end of the matrix construction module 3 is connected to the input end of the latent feature extraction module 4, the output end of the latent feature extraction module 4 is connected to the input end of the output module 5, and the output end of the output module 5 is connected to the second input end of the storage module 2.

[0078] Wherein, the data acquisition module 1 is used to receive historical algal bloom monitoring data from the server.

[0079] The storage module 2 is used to store the received historical water bloom monitoring data and the interpolated values ​​of the water bloom monitoring data.

[0080] In this embodiment, Figure 3 As shown, the storage module 2 includes a first storage unit 21 and a second storage unit 22;

[0081] The first storage unit 21 is used to store the received historical water bloom monitoring data, and is stored and protected in a triple data structure. The triple data structure is in the form of Y=(t, s, v), where t represents the time of data collection, s represents the specific indicators monitored by various sensors, and v represents the specific values ​​of the indicators of type s monitored by various water bloom monitoring sensors at time t;

[0082] The second storage unit 22 is used to store the interpolation value of the water bloom monitoring data output by the output module 5 .

[0083] The matrix construction module 3 is used to construct an algae bloom monitoring data matrix according to the stored historical algae bloom monitoring data; specifically:

[0084] All triples Y = (t, s, v) in the water bloom monitoring data storage unit are constructed into a water bloom monitoring data matrix V of size I × J, where I is the number of data collection times, J is the number of various sensor monitoring indicators, and the v in the matrix ij represents the specific value of each type of water bloom monitoring sensor monitoring the j type of indicator at time i, 1≤i≤I, 1≤j≤J, and Γ is the set of known elements contained in the water bloom monitoring data matrix V. Since various sensors may lose data during data transmission and it is impossible to synchronously collect all indicators at all time points, the constructed water bloom monitoring data matrix V is a sparse incomplete matrix.

[0085] The latent feature extraction module 4 is used to construct a target loss function based on the historical algal bloom monitoring data and the algal bloom monitoring data matrix to extract latent features.

[0086] like Figure 4 As shown, the latent feature extraction module 4 includes an initialization unit 41, a target loss function construction unit 42, and a training unit 43;

[0087] An initialization unit 41 is used to initialize process parameters involved in the interpolation process of water bloom monitoring data;

[0088] The target loss function construction unit 42 is used to construct the target loss function according to the historical algal bloom monitoring data, the algal bloom monitoring data matrix, and the process parameters;

[0089] The training unit 43 is used to train and optimize the target loss function and extract latent features.

[0090] The output module 5 is used to output the interpolation value of the algal bloom monitoring data according to the extracted latent features.

[0091] The present invention provides a water bloom monitoring data interpolation method and device based on second-order optimization latent feature analysis, which is specifically used for water bloom monitoring data and can interpolate water bloom monitoring data that conforms to statistical laws and has high accuracy, so as to interpolate missing water bloom monitoring data. It can be widely used in the fields of water bloom monitoring, environmental monitoring, etc.

[0092] The present invention also provides an electronic device, comprising a processor, wherein the processor is used to run a computer program stored in a memory so that the electronic device implements the steps of the water bloom monitoring data interpolation method based on second-order optimized latent feature analysis in the above-mentioned embodiment.

[0093] The present invention also provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed on a processor, the steps of the water bloom monitoring data interpolation method based on second-order optimized latent feature analysis in the above-mentioned embodiment are implemented.

[0094] A computer program includes computer program code, which may be in source code form, object code form, executable file or some intermediate form. Computer readable media may include at least: any entity or device capable of carrying computer program code to an electronic device, recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk or an optical disk. In some jurisdictions, according to legislation and patent practice, computer readable media cannot be electric carrier signals and telecommunication signals.

[0095] Those skilled in the art will appreciate that the above-mentioned embodiments are specific examples for implementing the present invention, and in actual applications, various changes may be made thereto in form and detail without departing from the spirit and scope of the present invention.

Claims

1. The algae bloom monitoring data interpolation method based on second-order optimization latent feature analysis is characterized by: The specific steps include: S1: Collect historical algal bloom monitoring data from the server and store it; S2: construct an algae bloom monitoring data matrix V based on the stored historical algae bloom monitoring data; S3: Construct a target loss function based on historical algal bloom monitoring data and the algal bloom monitoring data matrix V, and output the algal bloom monitoring data interpolation value with the highest accuracy.

2. The algae bloom monitoring data interpolation method based on second-order optimized latent feature analysis according to claim 1, characterized in that: In S1, the historical algal bloom monitoring data is represented in the form of a triplet Y=(t, s, v), where t represents the time of data collection, s represents the indicators monitored by various sensors, and v represents the values ​​of the indicators of type s monitored by various algal bloom monitoring sensors at time t.

3. The algae bloom monitoring data interpolation method based on second-order optimized latent feature analysis according to claim 1, characterized in that: In S2, the construction method of the algal bloom monitoring data matrix V is: All triples Y = (t, s, v) are constructed into an I × J algae bloom monitoring data matrix V, where I is the number of data collection times, J is the number of various sensor monitoring indicators, and the v in the matrix ij represents the specific value of s type indicator monitored by various types of algal bloom monitoring sensors at time t, 1≤i≤I, 1≤j≤J, and Γ is the set of known elements contained in the algal bloom monitoring data matrix V.

4. The algae bloom monitoring data interpolation method based on second-order optimized latent feature analysis according to claim 1, characterized in that: The S3 includes: S3-1: Initialize the process parameters involved in the interpolation of algal bloom monitoring data; S3-2: construct the target loss function ε; S3-3: Iterate the training of the constructed target loss function ε; S3-4: Determine whether the loss function ε converges during the training iteration on Γ. If so, terminate the iteration; otherwise, continue the iteration. S3-5: Output the interpolated value of the algal bloom monitoring data and store it.

5. The algae bloom monitoring data interpolation method based on second-order optimized latent feature analysis according to claim 4 is characterized in that: In S3-1, the process parameters include the latent feature matrices T and S; the first-order gradient term g ε Including g ε (t i ) and g ε (s j ), corresponding to the first-order gradient components of the two latent feature matrices T and S respectively; the update increment vector Δε includes Δε(t i ) and Δε(s j ), corresponding to the updated incremental component vector of each row of the two latent feature matrices T and S respectively; the dimension of the latent feature matrix D; the convergence termination threshold δ; and the damping term coefficient γ.

6. The water bloom monitoring data interpolation method based on second-order optimized latent feature analysis according to claim 4 is characterized in that: In S3-2, the target loss function ε is: In formula (1), Γ is the set of known elements contained in the algal bloom monitoring data matrix V; v ij represents the monitoring value of the jth monitoring indicator of each type of algae bloom monitoring sensor in the algae bloom monitoring data matrix V at the i-th time; t i represents the i-th row vector in the temporal latent feature matrix T; s j It represents the j-th row vector in the latent feature matrix S of various sensor monitoring indicators; T represents the transposed matrix.

7. The algae bloom monitoring data interpolation method based on second-order optimized latent feature analysis according to claim 4 is characterized in that: In S3-3, the iteration formula is: In formula (2), represents the effect of ε on t in τ-1 training iterations i The first-order negative gradient information; t i Represents the i-th row vector in the temporal latent feature matrix T; represents the training iteration round s of ε in τ-1 j The first-order negative gradient information; s j represents the jth row vector in the latent feature matrix S of various sensor monitoring indicators; g ε Indicates that all and The first-order negative gradient information composed of the component vectors; Δ ε Indicates that g ε Corresponding to all and The component vectors are combined into an update increment vector; CGD represents the conjugate gradient second-order solver; γ is the damping term coefficient; ρ represents the convergence accuracy of the conjugate gradient second-order solver; M represents the maximum iteration round of the conjugate gradient second-order solver; Representation and Correspondingly, it is used to update the latent feature vector t i The update increment of Representation and Correspondingly, used to update the feature vector s j The update increment of Represents the value of the τth training iteration round of the i-th row vector in the temporal latent feature matrix T; Represents the value of the τ-1th training iteration round of the i-th row vector in the temporal latent feature matrix T; Represents the value of the j-th row vector in the latent feature matrix S of various sensor monitoring indicators in the τ-th training iteration round; It represents the value of the j-th row vector in the latent feature matrix S of various sensor monitoring indicators in the τ-1th training iteration round.

8. The algae bloom monitoring data interpolation method based on second-order optimized latent feature analysis according to claim 4 is characterized in that: In S3-5, the calculation formula for the interpolation value of the algal bloom monitoring data is: In formula (3), represents the interpolation value of algal bloom monitoring data; t i represents the i-th row vector in the temporal latent feature matrix T; s j Represents the j-th row vector in the latent feature matrix S of various sensor monitoring indicators.

9. A water bloom monitoring data interpolation device based on second-order optimized latent feature analysis based on the method according to any one of claims 1 to 8, characterized in that: include: A data acquisition module, used for receiving historical algal bloom monitoring data from a server; A storage module, used for storing received historical water bloom monitoring data and water bloom monitoring data interpolation values; A matrix construction module, used to construct an algae bloom monitoring data matrix V according to the stored historical algae bloom monitoring data; Hidden feature extraction module, used to construct target loss function based on historical algal bloom monitoring data and algal bloom monitoring data matrix, and extract hidden features; The output module is used to output the interpolation value of the algal bloom monitoring data according to the extracted latent features.

10. The algae bloom monitoring data interpolation device based on second-order optimization latent feature analysis according to claim 9, characterized in that: The latent feature extraction module includes an initialization unit, a target loss function construction unit, and a training unit; An initialization unit, used to initialize process parameters involved in the interpolation process of algal bloom monitoring data; A target loss function construction unit is used to construct a target loss function according to historical algal bloom monitoring data, algal bloom monitoring data matrix, and process parameters; The training unit is used to train and optimize the target loss function and extract latent features.

Images