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

Abstract

The application discloses a water bloom monitoring data interpolation method and device based on second-order optimization hidden feature analysis, and 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; S3: constructing a target loss function according to the 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 application is specially used for water bloom monitoring data, can perform water bloom monitoring data interpolation in conformity with statistical rules and with high accuracy, and realizes the interpolation of missing water bloom monitoring data. The application can be widely applied in the fields of water bloom monitoring and environmental monitoring.

Description

Technical Field

[0001] This invention relates to the field of environmental protection technology, and in particular to a method and apparatus for interpolating algal bloom monitoring data based on second-order optimization latent feature analysis. Background Technology

[0002] With the continuous changes in aquatic environments and the increasing severity of global climate problems, algal blooms are occurring frequently in various regions, highlighting the growing importance of algal bloom monitoring. Algal bloom monitoring data, as a crucial indicator for measuring water health and environmental changes, reflects the degree of water pollution and the health of the aquatic ecosystem within the monitored area. Environmental management agencies and researchers can adjust water quality management plans and remediation measures based on algal bloom monitoring data. However, due to multi-source sensor malfunctions or communication issues leading to missing monitoring data, accurately analyzing and interpolating algal bloom monitoring data has become a key challenge in water protection and environmental management.

[0003] Currently, one feasible approach to imputing missing algal bloom monitoring data is to achieve accurate imputation using a representation learning method based on latent feature analysis. This method uses known historical algal bloom monitoring data to infer the missing data. During latent feature analysis, an incomplete algal bloom monitoring data matrix needs to be constructed between multi-source sensor monitoring data and the time series. Based on the known data in the matrix, two low-dimensional latent feature embedding representation matrices are trained, ensuring that the product of these two low-dimensional latent feature embedding matrices approximates the known algal bloom monitoring data matrix. However, due to the need for accurate imputation of the incomplete algal bloom monitoring data matrix, and the non-convex bilinear nature of the optimization objective of the latent feature analysis representation learning method, conventional first-order optimization-based latent feature analysis methods, which only consider first-order gradient information, are insufficient for accurate imputation of the incomplete algal bloom monitoring data matrix. Therefore, how to more accurately imput the incomplete algal bloom monitoring data matrix has become a critical problem that urgently needs to be solved. Summary of the Invention

[0004] To address the technical problem of missing data in existing algal bloom monitoring technologies, this invention proposes an algal bloom monitoring data interpolation method and apparatus based on second-order optimized latent feature analysis.

[0005] To achieve the above objectives, the present invention provides the following technical solution:

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

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

[0008] S2: constructing a water bloom monitoring data matrix V according to the stored historical water bloom monitoring data;

[0009] S3: constructing a target loss function according to the historical water bloom monitoring data and the water bloom monitoring data matrix V, and outputting the water bloom monitoring data interpolation value with the highest precision.

[0010] Preferably, in S1, the historical water bloom monitoring data is in the form of a triple Y=(t, s, v), t represents the time of data collection, s represents the index monitored by each type of sensor, and v represents the value monitored by each type of water bloom monitoring sensor at t time for s type index.

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

[0012] All triples Y=(t, s, v) 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 sensor monitoring indexes, and the element in the matrix represents the specific value monitored by each type of water bloom monitoring sensor at t time for s type index, 1≤i≤I, 1≤j≤J, and Γ is a set of known elements contained in the water bloom monitoring data matrix V.

[0013] Preferably, S3 includes:

[0014] S3-1: initializing the process parameters involved in the water bloom monitoring data interpolation process;

[0015] S3-2: constructing a target loss function ε;

[0016] S3-3: training iteration on the constructed target loss function ε;

[0017] S3-4: judging whether the loss function e converges in the iteration process on Γ, if yes, terminating the iteration, and if no, continuing the iteration;

[0018] S3-5: outputting the water bloom monitoring data interpolation value and storing.

[0019] Preferably, in S3-1, the process parameters include hidden feature matrices T and S; a first-order gradient term includes and , respectively corresponding to the first-order gradient components of the two hidden feature matrices T and S; an update increment vector includes and , respectively corresponding to the update increment component vectors of each row of the two hidden feature matrices T and S; a hidden feature matrix dimension D; a convergence termination threshold δ; and a damping term coefficient γ. ​

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

[0021]

[0022] where Γ is a set of known elements contained in the water bloom monitoring data matrix V; represents the monitoring value of each type of water bloom monitoring sensor at the i-th time at the j-th monitoring index in the water bloom monitoring data matrix V; represents the i-th row vector in the time hidden feature matrix T; represents the j-th row vector in the hidden feature matrix S of each type of sensor monitoring index.

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

[0024]

[0025] where, represents e In t -1 training iteration round of the first-order negative gradient information of ; represents the i-th row vector in the time hidden feature matrix T; represents e In t -1 training iteration round of the first-order negative gradient information of ; represents the j-th row vector in the hidden feature matrix S of each type of sensor monitoring index; represents the first-order negative gradient information composed of all and component vectors; represents the update increment vector corresponding to composed of all and component vectors; g is a damping term coefficient; p represents the convergence precision of the conjugate gradient second-order solver; M represents the maximum iteration round of the conjugate gradient second-order solver; represents the update increment corresponding to for updating the hidden feature vector ; represents the update increment corresponding to for updating the feature vector ; represents the value of the i-th row vector in the time hidden feature matrix T at the τ-th training iteration round; denotes the value of the i-th row vector in the time hidden feature matrix T in the (τ-1)-th training iteration round; denotes the value of the j-th row vector in the sensor monitoring index hidden feature matrix S in the τ-th training iteration round; denotes the value of the j-th row vector in the sensor monitoring index hidden feature matrix S in the (τ-1)-th training iteration round.

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

[0027]

[0028] wherein, denotes the algal bloom monitoring data interpolation value; denotes the i-th row vector in the time hidden feature matrix T; denotes the j-th row vector in the sensor monitoring index hidden feature matrix S.

[0029] The application also provides an algal bloom monitoring data interpolation device based on second-order optimization hidden feature analysis, comprising:

[0030] a data acquisition module configured to receive historical algal bloom monitoring data from a server;

[0031] a storage module configured to store the received historical algal bloom monitoring data and the algal bloom monitoring data interpolation value;

[0032] a matrix construction module configured to construct an algal bloom monitoring data matrix V according to the stored historical algal bloom monitoring data;

[0033] a hidden feature extraction module configured to construct a target loss function according to the historical algal bloom monitoring data and the algal bloom monitoring data matrix, and extract hidden features;

[0034] an output module configured to output the algal bloom monitoring data interpolation value according to the extracted hidden features.

[0035] Preferably, the hidden feature extraction module comprises an initialization unit, a target loss function construction unit and a training unit.

[0036] The initialization unit is configured to initialize process parameters involved in the algal bloom monitoring data interpolation process.

[0037] The target loss function construction unit is configured to construct a target loss function according to the historical algal bloom monitoring data and the algal bloom monitoring data matrix, and the process parameters.

[0038] The training unit is configured to train and optimize the target loss function, and extract hidden features.

[0039] In conclusion, compared with the prior art, the application has at least the following beneficial effects:

[0040] The application provides a water bloom monitoring data interpolation method and device based on second-order optimization hidden feature analysis, which is specially used for water bloom monitoring data and can perform water bloom monitoring data interpolation in accordance with statistical rules and with high accuracy, so as to realize interpolation of missing water bloom monitoring data. The application can be widely applied in the fields of water bloom monitoring and environmental monitoring. BRIEF DESCRIPTION OF DRAWINGS

[0041] Figure 1 The figure is a schematic diagram of the water bloom monitoring data interpolation method based on second-order optimization hidden feature analysis according to an exemplary embodiment of the application.

[0042] Figure 2 The figure is a schematic diagram of the water bloom monitoring data interpolation device based on second-order optimization hidden feature analysis according to an exemplary embodiment of the application.

[0043] Figure 3 The figure is a schematic diagram of the storage module according to an exemplary embodiment of the application.

[0044] Figure 4 The figure is a schematic diagram of the hidden feature extraction module according to an exemplary embodiment of the application. DETAILED DESCRIPTION

[0045] The application will be further described in detail below in combination with embodiments and specific implementation manners. However, it should not be understood that the scope of the above-mentioned subject matter of the application is limited to the following embodiments. Any technology realized based on the content of the application belongs to the scope of the application.

[0046] In the description of the application, it should be understood that the terms "longitudinal", "transverse", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer" and the like indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only used to facilitate the description of the application and simplify the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as a limitation of the application.

[0047] As shown in Figure 1 The application provides a water bloom monitoring data interpolation method based on second-order optimization hidden feature analysis, which specifically includes the following steps:

[0048] S1: receiving an instruction of interpolating water bloom monitoring data sent by a server, collecting historical water bloom monitoring data from the server, and storing the historical water bloom monitoring data in the form of a triple.

[0049] The server sends instructions to the water bloom monitoring data receiving module in the following ways: periodically, or according to the notification of the device, or according to the notification of a certain server, and sends instructions to the water bloom monitoring data receiving module.

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

[0051] S2: Construct a water bloom monitoring data matrix V according to the stored historical water bloom monitoring data.

[0052] In this embodiment, all triplets Y=(t, s, v) 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 indicators monitored by various sensors, and the elements in the matrix represent the specific values monitored by various water bloom monitoring sensors at t time for s type indicators, 1≤i≤I, 1≤j≤J, and Γ is a set of known elements contained in the water bloom monitoring data matrix V. Since various sensors cannot collect all indicators at all time points synchronously, and data loss may occur during data transmission, the constructed water bloom monitoring data matrix V is a sparse and incomplete matrix.

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

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

[0055] In this embodiment, the process parameters include the latent feature matrices T and S; the first-order gradient term includes and , which correspond to the first-order gradient components of the two latent feature matrices T and S, respectively; the update increment vector includes and , which correspond to the update increment component vectors of each row of the two latent feature matrices T and S, respectively; the latent feature matrix dimension D; the convergence termination threshold δ; the damping term coefficient γ;

[0056] wherein the latent feature matrix dimension D determines the dimension of the latent feature space of the latent feature matrices T and S, and is initialized as a positive integer;

[0057] The size of the hidden feature matrix T and S is determined by the dimension value of each order of the corresponding water bloom monitoring data matrix V, that is, T is a hidden feature matrix of I rows and D columns, S is a hidden feature matrix of J rows and D columns, and the two hidden feature matrices are initialized with smaller random positive numbers;

[0058] The convergence termination threshold δ is a parameter for judging whether the iteration process has converged, and is initialized with a very small positive number;

[0059] The damping term coefficient γ is a constant for controlling the positive definiteness of the second-order curvature matrix and balancing the degree of first-order and second-order approximation, and is initialized as a larger positive number.

[0060] S3-2: For the known data set Γ of the water bloom monitoring data matrix V, a target loss function ε is constructed to train the values of the hidden feature matrices T and S of the historical water bloom monitoring data; the target loss function ε is represented by the following formula:

[0061]

[0062] Where Γ is a set of known elements contained in the water bloom monitoring data matrix V; represents the monitoring value of each type of water bloom monitoring sensor at the i-th time at the j-th monitoring index in the water bloom monitoring data matrix V; represents the i-th row vector in the time hidden feature matrix T; represents the j-th row vector in the hidden feature matrix S of each type of sensor monitoring index.

[0063] The target loss function takes the Euclidean distance as the optimization objective.

[0064] S3-3: In the training process, the optimization target loss function e The first-order gradient information of the hidden feature matrix in the current training iteration round ; then in the training process, the optimization target loss function e The second-order curvature information of the hidden feature matrix in the current training iteration round, which is obtained by nonlinear scaling of the first-order gradient information obtained by the first-order optimization unit to obtain a more accurate update of the optimization objective ;

[0065] According to the initialized loss function and related parameters involved in the water bloom monitoring data interpolation process, the values of the hidden feature matrices T 、 S of the historical water bloom monitoring data are trained;

[0066] to minimize the optimization target loss function e so that the value of the optimization target loss function e can approach the second-order stable point as much as possible, and the training iteration formula is as follows:

[0067]

[0068] wherein, denotes e the first-order negative gradient information of the objective function t with respect to the time hidden feature vector in the time hidden feature matrix T at the τth training iteration round; denotes the ith row vector in the time hidden feature matrix T; denotes the jth row vector in the sensor monitoring index hidden feature matrix S; e the first-order negative gradient information of the objective function t with respect to the sensor monitoring index hidden feature vector in the sensor monitoring index hidden feature matrix S at the τth training iteration round; denotes the jth row vector in the sensor monitoring index hidden feature matrix S; denotes the first-order negative gradient information composed of all and component vectors; denotes the update increment vector corresponding to composed of all and component vectors; CGD denotes the conjugate gradient second-order solver; g is the damping term coefficient; p denotes the convergence precision of the conjugate gradient second-order solver; M denotes the maximum iteration round of the conjugate gradient second-order solver; denotes the update increment corresponding to for updating the hidden feature vector ; denotes the update increment corresponding to for updating the feature vector ; denotes the value of the ith row vector in the time hidden feature matrix T at the τth training iteration round; denotes the value of the ith row vector in the time hidden feature matrix T at the (τ-1)th training iteration round; denotes the value of the jth row vector in the sensor monitoring index hidden feature matrix S at the τth training iteration round; denotes the value of the jth row vector in the sensor monitoring index hidden feature matrix S at the (τ-1)th training iteration round.

[0069] S3-4: Determine whether the loss function e converges in the training iteration process on Γ, if yes, terminate the iteration, if not, continue the iteration.

[0070] The convergence determination condition is the value of the target loss function e calculated and the value of the current round target loss function of the continuous 10 rounds of training processe the value of the target loss function of the last round e the absolute value of the difference between the value of the target loss function of the last round d .

[0071] S3-5: output the algal bloom monitoring data interpolation value and store it.

[0072] In this embodiment, when the target loss function e converges on the known data set Γ, the hidden feature matrix e trained when the target loss function T reaches a minimum value is used to calculate the data acquisition time S with the highest accuracy algal bloom monitoring sensor data indicators algal bloom monitoring data interpolation value v̂ ij , and the calculated algal bloom monitoring data interpolation value is output to the algal bloom monitoring data interpolation value storage unit for storage, and the calculation formula is:

[0073]

[0074] wherein, represents the algal bloom monitoring data interpolation value; represents the i-th row vector in the time hidden feature matrix T; represents the j-th row vector in the sensor monitoring indicator hidden feature matrix S.

[0075] Based on the above-mentioned algal bloom monitoring data interpolation method based on second-order optimization hidden feature analysis, as shown in Figure 2 , the application also provides an algal bloom monitoring data interpolation device based on second-order optimization hidden feature analysis, which comprises a data acquisition module 1, a storage module 2, a matrix construction module 3, a hidden feature extraction module 4, and an output module 5.

[0076] The output end of the data acquisition module 1 is connected to the first input end of the storage module 2, and the output end of the storage module 2 is connected to the input end of the matrix construction module 3; the output end of the matrix construction module 3 is connected to the input end of the hidden feature extraction module 4, the output end of the hidden 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.

[0077] The data acquisition module 1 is configured to receive historical algal bloom monitoring data from a server.

[0078] The storage module 2 is configured to store the received historical algal bloom monitoring data and algal bloom monitoring data interpolation value.

[0079] In this embodiment, as shown in Figure 3As shown, the storage module 2 includes a first storage unit 21 and a second storage unit 22;

[0080] The first storage unit 21 is used to store the received historical algal bloom monitoring data, and the storage is protected by a triplet data structure. The triplet data structure is in the form of Y=(t, s, v), where t represents the data acquisition time, s represents the specific index monitored by various sensors, and v represents the specific value of the various algal bloom monitoring sensors for the s-type index at time t;

[0081] The second storage unit 22 is used to store the interpolated values ​​of the algal bloom monitoring data output by the output module 5.

[0082] Matrix construction module 3 is used to construct an algal bloom monitoring data matrix based on stored historical algal bloom monitoring data; specifically:

[0083] All triples Y=(t, s, v) in the algal bloom monitoring data storage unit are used to construct an I×J algal bloom monitoring data matrix V, where I represents the number of data acquisition times, J represents the number of monitoring indicators of various sensors, and v in the matrix represents the number of data points. ij Let Γ represent the specific values ​​of various algal bloom monitoring sensors for index j at time i, where 1 ≤ i ≤ I, 1 ≤ j ≤ J, and Γ is the set of known elements contained in the algal bloom monitoring data matrix V. Since data loss may occur during data transmission from various sensors, and it is impossible to synchronously collect all indices at all time points, the constructed algal bloom monitoring data matrix V is a sparse and incomplete matrix.

[0084] The latent feature extraction module 4 is used to construct a target loss function based on historical algal bloom monitoring data and algal bloom monitoring data matrix to extract latent features.

[0085] 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;

[0086] Initialization unit 41 is used to initialize the process parameters involved in the algal bloom monitoring data interpolation process.

[0087] The target loss function construction unit 42 is used to construct the target loss function based on historical algal bloom monitoring data, algal bloom monitoring data matrix, and process parameters.

[0088] Training unit 43 is used to train and optimize the target loss function and extract latent features.

[0089] Output module 5 is used to output interpolated values ​​of algal bloom monitoring data based on the extracted latent features.

[0090] The application provides a water bloom monitoring data interpolation method and device based on second-order optimization hidden feature analysis, which is specially used for water bloom monitoring data, can perform water bloom monitoring data interpolation in accordance with statistical rules and with high accuracy, and realizes interpolation of missing water bloom monitoring data. The application can be widely applied in the fields of water bloom monitoring, environmental monitoring and the like.

[0091] The application further provides an electronic device, which comprises a processor configured 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 optimization hidden feature analysis in the above embodiments.

[0092] The application further provides a computer readable storage medium, which stores a computer program, and the computer program implements the steps of the water bloom monitoring data interpolation method based on second-order optimization hidden feature analysis in the above embodiments when running on a processor.

[0093] The computer program comprises computer program code, which can be in the form of source code, object code, an executable file or some intermediate form. The computer readable medium can at least include any entity or device capable of carrying the computer program code to the electronic device, a recording medium, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal and a software distribution medium. For example, a U disk, a mobile hard disk, a magnetic disk or an optical disk.

[0094] Those skilled in the art can understand that the above embodiments are specific examples for implementing the application, and in actual applications, various changes can be made in form and details without departing from the spirit and scope of the application.

Claims

1. A data imputation method for monitoring algal blooms based on second-order optimization of latent feature analysis, characterized in that, Specifically comprising the following steps: S1: Collect historical water bloom monitoring data from the server and store them; S2: Construct a water bloom monitoring data matrix V according to the stored historical water bloom monitoring data; S3: Construct a target loss function according to the historical water bloom monitoring data and the water bloom monitoring data matrix V, and output the most accurate water bloom monitoring data interpolation value; The S3 comprises: S3-1: Initialize the process parameters involved in the water bloom monitoring data interpolation process; S3-2: Construct a target loss function ε; S3-3: Train and iterate the constructed target loss function ε; S3-4: judging the loss function ε whether the training iteration process on Γ converges, if yes, terminating the iteration, if not, continuing the iteration; S3-5: Output the water bloom monitoring data interpolation value and store it; In the S3-3, the training iteration formula is: wherein, denotes ε In τ -1 training iteration round of first-order negative gradient information; denotes the i-th row vector in the time hidden feature matrix T; denotes ε In τ -1 training iteration round of first-order negative gradient information; denotes the j-th row vector in the sensor monitoring index hidden feature matrix S; denotes combining all and component vectors into first-order negative gradient information; denotes the update increment vector corresponding to combined by all and component vectors into one; CGD denotes a conjugate gradient second-order solver; γ is a damping term coefficient; ρ denotes the convergence precision of the conjugate gradient second-order solver; M denotes the maximum iteration round of the conjugate gradient second-order solver; denotes the update increment corresponding to for updating the hidden feature vector ; denotes the update increment corresponding to for updating the feature vector ; denotes the value of the i-th row vector in the time hidden feature matrix T in the τ-th training iteration round; denotes the value of the i-th row vector in the time hidden feature matrix T in the τ-1-th training iteration round; denotes the value of the j-th row vector in the sensor monitoring index hidden feature matrix S in the τ-th training iteration round; denotes the value of the j-th row vector in the sensor monitoring index hidden feature matrix S in the τ-1-th training iteration round.

2. The method of bloom monitoring data imputation based on second order optimization latent feature analysis as claimed in claim 1, wherein, In the S1, the historical water bloom monitoring data is in the form of a triple Y=(t, s, v), t represents the time of data collection, s represents the indicators monitored by various sensors, and v represents the values monitored by various water bloom monitoring sensors at t time for s type indicators.

3. The method of bloom monitoring data imputation based on second order optimization latent feature analysis as claimed in claim 1, wherein, In the S2, the construction method of the water bloom monitoring data matrix V is: All the triplets Y = (t, s, v) are constructed into a bloom monitoring data matrix V of size I x J, where I is the number of data collection times, J is the number of various types of sensor monitoring indexes, and the matrix contains represents the specific value monitored by each type of bloom monitoring sensor at t time for s type index, 1≤i≤I, 1≤j≤J, and Γ is the set of known elements contained in the bloom monitoring data matrix V.

4. The method of bloom monitoring data imputation based on second order optimization latent feature analysis as claimed in claim 1, wherein, In the S3-1, the process parameters include an implicit feature matrix T and S; a first-order gradient term comprising and , respectively corresponding to the first-order gradient component of the two implicit feature matrices T and S; an update increment vector comprising and , respectively corresponding to the update increment component vector of each row of the two implicit feature matrices T and S; an implicit feature matrix dimension D; a convergence termination threshold δ; a damping term coefficient γ.

5. The second-order optimization based latent feature analysis for data imputation of algal bloom monitoring method of claim 1, wherein, In the S3-2, the target loss function ε is: where Γ is a set of known elements contained in the water bloom monitoring data matrix V; represents the monitoring value of each type of water bloom monitoring sensor in the water bloom monitoring data matrix V at the jth monitoring index at the ith time; represents the ith row vector in the time hidden feature matrix T. represents the jth row vector in the hidden feature matrix S of each type of sensor monitoring index.

6. The second-order optimization based latent feature analysis for data imputation of algal bloom monitoring method of claim 1, wherein, In the S3-5, the calculation formula of the water bloom monitoring data interpolation value is: wherein, represents the interpolated value of the algal bloom monitoring data; represents the i-th row vector in the time hidden feature matrix T; represents the j-th row vector in the hidden feature matrix S of various sensor monitoring indicators.

7. A device for data imputation of a water bloom monitoring based on second order optimization latent feature analysis based on the method according to any one of claims 1 to 6, characterized in that, Comprise: Data acquisition module, for receiving historical water bloom monitoring data from the server; Storage module, for storing the received historical water bloom monitoring data and water bloom monitoring data interpolation value; Matrix construction module, for constructing a water bloom monitoring data matrix V according to the stored historical water bloom monitoring data; Hidden feature extraction module, for constructing a target loss function according to the historical water bloom monitoring data and the water bloom monitoring data matrix, and extracting hidden features; Output module, for outputting the water bloom monitoring data interpolation value according to the extracted hidden features.

8. The apparatus for interpolating monitoring data of algal bloom in a second-order optimization hidden feature analysis as claimed in claim 7, wherein The hidden feature extraction module comprises an initialization unit, a target loss function construction unit and a training unit; The initialization unit is used to initialize the process parameters involved in the water bloom monitoring data interpolation process; The target loss function construction unit is used to construct a target loss function according to the historical water bloom monitoring data and the water bloom monitoring data matrix and the process parameters; The training unit is used to train and optimize the target loss function to extract hidden features.

Images