An insurance product development analysis method and system based on big data analysis

Abstract

The application relates to the technical field of big data analysis and information processing, and discloses an insurance product development analysis method and system based on big data analysis, which comprises the following steps: collecting real-time running state data of an electrical equipment, and establishing an actual state sequence; training a digital twin model based on historical data, and generating a predicted state sequence; comparing the actual state sequence with the predicted state sequence, constructing a state difference sequence, and extracting a structure fingerprint and a trend factor; using the structure fingerprint and the trend factor to form a joint vector, and establishing a credible function; generating a trend vector based on the trend characteristics of the credible function, judging the consistency of the trend vector, and dynamically adjusting the insurance strategy of the electrical equipment. The application realizes multi-grade dynamic response and adaptive linkage of the insurance strategy, forms a data-driven extended warranty strategy evolution closed loop, and significantly improves the intelligent level and forward-looking service capability of the guarantee strategy.

Description

Technical Field

[0001] This invention relates to the field of big data analysis and information processing technology, specifically to a method and system for developing insurance products based on big data analysis. Background Technology

[0002] With the increasing popularity of smart appliances and the growing demand for after-sales service from users, extending the lifespan of appliances and providing refined extended warranty strategies has become a key competitive advantage for home appliance companies and insurance platforms. Traditional extended warranty plans for appliances are usually based on uniform pricing and periodic configurations based on the equipment's factory parameters, years of use, or historical failure rates. They lack the perception and judgment of the actual operating status of the equipment and cannot reflect personalized and scenario-based dynamic protection capabilities.

[0003] While some existing extended warranty strategies incorporate equipment operation monitoring mechanisms, they often employ static threshold rules. For example, exceeding a certain power level or a temperature rise above a fixed temperature is considered a high-risk condition. This approach has significant limitations: first, it cannot accurately reflect the actual evolution trend of the equipment's condition, easily leading to misjudgments or omissions; second, it cannot identify the interaction relationships between different indicators; and third, the response mechanism is simplistic, with strategy adjustments often relying on manual intervention or static settings, lacking systematic automatic evolution capabilities.

[0004] Furthermore, current extended warranty service systems generally lack dynamic modeling methods to assess the deviation trend between the "predicted state" and the "actual state," making it impossible to use multi-dimensional indicators to collaboratively determine the stability of the device's operational trajectory, thus missing the optimal time for pre-emptive intervention. At the same time, policy triggering typically relies on threshold judgments, making it difficult to adapt to state fluctuations in complex scenarios, resulting in delayed service responses and impacting user experience and warranty quality.

[0005] Therefore, there is an urgent need for a new extended warranty strategy evolution method that integrates multi-source state perception, trend evolution judgment and strategy linkage mechanism, so as to realize the early perception of equipment state deviation, intelligent triggering and feedback optimization of protection plan, and build a dynamic protection system that can learn, adapt and close loop, so as to improve the intelligence level and technological advancement of extended warranty services. Summary of the Invention

[0006] In view of the above-mentioned problems, the present invention is proposed.

[0007] Therefore, the technical problem solved by this invention is that existing technologies cannot achieve accurate identification based on the evolution trend of device state.

[0008] To solve the above-mentioned technical problems, the present invention provides the following technical solution: an insurance product development and analysis method based on big data analysis, comprising: collecting real-time operating status data of electrical equipment and establishing an actual status sequence;

[0009] A digital twin model is trained based on historical data to generate a predicted state sequence;

[0010] The actual state sequence is compared with the predicted state sequence to construct a state difference sequence, and structural fingerprints and trend factors are extracted.

[0011] A reliable function is established by using the structural fingerprint and the trend factor to form a joint vector;

[0012] Trend vectors are generated based on the trend characteristics of the credibility function, the consistency of the trend vectors is judged, and the insurance strategy for electrical appliances is dynamically adjusted.

[0013] As a preferred embodiment of the insurance product development and analysis method based on big data analysis described in this invention, the real-time operating status data includes voltage, current, power, start-stop frequency, temperature, environmental disturbances, and user operations.

[0014] The real-time operating status data is integrated into an actual state vector at time t. Forming the actual state sequence S real (t).

[0015] As a preferred embodiment of the insurance product development and analysis method based on big data analysis described in this invention, the construction of the state difference sequence includes: comparing the actual state vector and the predicted state vector at the same moment, and calculating the difference vector Δ. t Within the time window w, the state difference sequence Δ(t) is formed;

[0016] The structural fingerprint includes encoding differential fluctuation symbols and identifying structural fragment patterns;

[0017] The encoded difference fluctuation symbol includes, within each time window w, the difference value at time i for the j-th state index dimension. The changing trend is used to encode symbols, resulting in the symbol sequence Σ. w ;

[0018] The identification of structural fragment patterns includes: extracting the symbol sequence into wave structure fragments according to time sequence, constructing a change pattern vector h(t); and encoding the change pattern vector within each time window w into a structural fingerprint vector.

[0019] The trend factors include the offset rate factor, the offset direction angle factor, and the trend reversal frequency factor.

[0020] The offset rate factor is the first derivative of the state difference vector;

[0021] The offset direction angle factor is the angle between the current state difference vector and the previous state difference vector.

[0022] The trend reversal frequency factor is the number of times the sign of the derivative of each dimension of the state difference vector changes within the time window w.

[0023] As a preferred embodiment of the insurance product development and analysis method based on big data analysis described in this invention, the construction of the state difference sequence includes, based on the structural fingerprint vector H state (t) and the trend factor vector q(t) constitute a joint input vector x(t) = [H state [(t),q(t)];

[0024] The establishment of the credibility function includes mapping the joint input vector to a high-dimensional feature space through a kernel function to construct a kernel envelope credibility region; the kernel function is a radial basis function kernel function.

[0025] As a preferred embodiment of the insurance product development and analysis method based on big data analysis described in this invention, the establishment of the credibility function further includes constructing an envelope B using historical health samples within the high-dimensional feature space. normal and with B normal The center vector μ B Using the reference as a benchmark, calculate the offset distance in the kernel space; and normalize the offset distance to construct the confidence function SCE(t).

[0026] As a preferred embodiment of the insurance product development and analysis method based on big data analysis described in this invention, the trend vector Z(t) = [η1(t), η2(t)] includes the credibility change rate feature and the credibility fluctuation frequency feature.

[0027] The credibility change rate feature η1(t) is the first-order change rate of the credibility function of the electrical equipment per unit time;

[0028] The reliability fluctuation frequency feature η2(t) represents the degree of frequent fluctuation within the time window w.

[0029] As a preferred embodiment of the insurance product development and analysis method based on big data analysis described in this invention, the consistency determination of the trend vector includes symbolically encoding the credibility change rate feature and the credibility fluctuation frequency feature as ρ(η). i (t)), and calculate the trend consistency index Ω(t);

[0030] When Ω(t) = 2, it indicates that all trend indicators are positive and the equipment condition is highly stable;

[0031] When Ω(t) = -2, it indicates that the equipment condition has seriously deviated from the trend and has consistently deteriorated;

[0032] When Ω(t) = 0, it indicates that there is a conflict between trend indicators and the trend of equipment status is unclear;

[0033] When Ω(t)≥1, the current equipment status trend is consistently positive, the equipment operation is highly reliable, and the insurance strategy adjustment is not triggered.

[0034] When -1≤Ω(t)<1, the current trend of the device is unclear, triggering a fine-tuning of the insurance strategy, pushing device operation suggestions to the user and reminding the user to check the device in advance to reduce the risk.

[0035] When Ω(t) < -1, it indicates that the current equipment trend has deteriorated significantly, immediately triggering the insurance strategy and coordinating maintenance service resources to intervene in advance, reducing the risk of actual equipment failure and subsequent losses.

[0036] An insurance product development and analysis system based on big data analytics, wherein:

[0037] The data module collects real-time operating status data of electrical equipment and establishes actual status sequences; it also trains a digital twin model based on historical data to generate predicted status sequences.

[0038] The difference sequence module compares the actual state sequence with the predicted state sequence to construct a state difference sequence and extracts structural fingerprints and trend factors.

[0039] The trusted function module uses the structural fingerprint and the trend factor to form a joint vector to establish a trusted function.

[0040] The strategy module generates trend vectors based on the trend characteristics of the credibility function, judges the consistency of the trend vectors, and dynamically adjusts the insurance strategy for electrical appliances.

[0041] A computer device includes: a memory and a processor; the memory stores a computer program, characterized in that: when the processor executes the computer program, it implements the steps of the method described in any one of the present invention.

[0042] A computer-readable storage medium having a computer program stored thereon, characterized in that: when the computer program is executed by a processor, it implements the steps of the method described in any one of the present invention.

[0043] The beneficial effects of this invention are as follows: The insurance product development analysis method based on big data analysis provided by this invention achieves real-time offset identification of the operating status of electrical equipment by constructing a dual-track comparison system containing actual state sequences and predicted state sequences; combined with structural fingerprints and trend factor extraction, it accurately characterizes the offset morphology and evolution direction. Furthermore, based on joint feature vectors, kernel space mapping and envelope modeling are introduced to construct a reliable function with disturbance compensation and trend adjustment capabilities, effectively improving the accuracy of anomaly identification. Finally, by evaluating the consistency of trends of multiple indicators through trend synergy matrices and dynamic adjustment factors, multi-level dynamic response and adaptive linkage of insurance strategies are achieved, forming a data-driven extended insurance strategy evolution closed loop, significantly improving the intelligence level and forward-looking service capabilities of the protection strategy. Attached Figure Description

[0044] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] Figure 1 The first embodiment of the present invention provides an overall flowchart of an insurance product development and analysis method based on big data analysis. Detailed Implementation

[0046] To make the above-mentioned objects, features, and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the protection scope of the present invention.

[0047] Example 1, referring to Figure 1 As an embodiment of the present invention, an insurance product development analysis method based on big data analysis is provided, comprising:

[0048] S1: Collect real-time operating status data of electrical equipment and establish an actual status sequence. Train a digital twin model based on historical data to generate a predicted status sequence.

[0049] The following data acquisition devices should be deployed on the insured electrical appliances and control panels:

[0050] Current sensor, voltage sensor, power meter. Temperature and humidity sensor. Displacement sensor (to detect movement). System control interface (to collect start / stop status and operation frequency). Network interface or IoT gateway module (for data transmission).

[0051] The real-time operating status data includes the voltage U at time t. t Current I t Power P t Start-stop frequency F t Temperature T t Environmental disturbances E t and user operation A t .

[0052] Voltage U is collected every 10 seconds. t Current I t Power P t Temperature T is collected every minute. t and environmental disturbances E t Start-stop frequency is calculated hourly. User operations are obtained from real-time operation logs.

[0053] The real-time operating status data is integrated into an actual state vector. Forming the actual state sequence S real (t):

[0054]

[0055] Wherein, the actual state vector of the electrical equipment at time t is represented. The structure is as follows:

[0056]

[0057] The historical data includes historical operating status data of electrical equipment of the same model; operating data of equipment known to be in "healthy condition"; and the complete lifecycle sequence of equipment labeled "normal and not faulty".

[0058] The preprocessing of the historical data includes removing fault points and extreme anomalies, aligning time, and standardizing the various indicators.

[0059] A multi-dimensional state vector for each time slice is extracted from the preprocessed historical data, including power, voltage, current, start / stop frequency, temperature, environmental disturbance values, and user operation behavior data, and a sliding window feature sequence is constructed. For any time point t, the state values ​​of its previous k time steps are used as the input feature vector X. t Predict the state value Y at the next time step t+1. t That is, construct the following regression objective:

[0060]

[0061] Using GBDT combined with X within the time window t and Y tTraining is performed to fit the state evolution of the device over continuous time to obtain the predicted state sequence.

[0062] Wherein, represents the predicted state vector of the electrical equipment at time t.

[0063]

[0064] in, This represents the device voltage value predicted by the digital twin model at time t. This represents the device operating current value predicted by the digital twin model at time t. This represents the device power consumption value predicted by the digital twin model at time t. This represents the device start-up and shutdown frequency (number of starts and stops per unit time) predicted by the digital twin model at time t. This represents the surface or core temperature value of the device predicted by the digital twin model at time t. This indicates the predicted intensity of disturbances (such as humidity or electromagnetic fluctuations) in the environment where the equipment is located. This represents the predicted user behavior characteristics (such as the number of times a user frequently controls a device, or abnormal control behavior).

[0065] Furthermore, by deploying multiple types of sensors in the electrical appliance and control module, high-frequency acquisition of multi-dimensional states such as voltage, current, power, and temperature is achieved. This data, combined with user behavior and environmental disturbance information, constructs an actual state sequence. Simultaneously, a digital twin model is trained using healthy samples to construct a predicted state evolution path, forming a basis for comparing real-time operation with the ideal trajectory. This provides data support for subsequent offset identification and trend modeling, overcoming the limitation of traditional extended warranty strategies in reflecting the individualized state evolution of equipment.

[0066] S2: Compare the actual state sequence with the predicted state sequence to construct a state difference sequence, and extract structural fingerprints and trend factors.

[0067] By comparing the actual state vector and the predicted state vector at the same moment, the difference vector Δ is calculated. t :

[0068]

[0069] Over the entire time span, a state difference sequence Δ(t) is formed:

[0070] Δ(t)={Δ1,Δ2,...,Δ n}

[0071] Where, Δ i This represents the state difference vector at the i-th time point.

[0072] The state difference sequence Δ(t) reflects the degree of deviation between the device and the digital twin predicted trajectory at each moment, and is a prerequisite for state stability and failure risk.

[0073] Based on the state difference sequence Δ(t), the structural fingerprint describing the state shift pattern of electrical appliances is extracted, including: encoding difference fluctuation symbols and identifying structural fragment patterns.

[0074] The encoded difference fluctuation symbols include, for a state difference sequence Δ(t), encoding the changing trend of each dimension of state difference value within any time window w.

[0075] The encoded difference fluctuation symbol includes, within each time window w, the difference value at time i for the j-th state index dimension. The changing trend is represented by symbolic encoding as follows:

[0076]

[0077] in, This represents the difference symbol encoding value under the j-th state index dimension at time i.

[0078] Obtain the symbol sequence within the time window w:

[0079]

[0080] Where d represents the dimension of the state vector.

[0081] The identification of structural fragment patterns includes extracting the symbol sequence into wave structure fragments in chronological order and constructing a change pattern vector h(t), expressed by the formula:

[0082] h(t) = [h1(t), h2(t), ..., h k (t)]

[0083] Among them, h i (t) represents the i-th wave structure characteristic factor, including symbol duration, period span, and mutation frequency. k represents the number of structural fingerprint dimensions.

[0084] The variation pattern vector within each time window w is encoded as a structural fingerprint vector, defined as:

[0085] H state (t)=[h1(t),h2(t),...,h k (t)]

[0086] Among them, H state (t) represents the structural fingerprint vector at time t.

[0087] The trend factors include the offset rate factor, the offset direction angle factor, and the trend reversal frequency factor.

[0088] The offset rate factor is the first derivative of the state difference vector, which characterizes the rate of change of each dimension of the state per unit time.

[0089] The offset direction angle factor is the angle between the current state difference vector and the previous state difference vector, which measures the degree of change in the state evolution direction.

[0090] The trend reversal frequency factor is the number of times the sign of the derivatives of each dimension of the state difference vector changes within a time window w. It characterizes whether the system is in an unstable state.

[0091] The formula for the trend factor q(t) is as follows:

[0092] q(t) = [q1(t), q2(t), q3(t)]

[0093] Where q1(t) is the offset rate factor, q2(t) is the direction angle factor, and q3(t) is the trend reversal frequency factor.

[0094] Furthermore, by calculating the hourly differences between the actual state sequence and the predicted sequence, the offset structural features (structural fingerprint) and evolution direction change indicators (trend factors) of the equipment operation can be extracted. This can accurately characterize the offset type, frequency, and rate, replacing static rules or fixed threshold judgment methods. This enables the equipment state changes to have a quantifiable and traceable semantic representation, providing a more stable and reliable input source for credibility modeling and behavior prediction.

[0095] S3: Use the structural fingerprint and the trend factor to form a joint vector and establish a reliable function.

[0096] The formula for the joint input vector is expressed as:

[0097] x(t)=[H state (t),q(t)]

[0098] Here, x(t) represents the joint state eigenvector at time t.

[0099] To determine whether the current state of the electrical equipment is within a reliable operating trajectory, a kernel space mapping mechanism is used to map the joint input vector to a high-dimensional feature space. The formula is as follows:

[0100] Φ(x(t))=K(x(t),·)

[0101] Where Φ(x(t)) represents the feature representation after mapping the joint state feature vector to the kernel space. K(·,·) represents the kernel function used, which is the radial basis function (RBF kernel), and the formula is expressed as:

[0102]

[0103] Where σ represents the bandwidth parameter of the kernel function, K(x) i ,x j ) represents the sample vector x i With x j Similarity calculation results in kernel space. i x j Let x and y represent any two joint state eigenvectors. i -x j ∥ represents vector x i With x j The Euclidean distance between them. exp(·) denotes the exponential function.

[0104] In a high-dimensional kernel space, a kernel envelope is constructed based on historical health state samples. Defining the confidence domain of electrical equipment under normal operating conditions, in order to center point μ B For reference, the distance of the current state in the kernel space is defined as:

[0105]

[0106] Where, μ B This indicates that Φ(x) is obtained through all historical health sample points. i The center vector is obtained by finding d. K (t) represents the reliable offset distance of the current state in the kernel space at time t.

[0107] The formula for the confidence function is defined as follows:

[0108]

[0109] Where SCE(t)∈[0,1] represents the reliability of the current operating state of the device. K τ(t) represents the offset distance of the current state in the kernel space. τ(t) is a dynamic normalization factor, representing the fluctuation range of the current kernel distance, constructed from the historical distance mean and standard deviation.

[0110]

[0111] Where τ(t) represents the dynamic tolerance factor used to standardize the current kernel distance. KStd(d) represents the average historical kernel distance, and λ represents the standard deviation adjustment coefficient, used to control the weight of the standard deviation term in the tolerance calculation. K ) represents the standard deviation of historical nuclear distance.

[0112] ε(t) represents the structural fingerprint H state (t) The degree of perturbation of the structural fingerprint within a preset time window is expressed by the formula:

[0113] ε(t)=λ1·Var(H state (t))+λ2·ν(t)

[0114] Among them, Var(H state ν(t) represents the variance of each dimension of the resulting fingerprint, ν(t) represents the frequency of alternation of state difference symbols (the number of times the symbol changes from +1 to -1 or -1 to +1), and λ1 and λ2 represent the adjustment coefficients.

[0115] ψ(t) reflects the positive / negative guiding effect of the trend factor q(t) on credibility, and is defined as:

[0116]

[0117] in, The angle factor represents the trend direction; the larger the angle, the more unstable the trend. α represents the trend sensitivity adjustment parameter.

[0118] Furthermore, by combining structural fingerprints and trend factors to form a joint state feature vector and introducing a kernel space mapping mechanism, a reliable envelope structure based on historical health state samples is established. A kernel envelope reliability function with disturbance compensation and trend adjustment capabilities is constructed, which can effectively improve the accuracy and sensitivity of equipment state anomaly identification, avoid the misjudgment and false alarm problems of traditional single-point threshold-based methods, and realize a more refined state reliability assessment mechanism.

[0119] S4: Generate trend vectors based on trend features of the credibility function, determine the consistency of trend vectors, and dynamically adjust the insurance strategy for electrical appliances.

[0120] The trend vector includes the confidence change rate feature and the confidence fluctuation frequency feature.

[0121] The reliability change rate feature η1(t) is the first-order change rate of the electrical equipment reliability function per unit time:

[0122]

[0123] Where SCE(t) represents the confidence function value at the current time step, and SCE(t-1) represents the confidence function value at the previous time step, δ t Indicates a time interval.

[0124] The reliability fluctuation frequency feature η2(t) represents the degree of frequent fluctuation within the time window w.

[0125]

[0126] in, The parentheses represent the indicator function, with a value of 1 when the condition within the parentheses is true, and 0 otherwise. γ represents the set threshold for significant fluctuations. SCE(i) represents the confidence function value at time i. SCE(i+1) represents the confidence function value at time i+1.

[0127] The trend vector formula is expressed as follows:

[0128] Z(t) = [η1(t), η2(t)]

[0129] Where Z(t) represents the trend feature vector at time t.

[0130] The credibility change rate feature and the credibility fluctuation frequency feature are respectively symbolically encoded as ρ(η) i (t) is defined as:

[0131] Further calculation of the trend consistency index:

[0132]

[0133] in:

[0134] Ω(t)∈[-2,+2];

[0135] When Ω(t) = 2, it indicates that the trend indicators are all positive and the equipment status is highly stable.

[0136] When Ω(t) = -2, it indicates that the equipment condition has deviated significantly from the trend and is consistently deteriorating.

[0137] When Ω(t) = 0, it indicates that there is a conflict between trend indicators and the trend of equipment status is unclear.

[0138] When Ω(t)≥1, the current equipment status trend is consistently positive, the equipment operation is highly reliable, and the insurance strategy adjustment is not triggered.

[0139] When -1≤Ω(t)<1, the current device trend is in an unclear state, triggering a fine-tuning of the insurance strategy, pushing device operation suggestions to the user and reminding the user to check the device in advance to reduce the risk.

[0140] When Ω(t) < -1, it indicates that the current equipment trend has deteriorated significantly, immediately triggering the insurance strategy and coordinating maintenance service resources to intervene in advance, reducing the risk of actual equipment failure and subsequent losses.

[0141] Furthermore, by extracting multidimensional trend features (rate, fluctuation frequency, directional stability) from the reliability function, a trend vector is constructed. Based on the inter-trend synergy matrix and disturbance adjustment factors, a comprehensive judgment is made on whether the equipment operation trend is stable and consistent, forming a multi-level response mechanism. Micro-insurance, push notifications, or silent control can be automatically triggered based on trend consistency, realizing a closed-loop link from equipment status perception and risk assessment to strategy feedback, thereby improving the proactiveness and intelligence of insurance services.

[0142] Example 2, an embodiment of the present invention, provides an insurance product development and analysis system based on big data analysis, comprising:

[0143] The data module collects real-time operating status data of electrical equipment and establishes actual status sequences. Based on historical data, it trains a digital twin model to generate predicted status sequences.

[0144] The difference sequence module compares the actual state sequence with the predicted state sequence to construct a state difference sequence and extracts structural fingerprints and trend factors.

[0145] The trusted function module uses the structural fingerprint and the trend factor to form a joint vector to establish a trusted function.

[0146] The strategy module generates trend vectors based on the trend characteristics of the credibility function, judges the consistency of the trend vectors, and dynamically adjusts the insurance strategy for electrical appliances.

[0147] Example 3, an embodiment of the present invention, differs from the previous two embodiments in that:

[0148] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0149] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as a ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (such as a computer-based system, a processor-included system, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device).

[0150] "Computer-readable medium" can be any means that can contain, store, communicate, propagate or transmit programs for use by or in conjunction with an instruction execution system, apparatus or device.

[0151] More specific examples of computer-readable media (a non-exhaustive list) include: electrical connections (electronic devices) having one or more wires, portable computer disk drives (magnetic devices), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, computer-readable media can even be paper or other suitable media on which the program can be printed, because the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in computer memory.

[0152] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0153] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for developing and analyzing insurance products based on big data analytics, characterized in that, include: Collect real-time operating status data of electrical equipment and establish actual status sequences; A digital twin model is trained based on historical data to generate a predicted state sequence; The actual state sequence is compared with the predicted state sequence to construct a state difference sequence, and structural fingerprints and trend factors are extracted. A reliable function is established by using the structural fingerprint and the trend factor to form a joint vector; Trend vectors are generated based on the trend features of the credibility function, the consistency of the trend vectors is judged, and the insurance strategy for electrical appliances is dynamically adjusted. The construction of the state difference sequence includes comparing the actual state vector and the predicted state vector at the same moment, and calculating the difference vector. During the time window Within this sequence, the difference vectors Δt at each time point are combined in chronological order to form the state difference sequence. ; The structural fingerprint includes encoding differential fluctuation symbols and identifying structural fragment patterns; The encoded difference fluctuation symbol includes, in each time window Inside, for the first The first state indicator dimension, the Difference value at time The changing trend is used to encode symbols and obtain symbol sequences. ; The identification of structural fragment patterns includes extracting the symbol sequence into wave structure fragments according to time sequence and constructing a change pattern vector. ; for each time window The variation pattern vector within is encoded as a structural fingerprint vector; The trend factors include the offset rate factor, the offset direction angle factor, and the trend reversal frequency factor. The offset rate factor is the first derivative of the state difference vector; The offset direction angle factor is the angle between the current state difference vector and the previous state difference vector. The trend reversal frequency factor is the number of times the sign of the derivative of each dimension of the state difference vector changes within the time window W. The trend vector This includes the credibility change rate characteristic and the credibility fluctuation frequency characteristic; The reliability change rate feature Let be the first-order rate of change of the reliability function of the electrical equipment per unit time; The credibility fluctuation frequency characteristic The degree of frequent fluctuation within the time window W.

2. The insurance product development and analysis method based on big data analysis as described in claim 1, characterized in that: The real-time operating status data includes voltage, current, power, start / stop frequency, temperature, environmental disturbances, and user operations. The real-time operating status data is integrated into an actual state vector at time t. to form the actual state sequence .

3. The insurance product development and analysis method based on big data analysis as described in claim 2, characterized in that: The constructed state difference sequence includes those based on structural fingerprint vectors. With trend factor vector Construct joint input vector ; The establishment of the credibility function includes mapping the joint input vector to a high-dimensional feature space through a kernel function to construct a kernel envelope credibility region; the kernel function is a radial basis function kernel function.

4. The insurance product development and analysis method based on big data analysis as described in claim 3, characterized in that: The establishment of the credibility function also includes constructing an envelope using historical health samples within the high-dimensional feature space. and with center vector Using this as a benchmark, calculate the offset distance in the kernel space; and normalize the offset distance to construct a confidence function. .

5. The insurance product development and analysis method based on big data analysis as described in claim 4, characterized in that: Determining the consistency of the trend vector includes symbolically encoding the confidence rate of change feature and the confidence fluctuation frequency feature, respectively. And calculate the trend consistency index. ; when This indicates that all trend indicators are positive and the equipment status is highly stable. when This indicates that the equipment status has deviated significantly from the trend and has consistently worsened; when This indicates a conflict between trend indicators, and the equipment status trend is unclear. when If the current equipment status trend is consistently positive, the equipment operation is highly reliable, and no insurance strategy adjustment will be triggered; when If the current device trend is unclear, the insurance strategy will be fine-tuned, pushing device operation suggestions to the user and reminding the user to check the device in order to reduce the risk in advance. when This indicates a significant deterioration in the current equipment trend, immediately triggering insurance strategies and coordinating with maintenance service resources to intervene in advance, reducing the risk of actual equipment failure and subsequent losses.

6. An insurance product development and analysis system based on big data analysis, employing the method described in any one of claims 1-5, characterized in that: The data module collects real-time operating status data of electrical equipment and establishes actual status sequences; it also trains a digital twin model based on historical data to generate predicted status sequences. The difference sequence module compares the actual state sequence with the predicted state sequence to construct a state difference sequence and extracts structural fingerprints and trend factors. The trusted function module uses the structural fingerprint and the trend factor to form a joint vector to establish a trusted function. The strategy module generates trend vectors based on the trend characteristics of the credibility function, judges the consistency of the trend vectors, and dynamically adjusts the insurance strategy for electrical appliances.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the insurance product development and analysis method based on big data analysis as described in any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the insurance product development and analysis method based on big data analysis as described in any one of claims 1 to 5.

Images