Data prediction method and device based on artificial intelligence, computer equipment and medium

Abstract

The invention belongs to the technical field of artificial intelligence, and relates to a data prediction method and device based on artificial intelligence, computer equipment and a storage medium, and the method comprises the steps: obtaining to-be-processed initial typhoon data; performing data arrangement on the initial typhoon data to obtain typhoon data; performing feature screening processing on the typhoon data based on a quantum feature combination screening strategy to obtain feature data; performing classification processing on the feature data based on a quantum classification model to obtain category data; performing fusion processing on the category data and the feature data to obtain target feature data; performing typhoon loss prediction processing on the target feature data based on a preset quantum cascade time sequence regression prediction model to obtain a corresponding loss prediction result; and carrying out output processing on the loss prediction result. In addition, the invention also relates to a block chain technology, and the loss prediction result can be stored in a block chain. The method can be applied to typhoon loss prediction scenes in the field of financial science and technology, and the processing efficiency and accuracy of typhoon loss prediction are improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology and can be applied to the financial technology field, particularly to data prediction methods, devices, computer equipment and storage media based on artificial intelligence. Background Technology

[0002] In the insurance industry, the scale of losses caused by natural disasters is extremely large. Typhoons, as a common and highly destructive natural disaster, bring enormous pressure to insurance companies in terms of payouts. Whenever a typhoon strikes, the damage to houses, infrastructure, and crop failures results in various losses, requiring insurance companies to bear high payout amounts after the disaster. This poses a significant challenge to the financial stability and operational sustainability of insurance companies.

[0003] To effectively mitigate claims risks, accurately predicting typhoon-related losses is a crucial step for insurance companies in responding to typhoon disasters. Accurate forecasting of typhoon damage not only helps insurance companies prepare financial reserves and claims plans in advance, but also optimizes insurance product pricing strategies and enhances the company's market competitiveness.

[0004] Currently, traditional regression methods, such as linear regression and multinomial regression, are mainly used in the field of typhoon loss prediction. However, these traditional methods have significant limitations in practical applications. Typhoon losses are influenced by a variety of complex factors, including the typhoon's intensity, path, landfall location, rainfall, the geographical environment of the affected area, building structure, and population density. These factors exhibit complex nonlinear relationships. Traditional regression methods often rely on simplistic assumptions, and when data is insufficient, the models struggle to adequately fit the data and accurately capture the complex intrinsic connections between typhoon losses and various factors. This results in low accuracy in typhoon loss prediction, failing to meet the needs of insurance companies in their actual operations.

[0005] Therefore, developing a method that can more accurately predict typhoon losses is of great practical significance. Summary of the Invention

[0006] The purpose of this application is to propose a data prediction method, apparatus, computer equipment, and storage medium based on artificial intelligence, in order to solve the technical problem that existing typhoon loss prediction methods have low accuracy.

[0007] Firstly, an artificial intelligence-based data prediction method is provided, including:

[0008] Obtain the initial typhoon data to be processed;

[0009] The initial typhoon data is processed to obtain the corresponding typhoon data;

[0010] perform feature screening processing on the typhoon data based on a preset quantum feature combination screening strategy to obtain corresponding feature data;

[0011] perform classification processing on the feature data based on a preset quantum classification model to obtain corresponding category data;

[0012] perform fusion processing on the category data and the feature data to obtain corresponding target feature data;

[0013] perform typhoon loss prediction processing on the target feature data based on a preset quantum cascade time series regression prediction model to obtain a corresponding loss prediction result;

[0014] perform output processing on the loss prediction result.

[0015] In a second aspect, a data prediction apparatus based on artificial intelligence is provided, and includes:

[0016] a first acquisition module configured to acquire initial typhoon data to be processed;

[0017] a collation module configured to perform data collation on the initial typhoon data to obtain corresponding typhoon data;

[0018] a screening module configured to perform feature screening processing on the typhoon data based on a preset quantum feature combination screening strategy to obtain corresponding feature data;

[0019] a classification module configured to perform classification processing on the feature data based on a preset quantum classification model to obtain corresponding category data;

[0020] a fusion module configured to perform fusion processing on the category data and the feature data to obtain corresponding target feature data;

[0021] a prediction module configured to perform typhoon loss prediction processing on the target feature data based on a preset quantum cascade time series regression prediction model to obtain a corresponding loss prediction result;

[0022] an output module configured to perform output processing on the loss prediction result.

[0023] In a third aspect, a computer device is provided, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the steps of the above-mentioned data prediction method based on artificial intelligence when executing the computer program.

[0024] In a fourth aspect, a computer readable storage medium is provided, which stores a computer program, and the computer program implements the steps of the above-mentioned data prediction method based on artificial intelligence when executed by a processor.

[0025] In the scheme implemented by the above artificial intelligence-based data prediction method, device, computer equipment and storage medium, first, the initial typhoon data to be processed is obtained; the initial typhoon data is arranged to obtain corresponding typhoon data; then, the typhoon data is subjected to feature screening processing based on a preset quantum feature combination screening strategy to obtain corresponding feature data; then, the feature data is subjected to classification processing based on a preset quantum classification model to obtain corresponding category data; subsequently, the category data and the feature data are subjected to fusion processing to obtain corresponding target feature data; further, the target feature data is subjected to typhoon loss prediction processing based on a preset quantum cascade time series regression prediction model to obtain a corresponding loss prediction result; finally, the loss prediction result is subjected to output processing. Based on the above automatic processing procedure, the initial typhoon data obtained is arranged to obtain typhoon data, then the typhoon data is subjected to feature screening processing based on the use of the quantum feature combination screening strategy to obtain feature data, then the feature data is subjected to classification processing based on the use of the quantum classification model to obtain category data, and the category data and the feature data are subjected to fusion processing to obtain target feature data, and then the target feature data is subjected to typhoon loss prediction processing based on the use of the quantum cascade time series regression prediction model to obtain a loss prediction result, and finally the loss prediction result is subjected to output processing. In this way, the quantum feature combination screening strategy, the quantum classification model and the quantum cascade time series regression prediction model are used in combination to process the typhoon data by taking advantage of quantum computing, thereby realizing high-precision and high-efficiency typhoon loss prediction processing, effectively improving the processing efficiency of typhoon loss prediction, and ensuring the accuracy and reliability of the generated loss prediction result. BRIEF DESCRIPTION OF DRAWINGS

[0026] In order to more clearly illustrate the schemes in the present application, the drawings needed in the description of the embodiments of the present application will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0027] Figure 1 is an exemplary system architecture diagram to which the present application can be applied;

[0028] Figure 2 is a flowchart of one embodiment of the artificial intelligence-based data prediction method according to the present application;

[0029] Figure 3 is a structural schematic diagram of one embodiment of the artificial intelligence-based data prediction device according to the present application;

[0030] Figure 4 is a structural schematic diagram of one embodiment of the computer device according to the present application. DETAILED DESCRIPTION

[0031] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs; the terminology used in the description herein is for describing particular embodiments only and is not intended to be limiting of the application; the use herein of terms such as "comprise" and "have" and any variations thereof are intended to cover a non-exclusive inclusion; the use herein of terms such as "first", "second" and the like are intended to distinguish between similar objects unless the context indicates otherwise.

[0032] Reference herein to "embodiment" means that a particular feature, structure, or characteristic described in connection with the embodiment can be included in at least one embodiment of the application. The appearances of the phrase that the phrase in the specification appear in various places in the specification are not necessarily all referring to the same embodiment, or are necessarily referring to different or alternative embodiments.

[0033] In order to make the technical personnel in the art better understand the scheme of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings.

[0034] As shown in Figure 1 The system architecture 100 can include a terminal device 101, a network 102 and a server 103, and the terminal device 101 can be a notebook computer 1011, a tablet computer 1012 or a mobile phone 1013. The network 102 is a medium for providing a communication link between the terminal device 101 and the server 103. The network 102 can include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0035] A user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0036] The terminal device 101 can be various electronic devices with a display screen and supporting web browsing, in addition to the notebook computer 1011, the tablet computer 1012 or the mobile phone 1013, the terminal device 101 can also be an electronic book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 player (Moving Picture Experts Group Audio Layer IV), a laptop computer and a desktop computer, etc.

[0037] The server 103 can be a server providing various services, for example, a background server providing support for the page displayed on the terminal device 101.

[0038] It should be noted that the data prediction method based on artificial intelligence provided by the embodiments of the present application is generally executed by a server / terminal device, and accordingly, the data prediction device based on artificial intelligence is generally arranged in a server / terminal device.

[0039] It should be understood that Figure 1 The number of terminal devices, networks and servers in

[0040] With reference to Figure 2 , a flowchart of one embodiment of the data prediction method based on artificial intelligence according to the present application is shown. The order of the steps in the flowchart can be changed according to different needs, and some steps can be omitted. The data prediction method based on artificial intelligence provided by the embodiments of the present application can be applied to any scenario requiring typhoon loss prediction, and then the data prediction method based on artificial intelligence can be applied to products in these scenarios, for example, typhoon loss prediction scenarios in the field of financial insurance. The data prediction method based on artificial intelligence comprises the following steps:

[0041] In step S201, initial typhoon data to be processed is obtained.

[0042] In the present embodiment, the electronic device (for example Figure 1The server / terminal device shown) can obtain the initial typhoon data to be processed through a wired connection or a wireless connection. It should be noted that the wireless connection can include but is not limited to 3G / 4G / 5G connection, WiFi connection, Bluetooth connection, WiMAX connection, Zigbee connection, UWB (ultra wideband) connection, and other now known or future developed wireless connection. The execution subject of the present application is a data prediction system, which can be referred to as a system. The present application can be applied to the typhoon loss prediction scene in the field of finance and insurance. The above-mentioned initial typhoon data refers to multi-dimensional data related to the area to be measured, which can be obtained by extracting the corresponding typhoon feature data from the meteorological department, geographic information system (GIS) and historical disaster database. At least can include: meteorological parameters: wind speed, air pressure, precipitation, sea surface temperature anomaly, etc. (continuous variable). Geographic information: coastal population density, building density, terrain elevation (continuous / discrete mixed). Historical records: the same area typhoon loss amount and disaster area in the past specified period (such as 5 years) (time series related features).

[0043] Step S202, data arrangement is performed on the initial typhoon data to obtain corresponding typhoon data.

[0044] In the embodiment, the specific implementation process of the above-mentioned data arrangement on the initial typhoon data to obtain the corresponding typhoon data will be further described in detail in the subsequent specific embodiments, and will not be described here.

[0045] Step S203, feature screening processing is performed on the typhoon data based on a preset quantum feature combination screening strategy to obtain corresponding feature data.

[0046] In the embodiment, the specific implementation process of the above-mentioned feature screening processing on the typhoon data based on the preset quantum feature combination screening strategy to obtain the corresponding feature data will be further described in detail in the subsequent specific embodiments, and will not be described here.

[0047] Step S204, classification processing is performed on the feature data based on a preset quantum classification model to obtain corresponding category data.

[0048] In the embodiment, the specific implementation process of the above-mentioned classification processing on the feature data based on the preset quantum classification model to obtain the corresponding category data will be further described in detail in the subsequent specific embodiments, and will not be described here.

[0049] Step S205, fusion processing is performed on the category data and the feature data to obtain corresponding target feature data.

[0050] In the embodiment, the specific implementation process of fusing the category data and the feature data to obtain the corresponding target feature data will be further described in details in subsequent embodiments, and will not be described in details here.

[0051] In step S206, the target feature data is processed by the preset quantum cascade time series regression prediction model to obtain a corresponding loss prediction result.

[0052] In the embodiment, the specific implementation process of processing the target feature data by the preset quantum cascade time series regression prediction model to obtain the corresponding loss prediction result will be further described in details in subsequent embodiments, and will not be described in details here.

[0053] In step S207, the loss prediction result is outputted.

[0054] In the embodiment, the generated loss prediction result is sent to the corresponding insurance company's claim processing personnel to complete the output of the loss prediction result, and facilitate the subsequent insurance claim processing of the claim processing personnel according to the loss prediction result corresponding to the initial typhoon data.

[0055] The application first acquires initial typhoon data to be processed, and performs data arrangement on the initial typhoon data to obtain corresponding typhoon data. Then, the typhoon data is subjected to feature screening processing based on a preset quantum feature combination screening strategy to obtain corresponding feature data. Subsequently, the feature data is subjected to classification processing based on a preset quantum classification model to obtain corresponding category data. Subsequently, the category data and the feature data are subjected to fusion processing to obtain corresponding target feature data. Further, the target feature data is subjected to typhoon loss prediction processing based on a preset quantum cascaded time series regression prediction model to obtain a corresponding loss prediction result. Finally, the loss prediction result is subjected to output processing. Based on the above automatic processing procedure, the application obtains typhoon data by performing data arrangement on the acquired initial typhoon data, and then obtains feature data by performing feature screening processing on the typhoon data based on the use of the quantum feature combination screening strategy. Subsequently, category data is obtained by performing classification processing on the feature data based on the use of the quantum classification model, and target feature data is obtained by performing fusion processing on the category data and the feature data. Further, a loss prediction result is obtained by performing typhoon loss prediction processing on the target feature data based on the use of the quantum cascaded time series regression prediction model. Finally, the loss prediction result is subjected to output processing. In this way, the application uses the combination of the quantum feature combination screening strategy, the quantum classification model and the quantum cascaded time series regression prediction model to process the typhoon data by taking advantage of quantum computing, thereby realizing high-precision and high-efficiency typhoon loss prediction processing, effectively improving the processing efficiency of typhoon loss prediction, and ensuring the accuracy and reliability of the generated loss prediction result.

[0056] In some optional implementations, step S204 includes the following steps:

[0057] A preset feature dimension reduction strategy is acquired.

[0058] In this embodiment, the strategy content of the feature dimension reduction strategy includes reducing the dimension of the screened feature data from the original dimension to the dimension determined in the model training process. Specifically, the feature data can be reduced to the specified dimension (e.g., from 600 dimensions to 50 dimensions) determined in the training stage by quantum principal component analysis (QPCA). The projection measurement of the quantum state is used to realize efficient dimension reduction and preserve the main variation mode of the data.

[0059] The feature data is subjected to dimension reduction processing based on the feature dimension reduction strategy to obtain corresponding first feature data.

[0060] In this embodiment, the dimension reduction processing of the feature data can be performed based on the strategy content of the feature dimension reduction strategy to obtain the corresponding first feature data.

[0061] Encode the first feature data based on a preset quantum state encoding mode to obtain corresponding second feature data.

[0062] In this embodiment, the dimension-reduced feature values (first feature data) are mapped to quantum states through angle encoding (Angle Embedding) by using the same quantum state encoding mode as in the training phase, so as to obtain corresponding second feature data. For example, a feature value xi is converted into a rotation angle θi = arctan(xi) and applied to an Ry(θi) gate of a quantum bit.

[0063] Classify the second feature data based on the quantum classification model to obtain corresponding output results.

[0064] In this embodiment, the obtained second feature data is input into a pre-trained quantum classification model, which outputs a loss level label yclass ∈ {0, 1, …, 9} (for example, level 3 represents medium loss) based on the quantum kernel function (such as quantum radial basis kernel) optimized during training, that is, the output results.

[0065] Take the output results as the category data of the feature data.

[0066] In this embodiment, the quantum classification model can effectively improve the flexibility of the classification boundary by utilizing the high-dimensional space expression capability of quantum states, and is particularly suitable for non-linearly separable loss level division.

[0067] The present application obtains a preset feature dimension reduction strategy, and dimension-reduces the feature data based on the feature dimension reduction strategy to obtain corresponding first feature data. Then, the first feature data is encoded based on a preset quantum state encoding mode to obtain corresponding second feature data. Subsequently, the second feature data is classified based on the quantum classification model to obtain corresponding output results. Finally, the output results are taken as the category data of the feature data. Based on the above processing procedure, the present application dimension-reduces the feature data to obtain first feature data based on the use of a feature dimension reduction strategy, then encodes the first feature data based on the use of a quantum state encoding mode to obtain second feature data, and then classifies the second feature data based on the use of a quantum classification model, so as to efficiently and accurately generate the required category data and improve the data accuracy of the generated category data.

[0068] In some optional implementations of this embodiment, step S205 includes the following steps:

[0069] Obtain a plurality of preset splicing strategies.

[0070] In the embodiment, the selection of the splicing strategy is not specifically limited, and can be determined according to actual business requirements. The splicing strategy at least includes a direct splicing strategy and a weighted splicing strategy. The strategy content of the direct splicing strategy is to merge the category data and the feature data in sequence. The strategy content of the weighted splicing strategy is to emphasize the weight of the category data, determine the weight coefficient of the category data through cross-validation, and then splice the category data and the product of the category data and the weight coefficient in sequence.

[0071] The target splicing strategy is selected from all the splicing strategies.

[0072] In the embodiment, one strategy can be randomly selected from all the splicing strategies as the target splicing strategy.

[0073] The category data and the feature data are spliced based on the target splicing strategy to obtain corresponding splicing data.

[0074] In the embodiment, the splicing of the category data and the feature data can be performed according to the strategy content of the selected target splicing strategy, and the obtained splicing data is used as the target feature data.

[0075] The splicing data is used as the target feature data.

[0076] In the embodiment, the category data is added to the original feature data as new features, so that the feature information can be further enriched.

[0077] The application obtains a plurality of preset splicing strategies, selects a target splicing strategy from all the splicing strategies, splices the category data and the feature data based on the target splicing strategy to obtain corresponding splicing data, and then uses the splicing data as the target feature data. Based on the above processing procedure, the application selects a target splicing strategy from a plurality of splicing strategies, and then splices the category data and the feature data based on the use of the target splicing strategy, so that the fusion processing between the category data and the feature data can be efficiently and accurately completed, and the richness and accuracy of the generated target feature data are effectively improved.

[0078] In some optional implementation manners, the quantum cascade time sequence regression prediction model at least includes a variational quantum circuit and a full connection layer; and step S206 includes the following steps:

[0079] The target feature data is subjected to quantum state preparation processing based on the quantum cascade time sequence regression prediction model to obtain corresponding quantum state data.

[0080] In the embodiment, the preparation process of the quantum state includes initialization, superposition state generation, and rotation encoding. Specifically, the initialization includes: allocating 1 qubit (quantum bit) for each element in the target feature data, a total of n qubits. The superposition state generation includes: applying a Hadamard gate to all qubits to generate an unbiased superposition state, i.e., a quantum superposition state. The rotation encoding includes: generating two rotation angles for each element: θi,1=arctan(xi), θi,2=arctan(xi2). In turn, use the Ry(θi,1) gate to rotate the quantum state along the y-axis, and use the Rz(θi,2) gate to rotate the quantum state along the z-axis, and embed the classical data into the quantum state.

[0081] Based on the preset quantum gating mechanism, the quantum state data is calculated and processed by the variational quantum circuit to obtain corresponding processing data.

[0082] In the embodiment, the design of the variational quantum circuit includes: the structure of the variational quantum circuit is composed of three layers of repeating units, each layer includes: entanglement generation: 2n CNOT gates form a chain entanglement to establish quantum correlation. Parameterized rotation: let each qubit pass through three single-qubit rotation gates in turn: use the Rx(θ1) gate to rotate the quantum state along the x-axis, use the Ry(θ2) gate to rotate the quantum state along the y-axis, and use the Rz(θ3) gate to rotate the quantum. The rotation angles are updated iteratively in the gradient descent algorithm.

[0083] The quantum gating mechanism includes: the quantum cascade time sequence regression prediction model (QLSTM) realizes three types of gating (forgetting gate ft, input gate it, and output gate ot) through the VQC module: calculate the forgetting gate: ft-qlstm=σ(VQC1(vt)), calculate the input gate: it-qlstm=σ(VQC2(vt)), calculate the cell state at this time: Ct-qlstm=ft-qlstm*ct-1+it-qlstm*tanh(VQC3(vt)), calculate the output gate: ot-qlstm=σ(VQC4(vt)), calculate the hidden layer state of the network at this time: ht-qlstm=VQC5(ot-qlstm*tanh(Ct-qlstm)).

[0084] Based on the full connection layer, the processing data is predicted to obtain a corresponding prediction result.

[0085] In the embodiment, the hidden state of the final time step, i.e., the processing data ht-qlstm, is input into the full connection layer, so as to output the prediction result of the typhoon loss value: ypred=W·VQC6(ht-qlstm)+b.

[0086] The prediction result is taken as the loss prediction result.

[0087] In this embodiment, the goal of the constructed quantum cascade temporal regression prediction model is to integrate QSVM classification prediction results with QAOA screening features, and utilize a quantum long short-term memory network (QLSTM) to capture the temporal dependence of typhoon losses, thereby achieving high-precision regression prediction. Specifically, by inputting preprocessed structured temporal features into the constructed QLSTM regression model, weight updates and hyperparameter optimization are performed end-to-end on the training set. After sufficient iterative convergence, the network parameters are solidified, resulting in the output of a final model with accurate typhoon loss prediction capabilities, namely the quantum cascade temporal regression prediction model.

[0088] This application prepares quantum states for target feature data based on the quantum cascaded temporal regression prediction model to obtain corresponding quantum state data. Then, based on a preset quantum gating mechanism, the quantum state data is processed by the variable quantum circuit to obtain corresponding processed data. Subsequently, the processed data is predicted using the fully connected layer to obtain corresponding prediction results. These prediction results are then used as the loss prediction results. Based on this processing flow, this application prepares quantum states for target feature data using the quantum cascaded temporal regression prediction model to obtain corresponding quantum state data. Then, based on the quantum gating mechanism, the quantum state data is calculated using the variable quantum circuit to obtain corresponding processed data. Finally, the processed data is predicted using the fully connected layer. This allows for efficient and accurate typhoon loss prediction processing of target feature data, improving the processing efficiency of typhoon loss prediction and ensuring the accuracy of the obtained loss prediction results.

[0089] In some alternative implementations, step S203 includes the following steps:

[0090] Invoke the preset optimal feature subset filtering rules.

[0091] In this embodiment, the aforementioned optimal feature subset selection rule refers to a quantum feature combination selection strategy based on a multi-index weighted Q matrix, i.e., the optimal feature subset selection rule determined during model training. The optimal feature subset obtained through quantum optimization can be regarded as a "quantum compression" result of the data structure, possessing global optimality, and therefore is suitable for direct reuse.

[0092] The screening steps corresponding to the quantum feature combination screening strategy based on multi-index weighted Q matrix include:

[0093] Step 1: Data preprocessing and correlation matrix initialization.

[0094] 1) Input data standardization. Preprocess the data of each feature factor related to typhoon prediction to ensure the normalization and consistency of the data. The specific processing methods are as follows: (1) Label Encoding is used for discrete feature factors to convert them into numerical form to meet the requirements of model input; (2) Z-score Normalization is implemented for continuous feature factors to eliminate the dimensional differences between different features and improve the comparability of data and the training effect of the model.

[0095] 2) Multi-dimensional correlation coefficient calculation. Calculate three types of correlation indicators for each feature pair (xi, xj) and (xi, y): Pearson coefficient, mutual information, and Spearman rank correlation coefficient. Among them, the meaning of feature pair (xi, xj) includes: xi and xj are two different features in the original feature set, for example, in the typhoon loss feature, x1 may be "maximum wind speed" and x2 may be "path length". Function: Calculate the correlation between xi and xj, the purpose is to measure the redundancy between features (i.e., whether two features contain similar or repeated information). Correspondingly, the meaning of correlation indicators for feature pair (xi, xj) includes: Pearson coefficient (Pearson): measures linear correlation, the value range is [-1, 1], the closer the absolute value is to 1, the stronger the linear relationship. Spearman rank correlation coefficient: measures monotonic correlation (without linear), suitable for non-linear but monotonic relationships (such as exponential growth). Mutual information (Mutual Information, MI): measures statistical dependence, can capture any form of non-linear relationship, the value range is [0, +∞), the larger the value, the stronger the dependence.

[0096] In addition, the meaning of feature pair (xi, y) includes: xi is a feature in the original feature set, and y is the target variable (such as typhoon loss amount or loss category). Function: Calculate the correlation between xi and y, the purpose is to measure the discriminability of the feature pair to the target (i.e., whether the feature is helpful for predicting the target). Correspondingly, the meaning of correlation indicators for feature pair (xi, xj) includes: Pearson coefficient: if y is continuous, it can measure the linear prediction ability of xi and y. Spearman rank correlation coefficient: suitable for cases where y is an ordered category or a non-linear monotonic relationship. Mutual information: directly measures how much information xi contains about y, suitable for cases where y is a discrete category or a complex non-linear relationship.

[0097] 3) Construct a weighted comprehensive correlation matrix Q.

[0098] (1) Define the diagonal elements (feature-target correlation):

[0099] Q ii = α * |σ iy|+β*r s,iy +γ*I(x i ; y)

[0100] where α+β+γ=1, ρ is Pearson coefficient, r is Spearman rank correlation coefficient, I is mutual information.

[0101] Define off-diagonal elements (inter-feature redundancy penalty):

[0102] Q ij = -(α*|σ ij |+β*r s,ij +γ*I(x i ; x j ))

[0103] The purpose is to maximize Qii (strong correlation between feature and target) and minimize Qij (low redundancy between features).

[0104] Step 2: Build QUBO problem and quantum encoding.

[0105] 1) Define feature selection variables: introduce binary variables zi∈{0,1}, zi=1 represents selecting the ith feature, otherwise not.

[0106] 2) Build the target Hamiltonian H, convert the feature selection problem into maximizing the energy of Q matrix:

[0107]

[0108] where λ>0 is the redundancy penalty coefficient. The first term rewards, i.e., highly discriminative features (Qii is large, then zi=1 has high returns), the second term penalizes inter-feature redundancy (if Qij is strongly negatively correlated, then the penalty is heavier when zizj=1).

[0109] 3. Convert to QUBO form:

[0110]

[0111] where a i = -Q ii , b ij = λQ ij .

[0112] Step 3: QAOA quantum optimization solution.

[0113] 1) Quantum circuit initialization. (1) Map the feature screening problem to the quantum bit space using amplitude encoding or ground state encoding. Prepare the superposition state |ψ0> of N = 600 quantum bits. (2) Design a parameterized quantum circuit: construct the quantum annealing Hamiltonian H = HP + HD, where HP is the problem Hamiltonian; HD is the driving Hamiltonian, which is used to introduce quantum entanglement. The quantum annealing process evolves the system from the initial state to the target state through time evolution.

[0114] 2) Classical-quantum collaborative optimization. (1) Execute the parameterized circuit on the quantum processor, and measure the state |ψ0>. (2) Calculate the expected value E(θ) = <ψ(θ) | H | ψ(θ)>. (3) Update θ = {γ, β} by the classical optimizer (such as COBYLA) to minimize E(θ).

[0115] 3) Output the optimal feature subset. (1) Measure the final state |ψopt>, and obtain the binary string z* = (z1*, …, z600*); (2) Screen the features with zi* = 1 to form the combination Sopt = {xi | zi* = 1}, and finally output the optimal feature subset (containing k features).

[0116] Based on the optimal feature subset screening rule, the specified features matching the preset optimal feature subset are screened from the typhoon data.

[0117] In this embodiment, the optimal feature subset screening rule determined in the training process is directly used. According to the typhoon data after data processing, the features matching the optimal feature subset, i.e., the above-mentioned specified features, are screened.

[0118] Obtain the feature values corresponding to the specified features.

[0119] In this embodiment, if the optimal feature subset determined in the training process contains features A, B, and C, the values of the three features are automatically extracted from the typhoon data after data processing to obtain the above-mentioned feature values.

[0120] The feature values are taken as the feature data.

[0121] In this embodiment, the quantum-driven global optimal feature combination search (QAOA) provided in this embodiment: by encoding the high-dimensional typhoon loss feature subset search problem into a quadratic unconstrained binary optimization (QUBO), using a quantum approximate optimization algorithm to traverse 2 600A combination, approaching global optimality. Quantum superposition and tunneling effect simultaneously capture feature-target correlation and inter-feature nonlinear synergy, explicitly identify entangled features with "low individual weight-high combined value", breaking through the dimension disaster and local optimal bottleneck of classical methods. Filter out those feature subsets that are not important individually but contribute significantly to the prediction of y in combination. This is something that classical feature importance ranking cannot do.

[0122] The application filters out the specified features matching the preset optimal feature subset from the typhoon data based on the optimal feature subset filtering rule; then obtains the feature values corresponding to the specified features; and subsequently takes the feature values as the feature data. Based on the above processing flow, the application filters out the specified features matching the preset optimal feature subset from the typhoon data based on the use of the optimal feature subset filtering rule, and then obtains the feature values corresponding to the specified features as the required feature data, so as to automatically and accurately complete the feature filtering processing of the typhoon data and improve the accuracy and adaptability of the obtained feature data.

[0123] In some optional implementations of the embodiment, step S202 includes the following steps:

[0124] A preset standardization strategy is obtained.

[0125] In the embodiment, the policy content of the standardization strategy includes: discrete feature encoding: for discrete features such as typhoon occurrence area and typhoon grade, label encoding (Label Encoding) is used to convert them into numerical form. For example, "East China" is encoded as 1, and "South China" is encoded as 2, to ensure that the model can handle non-numeric data. Continuous feature standardization: for continuous features such as wind speed, air pressure, and economic loss, Z-score standardization is implemented. Calculate the mean (μ) and standard deviation (σ) of each feature, and convert the original value x to z=σx-μ, so that the data distribution meets the standard normal distribution (mean is 0, standard deviation is 1). This step eliminates the dimension difference and improves the model convergence speed.

[0126] The initial typhoon data is standardized based on the standardization strategy to obtain corresponding first typhoon data.

[0127] In the embodiment, the standardization processing of the initial typhoon data can be performed based on the policy content of the standardization strategy to obtain the corresponding first typhoon data.

[0128] The first typhoon data is subjected to time series information supplement processing to obtain corresponding second typhoon data.

[0129] In the embodiment, the time sequence information supplementing processing includes: taking the current typhoon to be predicted as the core, extracting historical data (such as loss value, path coordinates, etc.) of the last and next typhoon of the year in the region, and splicing the historical data and the first typhoon data into a complete time sequence record, and taking the historical data as the required second typhoon data. For example, when predicting the typhoon "Haiyan" in 2023, the data of the previous and next two typhoons is combined to form a feature vector containing time context. After the generation of the second typhoon data is completed, it can be further checked whether the time sequence order of the second typhoon data generated after splicing is strictly arranged in time to avoid causal inversion (such as "next" data should not be earlier than "current"). In addition, the time sequence supplementing can effectively enhance the model's ability to capture the evolution law of the typhoon, especially for loss prediction tasks with strong time dependence.

[0130] The second typhoon data is taken as the typhoon data.

[0131] In the embodiment, the original heterogeneous data is converted into a numerical form that can be processed by the model through feature engineering, and the standardization eliminates the dimension difference, and the time sequence splicing injects dynamic evolution information, thereby laying an accurate data foundation for subsequent typhoon loss prediction.

[0132] The application obtains a preset standardization strategy, then performs standardization processing on the initial typhoon data based on the standardization strategy to obtain corresponding first typhoon data, then performs time sequence information supplementing processing on the first typhoon data to obtain corresponding second typhoon data, and subsequently takes the second typhoon data as the typhoon data. Based on the above processing procedure, the application obtains the first typhoon data by standardizing the initial typhoon data based on the use of the standardization strategy, and then performs time sequence information supplementing processing on the first typhoon data, so that the data arrangement and processing of the initial typhoon data can be automatically and accurately completed, and the information richness and accuracy of the generated typhoon data are improved.

[0133] In some optional implementation manners of the embodiment, before step S204, the electronic device can further perform the following steps:

[0134] The historical typhoon loss data collected in advance is obtained.

[0135] In the embodiment, the historical typhoon loss data can refer to the typhoon loss data collected in a specified time period. The data collection process of the historical typhoon loss data can refer to the construction process of the typhoon data, which will not be described in detail here. In addition, the time selection of the specified time period is not specifically limited, and can be set according to actual business requirements, for example, it can be within the last 3 years.

[0136] The historical typhoon loss data is labeled to obtain corresponding typhoon sample data.

[0137] In this embodiment, the historical typhoon loss data is labeled in terms of loss level category labels by using manual labeling or machine labeling, and the labeled data is used as the required typhoon sample data.

[0138] A preset quantum support vector machine model is called.

[0139] In this embodiment, the model architecture of the quantum support vector machine model includes: quantum state preparation (feature mapping): by designing a parameterized quantum circuit, an input vector x is mapped to a quantum state φ(x) in a Hilbert space. With the superposition and entanglement characteristics of the quantum system, an exponentially high-dimensional feature representation is efficiently generated, laying the foundation for the classification task. Kernel function construction: the inner product of the quantum state is used to calculate the kernel matrix, which breaks through the dimensionality restriction problem of traditional classical kernel functions in high-dimensional space, and improves the efficiency and accuracy of the model in processing complex data.

[0140] Among them, according to the distribution of typhoon loss value, it can be divided into 10 segments in advance, and each segment corresponds to a loss level interval. For example, the first segment is the lowest loss level, and the tenth segment is the highest loss level. Each data is marked as the corresponding loss level category label according to the segment interval of its typhoon loss value, so as to convert the continuous loss value prediction problem into a 10-class classification problem.

[0141] Based on the preset quantum optimization algorithm, the typhoon sample data is used to train and optimize the quantum support vector machine model until a target model meeting the construction requirements is obtained.

[0142] In this embodiment, by inputting the typhoon sample data labeled with loss level category labels into the quantum support vector machine model, and optimizing the model parameters through the quantum optimization algorithm, the quantum support vector machine model can maximize the classification interval and minimize the classification error rate on the training data, until a trained target model is obtained and used as the required quantum classification model.

[0143] The target model is used as the quantum classification model.

[0144] In this embodiment, the trained and optimized quantum classification model can classify and predict the loss level of new typhoon loss data. By encoding the new data feature values through the same quantum state encoding method and inputting them into the quantum classification model, the quantum classification model will output the loss level category label corresponding to the new data feature values according to the classification decision function obtained by training, thereby realizing the classification prediction of typhoon loss. For each piece of data to be tested, the above quantum classification model is used for prediction, and the prediction category y_class∈{0,1,…,9} is output.

[0145] Among them, the typhoon loss is classified into 10 grades by quantum support vector machine (QSVM), and the loss level prediction label (y_class) output by the model is constructed as a new time series feature factor. This feature factor is not a simple artificial binning result, but an explicit expression of the nonlinear discrimination boundary learned by QSVM using quantum entangled state mapping and high-dimensional Hilbert space kernel function under small sample conditions, which contains comprehensive high-order information of sample loss degree. For the first time, the decision output of the quantum classification model is converted into a quantifiable dynamic feature factor, so that the subsequent time series regression model (QLSTM) can directly use the sample loss potential evaluation signal extracted by QSVM to realize the lossless fusion of classification and regression information.

[0146] In addition, by designing the QAOA→QSVM→QLSTM quantum cascade framework, the small sample overfitting is suppressed by the three quantum computing characteristics. For the first time, the three levels of quantum computing anti-noise characteristics (quantum tunneling), high-dimensional representation advantages (Hilbert space mapping), and dynamic coherence (quantum gating) are combined to form an end-to-end overfitting immune architecture for small sample time series prediction:

[0147] (1) Feature layer anti-overfitting: QAOA is based on quantum tunneling effect and superposition state parallel search, and selects a low-redundancy and high-synergy quantum set in a 600-dimensional feature space for global optimization, thereby avoiding overfitting caused by high-dimensional noise from the source;

[0148] (2) Classification layer anti-overfitting: QSVM uses quantum state entanglement to construct a hyperplane, and its Hilbert space mapping capability makes the classification boundary maintain generalization under small samples.

[0149] (3) Regression layer anti-overfitting: QLSTM enhances memory units through quantum bit coherent state, and dynamically fuses time series dependence relationship through quantum gating mechanism.

[0150] The application obtains historical typhoon loss data collected in advance, then performs labeling processing on the historical typhoon loss data to obtain corresponding typhoon sample data, then calls a preset quantum support vector machine model, subsequently trains and optimizes the quantum support vector machine model based on a preset quantum optimization algorithm using the typhoon sample data until a target model meeting the construction requirement is obtained, and finally uses the target model as the quantum classification model. Based on the above processing procedure, the application obtains historical typhoon loss data collected in advance, performs labeling processing on the historical typhoon loss data to obtain typhoon sample data, and then trains and optimizes the quantum support vector machine model based on a quantum optimization algorithm using the typhoon sample data, so that a quantum classification model meeting the requirement is efficiently and accurately constructed, the model construction efficiency of the quantum classification model is improved, and the model effect of the obtained quantum classification model is ensured.

[0151] In some optional implementations, the obtained user information seeks user consent and complies with relevant laws and relevant policies.

[0152] In addition, the non-company software tools or components appearing in the embodiments of the application are only examples and do not represent actual use.

[0153] It should be understood that the size of the serial number of each step in the above embodiments does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the application.

[0154] It should be emphasized that, in order to further ensure the privacy and security of the above loss prediction result, the loss prediction result can also be stored in a node of a blockchain.

[0155] The blockchain referred to in the application is a new application mode of distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm and other computer technologies. Blockchain, in essence, is a decentralized database, a series of data blocks associated using cryptographic methods, each data block containing information about a batch of network transactions, used to verify the validity (anti-fraud) of the information and generate the next block. The blockchain can include a blockchain underlying platform, a platform product service layer, and an application service layer.

[0156] The embodiments of the application can obtain and process related data based on artificial intelligence technology. Artificial intelligence (AI) is the use of digital computers or computer-controlled machines to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0157] The basic technology of artificial intelligence generally includes technologies such as sensors, special artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction system, mechatronics, etc. The software technology of artificial intelligence mainly includes computer vision technology, robot technology, biometric identification technology, speech processing technology, natural language processing technology, and machine learning / deep learning, etc.

[0158] A person of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by computer readable instructions instructing related hardware, and the computer readable instructions can be stored in a computer readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiments of each method. Among them, the storage medium can be a non-volatile storage medium such as a magnetic disc, an optical disc, a read-only memory (ROM), or a random access memory (RAM).

[0159] It should be understood that, although each step in the flowchart of the accompanying drawings is shown in sequence according to the direction of the arrow, these steps are not necessarily executed in sequence according to the direction of the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and they can be executed in other orders. Moreover, at least part of the steps in the flowchart of the accompanying drawings can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence is not necessarily sequential, but can be executed alternately or alternately with at least part of other steps or sub-steps or stages of other steps.

[0160] Further referring to Figure 3 , as an implementation of the method shown in the above Figure 2 , the present application provides an embodiment of an artificial intelligence-based data prediction device. The device embodiment corresponds to the method embodiment shown in Figure 2 , and the device can be specifically applied to various electronic devices.

[0161] As shown in Figure 3 , the artificial intelligence-based data prediction device 300 described in the embodiment includes a first acquisition module 301, an arrangement module 302, a screening module 303, a classification module 304, a fusion module 305, a prediction module 306, and an output module 307. Among them:

[0162] The first acquisition module 301 is configured to acquire initial typhoon data to be processed.

[0163] The collation module 302 is configured to perform data collation on the initial typhoon data to obtain corresponding typhoon data.

[0164] The screening module 303 is configured to perform feature screening processing on the typhoon data based on a preset quantum feature combination screening strategy to obtain corresponding feature data.

[0165] The classification module 304 is configured to perform classification processing on the feature data based on a preset quantum classification model to obtain corresponding category data.

[0166] The fusion module 305 is configured to perform fusion processing on the category data and the feature data to obtain corresponding target feature data.

[0167] The prediction module 306 is configured to perform typhoon loss prediction processing on the target feature data based on a preset quantum cascade time series regression prediction model to obtain a corresponding loss prediction result.

[0168] The output module 307 is configured to perform output processing on the loss prediction result.

[0169] In the embodiment, the above modules or units are respectively used for operations corresponding to the steps of the data prediction method based on artificial intelligence of the foregoing embodiments, and thus will not be described here.

[0170] In some optional implementations of the embodiment, the classification module 304 includes:

[0171] The first obtaining sub-module is configured to obtain a preset feature dimension reduction strategy.

[0172] The first processing sub-module is configured to perform dimension reduction processing on the feature data based on the feature dimension reduction strategy to obtain corresponding first feature data.

[0173] The second processing sub-module is configured to perform encoding processing on the first feature data based on a preset quantum state encoding manner to obtain corresponding second feature data.

[0174] The classification sub-module is configured to perform classification processing on the second feature data based on the quantum classification model to obtain a corresponding output result.

[0175] The first determining sub-module is configured to take the output result as category data of the feature data.

[0176] In the embodiment, the above modules or units are respectively used for operations corresponding to the steps of the data prediction method based on artificial intelligence of the foregoing embodiments, and thus will not be described here.

[0177] In some optional implementations of the embodiment, the fusion module 305 includes:

[0178] a second obtaining sub-module, configured to obtain a plurality of preset splicing strategies;

[0179] a first screening sub-module, configured to screen a target splicing strategy from all the splicing strategies;

[0180] a splicing sub-module, configured to perform splicing processing on the category data and the feature data based on the target splicing strategy, to obtain corresponding spliced data;

[0181] a second determining sub-module, configured to determine the spliced data as the target feature data.

[0182] In the embodiment, the above modules or units are respectively used for performing operations corresponding to steps of the data prediction method based on artificial intelligence of the foregoing embodiments, and thus will not be described herein again.

[0183] In some optional implementations of the embodiment, the quantum cascade time series regression prediction model at least includes a variational quantum circuit and a fully connected layer; and the prediction module 306 includes:

[0184] a preparation sub-module, configured to perform quantum state preparation processing on target feature data based on the quantum cascade time series regression prediction model, to obtain corresponding quantum state data;

[0185] a calculation sub-module, configured to perform calculation processing on the quantum state data through the variational quantum circuit based on a preset quantum gate mechanism, to obtain corresponding processing data;

[0186] a prediction sub-module, configured to perform prediction processing on the processing data based on the fully connected layer, to obtain corresponding prediction results;

[0187] a third determining sub-module, configured to determine the prediction results as the loss prediction results.

[0188] In the embodiment, the above modules or units are respectively used for performing operations corresponding to steps of the data prediction method based on artificial intelligence of the foregoing embodiments, and thus will not be described herein again.

[0189] In some optional implementations of the embodiment, the screening module 303 includes:

[0190] a calling sub-module, configured to call a preset optimal feature subset screening rule;

[0191] a second screening sub-module, configured to screen, based on the optimal feature subset screening rule, a specified feature matching a preset optimal feature subset from the typhoon data;

[0192] a third obtaining sub-module, configured to obtain a feature value corresponding to the specified feature;

[0193] A fourth determining sub-module is configured to determine the feature value as the feature data.

[0194] In the embodiments, the modules or units are respectively configured to perform operations corresponding to the steps of the data prediction method based on artificial intelligence, and details are not described herein.

[0195] In some optional implementations of the embodiments, the sorting module 302 includes:

[0196] A fourth obtaining sub-module is configured to obtain a preset standardization strategy.

[0197] A third processing sub-module is configured to perform standardization processing on the initial typhoon data based on the standardization strategy, to obtain corresponding first typhoon data.

[0198] A fourth processing sub-module is configured to perform time sequence information supplement processing on the first typhoon data, to obtain corresponding second typhoon data.

[0199] A fifth determining sub-module is configured to determine the second typhoon data as the typhoon data.

[0200] In the embodiments, the modules or units are respectively configured to perform operations corresponding to the steps of the data prediction method based on artificial intelligence, and details are not described herein.

[0201] In some optional implementations of the embodiments, the data prediction apparatus based on artificial intelligence further includes:

[0202] A second obtaining module is configured to obtain historical typhoon loss data collected in advance.

[0203] A labeling module is configured to perform labeling processing on the historical typhoon loss data, to obtain corresponding typhoon sample data.

[0204] A calling module is configured to call a preset quantum support vector machine model.

[0205] A processing module is configured to perform training and optimization processing on the quantum support vector machine model based on a preset quantum optimization algorithm, using the typhoon sample data, until a target model meeting a construction requirement is obtained.

[0206] A determining module is configured to determine the target model as the quantum classification model.

[0207] In the embodiments, the modules or units are respectively configured to perform operations corresponding to the steps of the data prediction method based on artificial intelligence, and details are not described herein.

[0208] To solve the above technical problems, the embodiment of the present application further provides a computer device. For details, please refer to Figure 4 , Figure 4 The basic structure block diagram of the computer device of the embodiment is shown in the figure.

[0209] The computer device 4 comprises a memory 41, a processor 42 and a network interface 43 which are connected to each other through a system bus. It should be pointed out that only the computer device 4 with components 41-43 is shown in the figure, but it should be understood that all the shown components are not required to be implemented, and more or less components can be alternatively implemented. Among them, the computer device herein can be understood by those skilled in the art as a device capable of automatically performing numerical calculation and / or information processing according to pre-set or stored instructions, and the hardware thereof includes but is not limited to a microprocessor, an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA), a digital signal processor (DSP), an embedded device, etc.

[0210] The computer device can be a desktop computer, a notebook computer, a palm computer, a cloud server and the like. The computer device can perform human-computer interaction with a user through a keyboard, a mouse, a remote controller, a touchpad, a voice control device and the like.

[0211] The memory 41 includes at least one type of readable storage medium, such as a flash memory, a hard disk, a multimedia card, a card-type memory (e.g., an SD or DX memory, etc.), a random access memory (RAM), a static random access memory (SRAM), a read-only memory (ROM), an electrically erasable programmable read-only memory (EEPROM), a programmable read-only memory (PROM), a magnetic memory, a magnetic disk, an optical disk, etc. In some embodiments, the memory 41 can be an internal storage unit of the computer device 4, such as a hard disk or a memory of the computer device 4. In other embodiments, the memory 41 can also be an external storage device of the computer device 4, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the computer device 4. Of course, the memory 41 can also include both the internal storage unit and the external storage device of the computer device 4. In this embodiment, the memory 41 is generally used to store an operating system and various application software installed on the computer device 4, such as computer readable instructions of the artificial intelligence-based data prediction method, etc. In addition, the memory 41 can also be used to temporarily store various data that have been output or will be output.

[0212] The processor 42 can be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor 42 is generally used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to run computer readable instructions or process data stored in the memory 41, such as computer readable instructions of the artificial intelligence-based data prediction method.

[0213] The network interface 43 can include a wireless network interface or a wired network interface, and is generally used to establish a communication connection between the computer device 4 and other electronic devices.

[0214] The present application also provides another embodiment, i.e., to provide a computer readable storage medium storing computer readable instructions, which can be executed by at least one processor to make the at least one processor perform the steps of the artificial intelligence-based data prediction method as described above.

[0215] Those skilled in the art can clearly understand the above-mentioned embodiment method can be realized by means of software and the necessary general hardware platform, of course, it can also be realized by hardware, but in many cases, the former is a better embodiment. Based on such understanding, the technical solutions of the present application can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes a plurality of instructions for making a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) execute the method described in each embodiment of the present application.

[0216] Obviously, the above-described embodiments are only some of the embodiments of the present application, not all the embodiments, and the drawings show the preferred embodiments of the present application, but do not limit the patent scope of the present application. The present application can be implemented in many different forms, and conversely, the purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions recorded in the foregoing specific embodiments, or make equivalent replacements to some of the technical features. Any equivalent structure made by using the content of the specification and drawings, directly or indirectly applied to other related technical fields, is also within the scope of the patent protection of the present application.

Claims

1. A data prediction method based on artificial intelligence, characterized in that, Includes the following steps: Obtain the initial typhoon data to be processed; The initial typhoon data is processed to obtain the corresponding typhoon data; The typhoon data is processed by a preset quantum feature combination screening strategy to obtain the corresponding feature data. The feature data is classified based on a preset quantum classification model to obtain the corresponding category data; The category data and the feature data are fused together to obtain the corresponding target feature data; The target feature data is processed for typhoon loss prediction based on a preset quantum cascade time-series regression prediction model to obtain the corresponding loss prediction results. The loss prediction results are then processed for output.

2. The data prediction method based on artificial intelligence according to claim 1, characterized in that, The step of classifying the feature data based on a preset quantum classification model to obtain the corresponding category data specifically includes: Obtain the preset feature dimensionality reduction strategy; The feature data is reduced in dimensionality based on the feature reduction strategy to obtain the corresponding first feature data. The first feature data is encoded based on a preset quantum state encoding method to obtain the corresponding second feature data; The second feature data is classified based on the quantum classification model to obtain the corresponding output results. The output results are used as the category data of the feature data.

3. The data prediction method based on artificial intelligence according to claim 1, characterized in that, The step of fusing the category data and the feature data to obtain the corresponding target feature data specifically includes: Obtain multiple preset splicing strategies; Select the target stitching strategy from all the described stitching strategies; Based on the target splicing strategy, the category data and the feature data are spliced ​​together to obtain the corresponding spliced ​​data; The spliced ​​data is used as the target feature data.

4. The data prediction method based on artificial intelligence according to claim 1, wherein the quantum cascaded time-series regression prediction model includes at least a variable quantum circuit and a fully connected layer; characterized in that, The step of performing typhoon loss prediction processing on the target feature data based on a preset quantum cascade time-series regression prediction model to obtain the corresponding loss prediction results specifically includes: Based on the quantum cascade time-series regression prediction model, the target feature data is processed to prepare quantum states, and the corresponding quantum state data is obtained. Based on a preset quantum gating mechanism, the quantum state data is processed by the variable quantum circuit to obtain the corresponding processed data. Based on the fully connected layer, the processed data is subjected to prediction processing to obtain the corresponding prediction result; The prediction result is used as the loss prediction result.

5. The data prediction method based on artificial intelligence according to claim 1, characterized in that, The step of performing feature filtering on the typhoon data based on a preset quantum feature combination filtering strategy to obtain corresponding feature data specifically includes: Invoke the preset optimal feature subset filtering rules; Based on the optimal feature subset filtering rules, specified features that match the preset optimal feature subset are filtered from the typhoon data. Obtain the feature value corresponding to the specified feature; The feature value is used as the feature data.

6. The data prediction method based on artificial intelligence according to claim 1, characterized in that, The step of processing the initial typhoon data to obtain the corresponding typhoon data specifically includes: Obtain the preset standardized strategy; The initial typhoon data is standardized based on the standardization strategy to obtain the corresponding first typhoon data. The first typhoon data is supplemented with time-series information to obtain the corresponding second typhoon data; The second typhoon data is used as the typhoon data.

7. The data prediction method based on artificial intelligence according to claim 1, characterized in that, Before the step of classifying the feature data based on a preset quantum classification model to obtain the corresponding category data, the method further includes: Acquire pre-collected historical typhoon loss data; The historical typhoon loss data is labeled to obtain the corresponding typhoon sample data; Invoke the preset quantum support vector machine model; Based on a preset quantum optimization algorithm, the quantum support vector machine model is trained and optimized using the typhoon sample data until a target model that meets the construction requirements is obtained. The target model is used as the quantum classification model.

8. A data prediction device based on artificial intelligence, characterized in that, include: The first acquisition module is used to acquire the initial typhoon data to be processed. The data processing module is used to process the initial typhoon data to obtain the corresponding typhoon data. The filtering module is used to perform feature filtering processing on the typhoon data based on a preset quantum feature combination filtering strategy to obtain corresponding feature data; The classification module is used to classify the feature data based on a preset quantum classification model to obtain the corresponding category data. The fusion module is used to fuse the category data and the feature data to obtain the corresponding target feature data; The prediction module is used to perform typhoon loss prediction processing on the target feature data based on a preset quantum cascade time-series regression prediction model, and obtain the corresponding loss prediction results. The output module is used to process the loss prediction results.

9. A computer device, characterized in that, The system includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the data prediction method based on artificial intelligence as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the data prediction method based on artificial intelligence as described in any one of claims 1 to 7.

Images