Fault evaluation method and device, storage medium and power utilization device

Abstract

The application provides a fault evaluation method and device, a storage medium and a power utilization equipment. The method comprises the following steps: obtaining operation data of an object to be evaluated; performing feature extraction on the operation data to obtain an evaluation feature value related to a fault of the object to be evaluated; inputting the evaluation feature value into a fault evaluation model, performing fault evaluation according to the evaluation feature value and a first feature parameter combination of the fault evaluation model, and outputting a fault evaluation result; wherein the fault evaluation model is constructed based on a fault evaluation initial model and the first feature parameter combination, the first feature parameter combination is calibrated based on an objective function, and the objective function is obtained based on at least one business function matched with at least one business requirement and a target weight combination. Since the fault evaluation model is related to the business requirement, the actual application scene can be adapted, and the accuracy of the fault evaluation result is improved.

Description

Technical Field

[0001] The present application relates to the field of fault detection technology, and in particular to a fault assessment method, device, storage medium, and electrical device. Background Art

[0002] Before a battery fails, various insulation and temperature characteristics will show certain changes, which can be detected and warned in advance. However, there are many types of battery failure detection models, and although they can all achieve battery failure detection, the detection effect is poor. Summary of the Invention

[0003] In view of the above problems, the present application provides a fault assessment method, device, storage medium and electrical device, which determines the objective function based on the business function, and calibrates the characteristic parameter combination of the fault assessment model based on the objective function. Since the fault assessment model is related to business needs, when fault assessment is performed through the fault assessment model, it can adapt to actual application scenarios, thereby improving the accuracy of the fault assessment results and enhancing the detection effect.

[0004] In a first aspect, the present application provides a fault assessment method, comprising: obtaining operating data of an object to be evaluated; performing feature extraction on the operating data to obtain a feature value to be evaluated related to the fault of the object to be evaluated; inputting the feature value to be evaluated into a fault assessment model, performing a fault assessment based on a combination of the feature value to be evaluated and a first feature parameter of the fault assessment model, and outputting a fault assessment result; wherein the fault assessment model is constructed based on an initial fault assessment model and a combination of the first feature parameter, the first feature parameter combination is calibrated based on an objective function, and the objective function is a combination of at least one business function matching at least one business requirement and a target weight.

[0005] In the technical solution of the embodiment of the present application, the fault assessment model is constructed based on the fault assessment initial model and the first characteristic parameter combination, the first characteristic parameter combination is calibrated based on the objective function, and the objective function is based on at least one business function matching at least one business requirement and the target weight combination. Since the fault assessment model is related to the business requirement, when fault assessment is performed through the fault assessment model, it can adapt to the actual application scenario, thereby improving the accuracy of the fault assessment results and enhancing the detection effect.

[0006] In some embodiments, the method for determining the first characteristic parameter combination includes: determining at least one second characteristic parameter combination for updating the initial fault assessment model to obtain at least one updated model corresponding to the at least one second characteristic parameter combination; inputting sample characteristic values ​​of at least one sample object into each of the at least one updated model; wherein the sample characteristic values ​​are obtained by extracting features from the operating data of the sample object; performing a fault assessment based on the sample characteristic values ​​of the at least one sample object and the second characteristic parameter combination, determining a target value of an objective function to obtain at least one target value; and determining the first characteristic parameter combination based on the at least one target value. In this manner, calibration based on the objective function can obtain the first characteristic parameter combination of the fault assessment model.

[0007] In some embodiments, determining the first characteristic parameter combination based on at least one target value includes: performing curve fitting based on at least one second characteristic parameter combination and at least one target value to obtain a fitted curve; and obtaining a solution to the fitted curve to obtain the first characteristic parameter combination. In this manner, the first characteristic parameter combination can be quickly obtained using a curve fitting approach.

[0008] In some embodiments, determining the first characteristic parameter combination based on at least one target value includes: using the second characteristic parameter combination corresponding to the target value that meets a preset condition in the at least one target value as the first characteristic parameter combination. In this way, the first characteristic parameter combination can be quickly obtained using a data comparison method.

[0009] In some embodiments, the method for determining the target weight combination includes: determining at least one candidate weight combination; matching different candidate weights in the candidate weight combination with different business needs; determining at least one candidate function based on at least one business function and each candidate weight combination in at least one candidate weight combination; training the initial fault assessment model based on the sample feature value of at least one sample object and each candidate function in at least one candidate function to obtain a third feature parameter combination of the initial fault assessment model; wherein the sample feature value is obtained by feature extraction of the operating data of the sample object; and taking the candidate weight combination corresponding to the third feature parameter combination that meets the preset conditions in at least one third feature parameter combination as the target weight combination. Since different candidate weights in the candidate weight combination match different business needs, the target weight combination also matches the business function, so that the target weight combination adapts to the actual application scenario, further improving the accuracy of the evaluation result of the fault assessment model; at the same time, by training the initial fault assessment model, the third feature parameter combination can be optimized, so that the target weight combination determined based on the optimal third feature parameter combination has a higher accuracy.

[0010] In some embodiments, a candidate weight combination corresponding to a third feature parameter combination that satisfies a preset condition in at least one third feature parameter combination is used as a target weight combination, including: if each third feature parameter in the third feature parameter combination is within a corresponding feature parameter range, then the candidate weight combination corresponding to the third feature parameter combination is used as the target weight combination. In this way, the rationality of the third feature parameter combination is ensured, and thus the rationality of the target weight combination is ensured.

[0011] In some embodiments, the objective function is used to replace the loss function of the initial fault assessment model. This can improve the speed and accuracy of fault assessment model construction and better adapt it to actual application scenarios.

[0012] In some embodiments, determining an initial fault assessment model includes: determining at least one candidate model; determining a Pareto optimal surface for each of the at least one candidate model based on at least one business function; and selecting the candidate model corresponding to the Pareto optimal surface that satisfies a preset condition as the initial fault assessment model. In this manner, a corresponding candidate model can be selected from multiple candidate models based on business needs to serve as the initial fault assessment model, thereby making the constructed fault assessment model more suitable for actual application scenarios.

[0013] In a second aspect, the present application provides a computer-readable storage medium having a program stored thereon, which implements the aforementioned fault assessment method when executed by a processor.

[0014] In the technical solution of the embodiment of the present application, a computer-readable storage medium executes a program to perform fault assessment on an object to be assessed through a fault assessment model. The fault assessment model is constructed based on an initial fault assessment model and a first characteristic parameter combination. The first characteristic parameter combination is calibrated based on an objective function. The objective function is based on a combination of at least one business function and a target weight that matches at least one business requirement. Since the fault assessment model is related to the business requirement, when performing fault assessment through the fault assessment model, it can adapt to actual application scenarios, thereby improving the accuracy of the fault assessment results and enhancing the detection effect.

[0015] In a third aspect, the present application provides an electrical device, comprising: a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the aforementioned fault assessment method is implemented.

[0016] In the technical solution of the embodiment of the present application, when the electrical equipment executes a program to perform fault assessment on the object to be assessed through a fault assessment model, the fault assessment model is constructed based on the fault assessment initial model and a first characteristic parameter combination, the first characteristic parameter combination is calibrated based on the objective function, and the objective function is based on at least one business function matching at least one business requirement and a target weight combination. Since the fault assessment model is related to the business requirement, when performing fault assessment through the fault assessment model, it can adapt to actual application scenarios, thereby improving the accuracy of the fault assessment results and enhancing the detection effect.

[0017] In a fourth aspect, the present application provides a fault assessment device, which includes: an acquisition module for acquiring operating data of an object to be evaluated; an extraction module for performing feature extraction on the operating data to obtain a characteristic value to be evaluated related to the fault of the object to be evaluated; an analysis module for inputting the characteristic value to be evaluated into a fault assessment model, performing fault assessment based on a combination of the characteristic value to be evaluated and a first characteristic parameter of the fault assessment model, and outputting a fault assessment result; wherein the fault assessment model is constructed based on a fault assessment initial model and a combination of the first characteristic parameter, the first characteristic parameter combination is calibrated based on an objective function, and the objective function is based on a combination of at least one business function and a target weight that matches at least one business requirement.

[0018] In the technical solution of the embodiment of the present application, the fault assessment model is constructed based on the fault assessment initial model and the first characteristic parameter combination, the first characteristic parameter combination is calibrated based on the objective function, and the objective function is based on at least one business function matching at least one business requirement and the target weight combination. Since the fault assessment model is related to the business requirement, when fault assessment is performed through the fault assessment model, it can adapt to the actual application scenario, thereby improving the accuracy of the fault assessment results and enhancing the detection effect.

[0019] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present application. The same reference numerals are used throughout the drawings to represent the same components. In the drawings:

[0021] Figure 1This is a schematic structural diagram of an electrical device according to some embodiments of the present application;

[0022] Figure 2 This is a flowchart of a fault assessment method according to some embodiments of the present application;

[0023] Figure 3 This is a schematic diagram of a process for obtaining a first characteristic parameter combination in some embodiments of the present application;

[0024] Figure 4 This is a schematic diagram of the process of obtaining target weight combinations in some embodiments of the present application;

[0025] Figure 5 A schematic diagram of a process for obtaining an initial model for fault assessment in some embodiments of the present application;

[0026] Figure 6 A schematic diagram of the construction process of a fault assessment model in some embodiments of the present application;

[0027] Figure 7 This is a schematic structural diagram of a fault assessment device according to some embodiments of the present application. DETAILED DESCRIPTION

[0028] The following embodiments of the technical solution of the present application will be described in detail with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present application and are therefore only examples and are not intended to limit the scope of protection of the present application.

[0029] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above-mentioned figure descriptions are intended to cover non-exclusive inclusions.

[0030] In the description of the embodiments of this application, the technical terms "first" and "second" are used only to distinguish different objects and should not be understood to indicate or imply relative importance or implicitly specify the quantity, specific order, or primary and secondary relationship of the indicated technical features. In the description of the embodiments of this application, the meaning of "plurality" is more than two, unless otherwise clearly and specifically defined.

[0031] References herein to "embodiments" mean that a particular feature, structure, or characteristic described in connection with the embodiments may be included in at least one embodiment of the present application. The appearance of this phrase in various places in the specification does not necessarily refer to the same embodiment, nor does it constitute an independent or alternative embodiment that is mutually exclusive of other embodiments. It is understood, both explicitly and implicitly, by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0032] In the description of the embodiments of this application, the term "and / or" is simply a description of the association relationship between associated objects, indicating that three relationships can exist. For example, A and / or B can represent the following three situations: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this document generally indicates that the associated objects are in an "or" relationship.

[0033] In the description of the embodiments of the present application, the term "multiple" refers to more than two (including two). Similarly, "multiple groups" refers to more than two groups (including two groups), and "multiple pieces" refers to more than two pieces (including two pieces).

[0034] Currently, market developments indicate that batteries are increasingly being used. They are not only used in energy storage systems such as hydropower, thermal, wind, and solar power plants, but are also widely used in electric vehicles like electric bicycles, electric motorcycles, and electric cars, as well as in military equipment and aerospace. As battery applications continue to expand, market demand is also growing.

[0035] The inventors of the present application have noticed that during the use of a battery, before an insulation failure occurs in the battery, the various insulation value characteristics, temperature difference characteristics, etc. of the battery will show certain changes, which can be discovered and warned in advance, but different parameter thresholds and models have different warning effects, and different parameter thresholds and models have their own advantages and disadvantages, so it is difficult to select appropriate parameter thresholds and models; and when performing battery insulation fault warning processing in actual business scenarios, there are various business requirements, such as wanting specific faults to be monitored as fully as possible, wanting a lower false alarm rate and as low a processing cost as possible, etc.

[0036] However, current fault detection models generally use a commonly used loss function, and then train based on a training set to obtain a converged model. For example, if a model is needed to detect battery insulation faults, battery data with and without insulation faults will be used as training sets for training. Based on the different training sets, the detection effect of the model after training to convergence is also different, which makes it difficult to adapt to different business needs.

[0037] Therefore, in order to adapt to different business needs, the inventors have found that the objective function can be determined based on different business needs, and the characteristic parameter combination of the fault assessment model can be obtained based on the calibration of the objective function. Among them, business needs include but are not limited to the ability to monitor all specific faults as much as possible, lower processing costs, etc. Since the fault assessment model is related to business needs, when performing fault assessment through the fault assessment model, it can adapt to actual application scenarios (i.e., actual business scenarios), thereby improving the accuracy of the fault assessment results and enhancing the detection effect.

[0038] Based on the above considerations, the inventors have conducted in-depth research and designed a fault assessment method, which can obtain an objective function in advance based on the weighted summation of at least one business function and a target weight combination that matches at least one business requirement, and then obtain the first characteristic parameter combination of the fault assessment model based on the objective function calibration, such as replacing the loss function of the initial fault assessment model with the objective function, and obtaining the first characteristic parameter combination of the fault assessment model based on the objective function calibration, and finally constructing a fault assessment model based on the initial fault assessment model and the first characteristic parameter combination; when applied, feature extraction is performed on the operating data of the object to be assessed to obtain a characteristic value to be assessed related to the fault of the object to be assessed, and the characteristic value to be assessed is input into the fault assessment model, and a fault assessment is performed based on the characteristic value to be assessed and the first characteristic parameter combination of the fault assessment model, and a fault assessment result is output. Since the fault assessment model is related to the business requirement, when performing fault assessment through the fault assessment model, it can adapt to the actual application scenario, thereby improving the accuracy of the fault assessment result and enhancing the detection effect.

[0039] The fault assessment method disclosed in the embodiments of the present application can not only be used to perform fault assessment on batteries, such as performing fault assessment on batteries in electrical devices or various energy storage systems, but also can perform fault assessment on components other than batteries in these devices or perform fault assessment on components or devices other than these devices. Electrical devices can be, but are not limited to, mobile phones, tablets, laptops, electric toys, electric tools, battery cars, electric cars, ships, spacecraft, and the like. Among them, electric toys can include fixed or mobile electric toys, such as game consoles, electric car toys, electric ship toys, and electric airplane toys, and the like, and spacecraft can include airplanes, rockets, space shuttles, and spacecraft, and the like.

[0040] For the convenience of description, the following embodiments are described by taking a vehicle as an example of an electrical device according to an embodiment of the present application.

[0041] Please refer to Figure 1 , Figure 1A schematic structural diagram of a vehicle 1000 provided for some embodiments of the present application. The vehicle 1000 may be a fuel vehicle, a gas vehicle or a new energy vehicle. The new energy vehicle may be a pure electric vehicle, a hybrid vehicle or an extended-range vehicle, etc. A battery 100 is provided inside the vehicle 1000. The battery 100 may be provided at the bottom, head or tail of the vehicle 1000. The battery 100 may be used to power the vehicle 1000. For example, the battery 100 may serve as an operating power source for the vehicle 1000. The vehicle 1000 may further include a controller 200 and a motor 300. The controller 200 is used to control the battery 100 to power the motor 300, for example, to meet the power requirements for starting, navigating and driving the vehicle 1000.

[0042] In some embodiments of the present application, the battery 100 can serve not only as an operating power source for the vehicle 1000, but also as a driving power source for the vehicle 1000, replacing or partially replacing fuel or natural gas to provide driving power for the vehicle 1000.

[0043] According to some embodiments of the present application, referring to Figure 2 , fault assessment methods include:

[0044] S201: Obtaining the operating data of the object to be evaluated.

[0045] For ease of explanation, let's use the example of evaluating whether a vehicle's battery has an insulation fault. Accordingly, the vehicle's battery is the object to be evaluated, and the operating data can be real-time data collected during the battery's operation within a preset time period (e.g., 1 minute, 1 hour) prior to the current moment, including but not limited to the battery's voltage, current, state of charge, insulation value, and so on.

[0046] S203: Extracting features from the operating data to obtain feature values ​​to be evaluated that are related to the fault of the object to be evaluated.

[0047] When evaluating whether a vehicle battery has an insulation fault, feature extraction is performed on the operating data. This can involve extracting insulation values, current, and the like from the operating data to obtain characteristic values ​​to be evaluated that are related to the battery insulation fault. For example, the insulation values ​​within a preset time period before the current moment can be extracted, and the insulation values ​​within the preset time period can be statistically analyzed (e.g., taking the average value, variance value, maximum value, minimum value, etc.) to obtain insulation value statistics, which are then used as insulation value characteristic values. For another example, the current within a preset time period before the current moment can be extracted, and the current statistics within the preset time period can be statistically analyzed (e.g., taking the average value, variance value, maximum value, minimum value, etc.) to obtain current statistics, which are then used as current characteristic values.

[0048] S205: Input the characteristic value to be evaluated into the fault assessment model, perform a fault assessment based on the characteristic value to be evaluated and a first characteristic parameter combination of the fault assessment model, and output a fault assessment result. The fault assessment model is constructed based on the initial fault assessment model and the first characteristic parameter combination, and the first characteristic parameter combination is calibrated based on an objective function, and the objective function is based on a combination of at least one business function matching at least one business requirement and a target weight.

[0049] The characteristic value to be evaluated is input into the fault assessment model, and the fault assessment is performed according to the characteristic value to be evaluated and the first characteristic parameter combination of the fault assessment model through the fault assessment model, and the fault assessment result is output. Among them, the first characteristic parameter combination may include one or more first characteristic parameters, and the first characteristic parameter may be, but is not limited to, characteristic thresholds, characteristic weights and other parameters related to battery insulation faults. For example, the first characteristic parameter combination may include an insulation value characteristic threshold or a current characteristic threshold, and may also include an insulation value characteristic threshold, a current characteristic threshold, an insulation value characteristic weight corresponding to the insulation value characteristic threshold, and a current characteristic weight corresponding to the current characteristic threshold. The specific details are not limited here. The fault assessment result may be, but is not limited to, the fault type and confidence of the battery, and whether the battery has a fault.

[0050] For example, the insulation value characteristic value to be evaluated can be input into the fault assessment model, and the fault assessment model is used to perform a fault assessment based on the size relationship between the insulation value characteristic value and the insulation value characteristic threshold, and output an assessment result of whether the battery has an insulation fault. For another example, the insulation value characteristic value and the current characteristic value to be evaluated can be input into the fault assessment model, and the fault assessment model is used to perform a fault assessment based on the size relationship between the insulation value characteristic value and the insulation value characteristic threshold, and the size relationship between the current characteristic value and the current characteristic threshold, and output an assessment result of whether the battery has an insulation fault.

[0051] The fault assessment model can be predetermined. The fault assessment model is constructed based on an initial fault assessment model and a combination of first characteristic parameters. The initial fault assessment model can be, but is not limited to, a machine learning model such as a decision tree model or a regression tree model. The first characteristic parameter combination is calibrated based on an objective function, and the objective function is a weighted summation of at least one business function matching at least one business requirement and a target weight combination.

[0052] At least one business requirement is determined based on the actual application scenario. For example, when evaluating battery insulation faults, at least one business requirement includes, but is not limited to, finding more faulty batteries, fewer false positives for healthy batteries, minimizing processing costs, and having a higher ability to identify specific severe fault types (i.e., ensuring that all specific faults are monitored as fully as possible).

[0053] At least one business function corresponds to at least one business requirement. Different business requirements can construct different business functions. For example, when the business requirement is to find more faulty batteries, the business function that matches it is the recall rate, that is, it is hoped to have a higher recall rate, that is, it is hoped that the number of faulty batteries detected accounts for a higher proportion of the total number of faulty batteries; when the business requirement is to have fewer false positives for normal batteries, the business function that matches it is the false alarm rate, that is, it is hoped to have a lower false alarm rate, that is, it is hoped that the number of normal batteries among the detected faulty batteries accounts for a lower proportion of the total number of detected faulty batteries; when the business requirement is to have the lowest possible processing cost, the business function that matches it is the warning ratio, that is, it is hoped to have a lower warning ratio, that is, it is hoped that the number of faulty batteries that issue warning prompts accounts for a lower proportion of the total number of batteries; when the business requirement is to have a higher recognition ability for specific serious fault types, the business function that matches it is the recognition rate of specific serious fault types, that is, it is hoped to have a higher recognition rate of specific serious fault types, that is, it is hoped that the number of faulty batteries of a specific serious fault type detected accounts for a higher proportion of the total number of faulty batteries with a specific serious fault type.

[0054] The target weight combination includes at least one weight, which corresponds one-to-one to at least one business function. The target weight combination can be determined based on the actual application scenario. For example, if higher battery safety is required, the recall rate and the recognition rate of specific severe fault types can be given higher weights, while other business functions can be given lower weights.

[0055] After obtaining at least one business function and a target weight combination, a weighted sum of at least one business function and at least one weight in the target weight combination may be performed to obtain a target function.

[0056] After obtaining the objective function, the first characteristic parameter combination of the fault assessment model can be obtained based on the objective function calibration. For example, the objective function can be used as the loss function of the initial fault assessment model. Then, based on the historical operating data of the object to be assessed, the initial fault assessment model is used to perform a fault assessment to obtain a fault assessment result. Based on the fault assessment result, the function values ​​of each business function are determined, and the target value of the objective function is obtained by weighted summation of the function values ​​and the target weight combination. When the model parameters of the initial fault assessment model change, the target value of the objective function will also change accordingly. Therefore, based on the target value, the most appropriate model parameters can be determined to obtain the first characteristic parameter combination of the fault assessment model.

[0057] Finally, a fault assessment model is constructed based on the initial fault assessment model and the first characteristic parameter combination. For example, the initial fault assessment model is a decision tree model, and the first characteristic parameter in the first characteristic parameter combination is a parameter on each layer of the model.

[0058] In the above embodiment, the fault assessment model is constructed based on the fault assessment initial model and the first characteristic parameter combination, the first characteristic parameter combination is calibrated based on the objective function, and the objective function is based on at least one business function matching at least one business requirement and the target weight combination. Since the fault assessment model is related to the business requirement, when fault assessment is performed through the fault assessment model, it can adapt to the actual application scenario, thereby improving the accuracy of the fault assessment results and enhancing the detection effect.

[0059] According to some embodiments of the present application, referring to Figure 3 , the method for determining the first characteristic parameter combination includes:

[0060] S301: Determine at least one second characteristic parameter combination for updating an initial fault assessment model to obtain at least one updated model corresponding to the at least one second characteristic parameter combination.

[0061] The second characteristic parameter in the second characteristic parameter combination corresponds one-to-one to the first characteristic parameter in the first characteristic parameter combination. For example, when the first characteristic parameter combination includes an insulation value characteristic threshold, the second characteristic parameter combination also includes the insulation value characteristic threshold; when the first characteristic parameter combination includes an insulation value characteristic threshold, a current characteristic threshold, and corresponding characteristic weights, the second characteristic parameter combination also includes the insulation value characteristic threshold, the current characteristic threshold, and corresponding characteristic weights, so that the first characteristic parameter combination can be determined based on the second characteristic parameter combination.

[0062] For example, characteristic thresholds, characteristic weights and other parameters related to battery insulation failure can be predetermined, including but not limited to insulation value characteristic thresholds, current characteristic thresholds, etc., and corresponding parameter ranges can be determined for each characteristic parameter. Then, based on the parameter range of each characteristic parameter, at least one second characteristic parameter combination can be determined. Assume that the first characteristic parameter combination finally determined is (k1, k2, ..., k a ), then at least one second characteristic parameter combination can be (k 11 ,k 21 ,…,k a1 )、(k 12 ,k 22 ,…,k a2 ),...,(k 1b ,k 2b ,…,k ab ), where a and b are positive integers.

[0063] Then, each second characteristic parameter combination may be input into the fault assessment initial model as a model parameter of the fault assessment initial model, thereby obtaining at least one updated model.

[0064] S303: Inputting a sample feature value of at least one sample object into each of the at least one update model, wherein the sample feature value is obtained by extracting features from the running data of the sample object.

[0065] The sample object is the battery on the sample vehicle. The process of obtaining the sample characteristic value is the same as the process of obtaining the characteristic value to be evaluated. The difference is that the sample characteristic value is derived from the historical operating data of the battery on the sample vehicle.

[0066] S305: Perform fault assessment based on a combination of a sample characteristic value and a second characteristic parameter of at least one sample object, and determine a target value of the target function to obtain at least one target value.

[0067] For each updated model, the sample feature value of at least one sample object can be input into the updated model. Through the updated model, the fault assessment is performed based on the sample feature value of at least one sample object and the combination of the second feature parameter to obtain a fault assessment result, and the target value of the objective function is determined based on the fault assessment result. For example, when the second feature parameter combination is (k 11 ,k 21 ,…,k a1 ), a target value S1 can be determined; when the second characteristic parameter combination is (k 12 ,k 22 ,…,k a2 ), a target value S2 can be determined; ...; when the second characteristic parameter combination is (k 1b ,k 2b ,…,k ab ), a target value S can be determined b .

[0068] S307: Determine a first characteristic parameter combination based on at least one target value.

[0069] For example, the target value S1, the target value S2, ..., the target value S b , determine the first characteristic parameter combination through curve fitting or data comparison.

[0070] In the above embodiment, calibration is performed based on the objective function to obtain the first characteristic parameter combination of the fault assessment model.

[0071] According to some embodiments of the present application, determining a first characteristic parameter combination based on at least one target value includes: performing curve fitting based on at least one second characteristic parameter combination and at least one target value to obtain a fitting curve; obtaining a solution of the fitting curve to obtain the first characteristic parameter combination.

[0072] For example, the second characteristic parameter combination (k11 ,k 21 ,…,k a1 )、(k 12 ,k 22 ,…,k a2 ),...,(k 1b ,k 2b ,…,k ab ) and target values ​​S1, S2, ..., S b Perform curve fitting to obtain the fitting curve f(S i ,k 1i ,k 2i ,…,k ai ), the fitting curve f(S i ,k 1i ,k 2i ,…,k ai ) is the second characteristic parameter in the second characteristic parameter combination (k 1i ,k 2i ,…,k ai ), the dependent variable is the target value S i ; Then, the fitting curve f(S i ,k 1i ,k 2i ,…,k ai ) in each second characteristic parameter (k 1i ,k 2i ,…,k ai ) is derived to obtain a equation, which are Finally, by solving these a equations, we can get the solution of each second characteristic parameter, and these solutions constitute the first characteristic parameter combination.

[0073] In the above embodiment, the first characteristic parameter combination can be quickly obtained by using a curve fitting method.

[0074] According to some embodiments of the present application, determining a first characteristic parameter combination based on at least one target value includes: taking a second characteristic parameter combination corresponding to a target value that meets a preset condition in the at least one target value as the first characteristic parameter combination.

[0075] For ease of explanation, let's take the second characteristic parameter combination including a daily average insulation value threshold as an example. Assume that the daily average insulation value threshold is 1500 ohms, and the corresponding target value of the objective function is 3.9; the daily average insulation value threshold is 2000 ohms, and the corresponding target value of the objective function is 4.02; and the daily average insulation value threshold is 2500 ohms, and the corresponding target value of the objective function is 3.9. Then, the daily average insulation value threshold of 2000 ohms, corresponding to the maximum target value of 4.02, can be used as the optimal threshold to form the first characteristic parameter combination. In this case, the first characteristic parameter combination includes the daily average insulation value threshold of 2000 ohms.

[0076] In the above embodiment, the first characteristic parameter combination can be quickly obtained by using a data comparison method.

[0077] According to some embodiments of the present application, referring to Figure 4 , the target weight combination is determined by:

[0078] S401: Determine at least one candidate weight combination; different candidate weights in the candidate weight combination match different business requirements.

[0079] The target weight combination is one of the at least one candidate weight combination. When the target weight combination is required to match the actual application scenario, each candidate weight combination in the at least one candidate weight combination must also match the actual application scenario. For example, when higher battery safety is required, for each candidate weight combination, the recall rate and specific fault recognition rate can be given higher weights, while other business functions can be given lower weights.

[0080] Assume that the business requirements include business requirement 1, business requirement 2, ..., business requirement n, and the corresponding business functions include business function obj1, business function obj2, ..., business function obj n , at least one candidate weight combination includes a candidate weight combination (W 11 ,W 21 ,...,W n1 ), candidate weight combination (W 12 ,W 22 ,...,W n2 ),..., candidate weight combination (W 1m ,W 2m ,...,W nm ), where n and m are positive integers.

[0081] S403: Determine at least one candidate function based on at least one business function and each candidate weight combination in at least one candidate weight combination.

[0082] For each candidate weight combination, a candidate function is obtained by performing a weighted summation on at least one weight in the candidate weight combination and at least one business function, thereby obtaining at least one candidate function.

[0083] For example, business functions obj1, obj2, ..., obj n Combined with the candidate weights (W 11 ,W 21 ,...,W n1 ) performs weighted summation to obtain the candidate function OBJ1=obj1*W 11+obj2*W 21 +...obj n *W n1 ; Set business functions obj1, obj2, ..., obj n Combined with the candidate weights (W 12 ,W 22 ,...,W n2 ) performs weighted summation to obtain the candidate function OBJ2=obj1*W 12 +obj2*W 22 +...obj n *W n2 ;...; business functions obj1, obj2,..., obj n Combined with the candidate weights (W 1m ,W 2m ,...,W nm ) to perform weighted summation and obtain the candidate function OBJ m =obj1*W 1m +obj2*W 2m +...obj n *W nm .

[0084] S405: Based on the sample feature value of the at least one sample object and each candidate function of the at least one candidate function, the initial fault assessment model is trained to obtain a third feature parameter combination of the initial fault assessment model. The sample feature value is obtained by extracting features from the operating data of the sample object.

[0085] When determining the third characteristic parameter combination of the fault assessment initial model, each candidate function of at least one candidate function can be used as the loss function of the fault assessment initial model, and the fault assessment initial model can be trained until the fault assessment initial model converges, thereby obtaining the optimal characteristic parameter combination of the fault assessment initial model corresponding to each candidate function, which is used as the third characteristic parameter combination of the fault assessment initial model corresponding to each candidate function.

[0086] The third characteristic parameter in the third characteristic parameter combination can be part of the first characteristic parameter in the first characteristic parameter combination. For example, when the first characteristic parameter combination includes an insulation value characteristic threshold, a current characteristic threshold, and their corresponding characteristic weights, the third characteristic parameter combination can include the insulation value characteristic threshold or the current characteristic threshold. This can accelerate the convergence of the model. At the same time, since the third characteristic parameter combination is mainly used to determine the target weight combination, it will not affect the accuracy of the first characteristic parameter combination. When selecting the third characteristic parameter in the third characteristic parameter combination, some easy-to-understand characteristic parameters can be selected, such as the daily insulation value minimum threshold.

[0087] For example, when the third characteristic parameter combination includes a daily insulation value minimum threshold, the daily insulation value minimum threshold can be used as a model parameter of the fault assessment initial model, and accordingly, the sample characteristic value is the daily insulation value minimum. Then, for the candidate function OBJ1, the candidate function OBJ1 is used as the loss function of the fault assessment initial model, and the fault assessment initial model is trained based on at least one daily insulation value minimum until the fault assessment initial model converges, thereby obtaining the optimal daily insulation value minimum threshold of the fault assessment initial model corresponding to the candidate function OBJ1, which is used as the third characteristic parameter in the third characteristic parameter combination of the fault assessment initial model corresponding to the candidate function OBJ1; and so on, the candidate functions OBJ2, ..., and candidate functions OBJ1 can be obtained. m The corresponding third characteristic parameter combination of the initial fault assessment model.

[0088] S407: Taking the candidate weight combination corresponding to the third feature parameter combination that meets the preset condition in at least one third feature parameter combination as the target weight combination.

[0089] After obtaining candidate function OBJ1, candidate function OBJ2, ..., candidate function OBJ m After the third characteristic parameter combination of the corresponding initial fault assessment model is combined, a candidate weight combination corresponding to the third characteristic parameter combination that best conforms to the actual situation can be selected from these third characteristic parameter combinations as the target weight combination. For example, assuming that the third characteristic parameter combination of the initial fault assessment model corresponding to the candidate function OBJ1 best conforms to the actual situation, then the candidate weight combination (W 11 ,W 21 ,...,W n1 ) as the target weight combination, and then based on the target weight combination and at least one business function, the target function can be determined as OBJ1=obj1*W 11 +obj2*W 21 +...obj n *W n1 .

[0090] In the above embodiment, since different candidate weights in the candidate weight combination match different business needs, the target weight combination also matches the business function, so that the target weight combination adapts to the actual application scenario, further improving the accuracy of the evaluation results of the fault assessment model; at the same time, by training the initial fault assessment model, the third feature parameter combination can be optimized, so that the target weight combination determined based on the optimal third feature parameter combination has higher accuracy.

[0091] According to some embodiments of the present application, a candidate weight combination corresponding to a third feature parameter combination that meets preset conditions in at least one third feature parameter combination is used as a target weight combination, including: if each third feature parameter in the third feature parameter combination is within the corresponding feature parameter range, then the candidate weight combination corresponding to the third feature parameter combination is used as the target weight combination.

[0092] A rationality analysis can be performed on each of the at least one third characteristic parameter combination to select a third characteristic parameter combination that best conforms to the actual situation from the at least one third characteristic parameter combination, and then the candidate weight combination corresponding to the third characteristic parameter combination is used as the target weight combination.

[0093] For example, based on business understanding, the characteristic parameter range corresponding to each third characteristic parameter in the third characteristic parameter combination can be determined, and then for each third characteristic parameter combination, it is determined whether each third characteristic parameter in each third characteristic parameter combination is within the corresponding characteristic parameter range. If so, the candidate weight combination corresponding to the characteristic parameter combination is used as the target weight combination. For example, when the third characteristic parameter combination includes a daily insulation value minimum threshold, assuming that the determined daily insulation value minimum thresholds are 300 ohms and 5000 ohms, then based on the range of the daily insulation value minimum threshold, it can be determined that 300 ohms is more reasonable. In this case, the candidate weight combination corresponding to 300 ohms is used as the target weight combination.

[0094] From the perspective of application scenarios, assuming that the third characteristic parameter combination of a candidate weight combination determined based on the sample object detects the most fault samples, and in the current application scenario, the consequences of a fault are more serious, and it is hoped to discover all potential fault samples as much as possible, then the candidate weight combination that can detect the most fault samples is most suitable for this application scenario.

[0095] In the above embodiment, the rationality of the third characteristic parameter combination is ensured, thereby ensuring the rationality of the target weight combination.

[0096] According to some embodiments of the present application, referring to Figure 5 ,The method of determining the initial model for fault assessment includes:

[0097] S501: Determine at least one candidate model.

[0098] The at least one candidate model may be, but is not limited to, a machine learning model such as a decision tree model or a regression tree model.

[0099] S503: Determine a Pareto optimal surface of each candidate model in at least one candidate model based on at least one business function.

[0100] The Pareto optimal surface is composed of the points where all Pareto optimal solutions are mapped in the coordinate system. There can be multiple Pareto optimal solutions. For each Pareto optimal solution, there is no other solution in the variable space that is better than this solution.

[0101] When determining the Pareto optimal surface of each candidate model, a first function may be generated based on at least one business function. The first function is obtained by assigning equal weights to each of the at least one business function and performing a weighted summation. Then, based on the first function, a Pareto optimal surface is generated for each candidate model. The same first function, when applied to different candidate models, generates different Pareto optimal surfaces.

[0102] S505: Using a candidate model corresponding to a Pareto optimal surface that meets a preset condition in at least one Pareto optimal surface as an initial model for fault assessment.

[0103] For example, the position of each Pareto optimal surface in at least one Pareto optimal surface is judged, where, in the same coordinate system, the higher the position of the Pareto optimal surface, the better the Pareto optimal surface, which means that the candidate model corresponding to the Pareto optimal surface is more suitable for the actual application scenario corresponding to at least one business function. Therefore, the candidate model corresponding to the Pareto optimal surface at the top can be used as the initial model for fault assessment.

[0104] In the above embodiment, a corresponding candidate model can be screened out from multiple candidate models based on business needs as an initial fault assessment model, so that the constructed fault assessment model is more suitable for actual application scenarios; and the determined initial fault assessment model can quickly adapt to the replacement of subsequent application scenarios. For example, application scenario 1 is a scenario with a relatively high mean insulation value. Through the above method, it is determined that candidate model 1 is better than candidate model 2 in this application scenario. If a scenario with a relatively high mean insulation value is encountered subsequently, candidate model 1 or a fault assessment model determined based on candidate model 1 can be directly used.

[0105] As an example, see Figure 6 , fault assessment methods may include:

[0106] S601: Based on the actual application scenario, n business requirements are determined, namely business requirement 1, business requirement 2, ..., business requirement n. For example, there are four business requirements: finding more faulty batteries, fewer false positives for normal batteries, minimizing processing costs, and identifying more specific severe fault types.

[0107] S603: Determine n business functions according to n business requirements. The business functions correspond to the business requirements one by one, namely business function obj1, business function obj2, ..., business function objn For example, the four business functions corresponding to the above four business requirements are recall rate, false alarm rate, warning ratio, and recognition rate of specific serious fault types.

[0108] S605: Determine m candidate weight combinations based on the actual application scenario, which are respectively the candidate weight combinations (W 11 ,W 21 ,...,W n1 ), candidate weight combination (W 12 ,W 22 ,...,W n2 ),..., candidate weight combination (W 1m ,W 2m ,...,W nm ).

[0109] S607: Based on the m candidate weight combinations and the n business functions, determine m candidate functions, namely, candidate function OBJ1=obj1*W 11 +obj2*W 21 +...obj n *W n1 ;Candidate function OBJ2=obj1*W 12 +obj2*W 22 +...obj n *W n2 ;...;Candidate function OBJ m =obj1*W 1m +obj2*W 2m +...obj n *W nm .

[0110] S609: Based on n business functions, determine a first function, the first function is AOBJ = obj1*w1+obj2*w2+...obj n *w n Among them, weights w1, w2, ...w n All the same.

[0111] S611: Determine d candidate models, namely candidate model 1, candidate model 2, ..., candidate model d. For example, two candidate models are a decision tree model and a regression model.

[0112] S613: Determine the Pareto optimal surface corresponding to each of the b candidate models, namely Pareto optimal surface 1, Pareto optimal surface 2, ..., Pareto optimal surface b. The Pareto optimal surface with the highest position is determined from the b Pareto optimal surfaces, and the candidate model corresponding to the highest Pareto optimal surface is used as the initial fault assessment model. For example, the initial fault assessment model is a decision tree model.

[0113] S615: Determine the third characteristic parameter combination corresponding to each candidate function in the m candidate functions, which are the third characteristic parameter combinations (k h11 ,k h21 ,…,k hc1 ), the third characteristic parameter combination (k h12 ,k h22 ,…,k hc2 ), ..., the third characteristic parameter combination (k h1m ,k h2m ,…,k hcm For example, a small number of key and understandable characteristic parameters, such as the minimum threshold value of the daily insulation value, can be selected as the model parameters of the initial fault assessment model. Then, for each of the m candidate functions, the candidate function is used as the loss function of the initial fault assessment model. The initial fault assessment model is trained based on the sample characteristic values ​​to obtain the optimal model parameter combination of the initial fault assessment model. The optimal model parameter combination is used as the third characteristic parameter combination of the initial fault assessment model corresponding to the candidate function, thereby obtaining m third characteristic parameter combinations.

[0114] S617: Based on the m third feature parameter combinations, determine the target weight combination. For example, perform a rationality analysis on the m third feature parameter combinations, specifically determine whether the third feature parameters in each third feature parameter combination are within the corresponding parameter range. If so, use the candidate weight combination corresponding to the third feature parameter combination as the target weight combination. For example, assuming the third feature parameter combination (k h11 ,k h21 ,…,k hc1 ) is the most reasonable, then the target weight combination is (W 11 ,W 21 ,...,W n1 ). For example, assuming that the third characteristic parameter combination only includes the daily insulation value minimum threshold, the daily insulation value minimum threshold of the fault assessment initial model corresponding to the candidate function OBJ1 determined by S615 is 300 ohms, and the daily insulation value minimum threshold of the fault assessment initial model corresponding to the candidate function OBJ2 is 5000 ohms. Based on business understanding, the daily insulation value minimum threshold of 300 ohms is more reasonable, so the candidate weight combination corresponding to the candidate function OBJ1 is the target weight combination.

[0115] S619: Determine the target function based on the target weight combination and n business functions. For example, when the target weight combination is (W 11 ,W 21 ,...,W n1 ), the objective function is OBJ1=obj1*W 11 +obj2*W 21 +...obj n *W n1 .

[0116] S621: calibrate the first characteristic parameter combination of the fault assessment model based on the objective function. For example, b second characteristic parameter combinations may be determined first, which are the second characteristic parameter combinations (k 11 ,k 21 ,…,k a1 ), the second characteristic parameter combination (k 12 ,k 22 ,…,k a2 ),..., the second characteristic parameter combination (k 1b ,k 2b ,…,k ab ); then, each of the b second characteristic parameter combinations is used as a model parameter of the initial fault assessment model, and the objective function is used as the loss function of the initial fault assessment model, the sample characteristic value is subjected to fault assessment, and a fault assessment result is obtained. The target value of the objective function is calculated based on the fault assessment result, thereby obtaining b target values, which are target value S1, target value S2, ..., target value S b Then, based on the second characteristic parameter combination (k 11 ,k 21 ,…,k a1 )、(k 12 ,k 22 ,…,k a2 ),...,(k 1b ,k 2b ,…,k ab ) and target values ​​S1, S2, ..., S b Perform curve fitting, and based on the fitting curve, use the derivative method to calculate the optimal solution of each second characteristic parameter, and these optimal solutions constitute the first characteristic parameter combination. Alternatively, for the target values ​​S1, S2, ..., S b A size comparison is performed, and the second characteristic parameter combination corresponding to the largest target value is used as the first characteristic parameter combination.

[0117] S623: Construct a fault assessment model based on the initial fault assessment model and the first feature parameter combination. Assuming that the initial fault assessment model is a decision tree model, the objective function can be used as the loss function of the decision tree model. The importance of each first feature parameter in the first feature parameter combination is then determined based on the target value of the objective function, and each first feature parameter is distributed across different layers based on importance. Each layer uses a greedy algorithm to obtain the current optimal solution, and then the next layer is added until the number of layers reaches a preset number or the target value exceeds a certain value, at which point the decision tree model converges to form the final fault assessment model.

[0118] S625: Perform a fault assessment on the object to be assessed based on the fault assessment model to obtain a fault assessment result. For example, the fault assessment result includes, but is not limited to, the fault type, confidence level, and whether a fault exists. The confidence level indicates whether the fault type is accurate. The higher the confidence level, the more accurate the fault type identification.

[0119] S627: Based on the fault assessment result, an early warning reminder is issued.

[0120] In the above embodiment, business requirements are determined based on the actual application scenario, and corresponding business functions and candidate weight combinations are determined based on the business requirements, and then the fault assessment initial model and the first characteristic parameter combination that are adapted to the actual application scenario are determined based on the business function and the candidate weight combination, so that the fault assessment model constructed based on the fault assessment initial model and the first characteristic parameter combination is more suitable for the actual application scenario, and the fault assessment effect is better.

[0121] According to some embodiments of the present application, a computer-readable storage medium stores a program, and when the program is executed by a processor, the aforementioned fault assessment method is implemented.

[0122] According to some embodiments of the present application, the electrical device includes: a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the aforementioned fault assessment method is implemented.

[0123] According to some embodiments of the present application, referring to Figure 7The fault assessment device 2000 includes an acquisition module 2100, an extraction module 2200, and an analysis module 2300. The acquisition module 2100 is used to acquire operating data of an object to be assessed; the extraction module 2200 is used to perform feature extraction on the operating data to obtain a feature value to be assessed related to the fault of the object to be assessed; and the analysis module 2300 is used to input the feature value to be assessed into a fault assessment model, perform a fault assessment based on a combination of the feature value to be assessed and a first feature parameter of the fault assessment model, and output a fault assessment result. The fault assessment model is constructed based on an initial fault assessment model and a combination of the first feature parameter, the first feature parameter combination being calibrated based on an objective function, and the objective function being a combination of at least one business function matching at least one business requirement and a target weight.

[0124] According to some embodiments of the present application, the analysis module 2300 is used to: determine at least one second characteristic parameter combination for updating the initial model of fault assessment to obtain at least one updated model corresponding to the at least one second characteristic parameter combination; input the sample characteristic value of at least one sample object into each updated model in the at least one updated model; wherein the sample characteristic value is obtained by feature extraction of the operating data of the sample object; perform fault assessment based on the sample characteristic value of at least one sample object and the second characteristic parameter combination, determine the target value of the objective function to obtain at least one target value; and determine the first characteristic parameter combination based on the at least one target value.

[0125] According to some embodiments of the present application, the analysis module 2300 is used to: perform curve fitting based on at least one second characteristic parameter combination and at least one target value to obtain a fitting curve; and obtain a solution of the fitting curve to obtain a first characteristic parameter combination.

[0126] According to some embodiments of the present application, the analysis module 2300 is configured to: use a second characteristic parameter combination corresponding to a target value that meets a preset condition among at least one target value as a first characteristic parameter combination.

[0127] According to some embodiments of the present application, the analysis module 2300 is used to: determine at least one candidate weight combination; different candidate weights in the candidate weight combination match different business needs; determine at least one candidate function based on at least one business function and each candidate weight combination in at least one candidate weight combination; train the fault assessment initial model based on the sample feature value of at least one sample object and each candidate function in at least one candidate function to obtain a third feature parameter combination of the fault assessment initial model; wherein the sample feature value is obtained by feature extraction of the operating data of the sample object; and use the candidate weight combination corresponding to the third feature parameter combination that meets the preset conditions in at least one third feature parameter combination as the target weight combination.

[0128] According to some embodiments of the present application, the analysis module 2300 is used to: if each third feature parameter in the third feature parameter combination is within the corresponding feature parameter range, then the candidate weight combination corresponding to the third feature parameter combination is used as the target weight combination.

[0129] According to some embodiments of the present application, the objective function is used to replace the loss function of the initial fault assessment model.

[0130] According to some embodiments of the present application, analysis module 2300 is configured to: determine at least one candidate model; determine the Pareto optimal surface of each of the at least one candidate model based on at least one business function; and select the candidate model corresponding to the Pareto optimal surface that satisfies a preset condition as the initial fault assessment model. In this way, a corresponding candidate model can be selected from multiple candidate models based on business needs to serve as the initial fault assessment model, making the constructed fault assessment model more suitable for actual application scenarios.

[0131] It should be noted that for the relevant descriptions of the device, storage medium and electrical equipment, please refer to the relevant descriptions of the method, which will not be repeated here.

[0132] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some or all of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present application, and they should all be included in the scope of the claims and specification of the present application. In particular, as long as there is no structural conflict, the various technical features mentioned in the various embodiments can be combined in any way. The present application is not limited to the specific embodiments disclosed herein, but includes all technical solutions that fall within the scope of the claims.

Claims

1. A fault assessment method, characterized in that: The method comprises: Obtaining the operating data of the object to be evaluated; Extracting features from the operating data to obtain feature values ​​related to the fault of the object to be evaluated; The characteristic value to be evaluated is input into a fault assessment model, a fault assessment is performed based on the characteristic value to be evaluated and a first characteristic parameter combination of the fault assessment model, and a fault assessment result is output; wherein the fault assessment model is constructed based on an initial fault assessment model and the first characteristic parameter combination, the first characteristic parameter combination is calibrated based on an objective function, and the objective function is based on at least one business function matching at least one business requirement and a target weight combination.

2. The method according to claim 1, characterized in that The method for determining the first characteristic parameter combination includes: determining at least one second characteristic parameter combination for updating the initial fault assessment model to obtain at least one updated model corresponding to the at least one second characteristic parameter combination; Inputting a sample feature value of at least one sample object into each of the at least one update model; wherein the sample feature value is obtained by extracting features from the operating data of the sample object; Performing a fault assessment based on a combination of a sample characteristic value of the at least one sample object and the second characteristic parameter, and determining a target value of the objective function to obtain at least one target value; The first characteristic parameter combination is determined based on the at least one target value.

3. The method according to claim 2, characterized in that The determining the first characteristic parameter combination based on the at least one target value includes: Performing curve fitting according to the at least one second characteristic parameter combination and the at least one target value to obtain a fitting curve; A solution of the fitting curve is obtained to obtain the first characteristic parameter combination.

4. The method according to claim 2, characterized in that The determining the first characteristic parameter combination based on the at least one target value includes: The second characteristic parameter combination corresponding to the target value that meets the preset condition in the at least one target value is used as the first characteristic parameter combination.

5. The method according to claim 1, wherein The method for determining the target weight combination includes: Determining at least one candidate weight combination; different candidate weights in the candidate weight combination match different business requirements; determining at least one candidate function based on the at least one business function and each candidate weight combination of the at least one candidate weight combination; Training the initial fault assessment model based on sample feature values ​​of at least one sample object and each candidate function of the at least one candidate function to obtain a third feature parameter combination of the initial fault assessment model; wherein the sample feature values ​​are obtained by feature extraction of operating data of the sample object; The candidate weight combination corresponding to the third feature parameter combination that meets the preset conditions in at least one third feature parameter combination is used as the target weight combination.

6. The method according to claim 5, characterized in that The step of taking a candidate weight combination corresponding to a third feature parameter combination that satisfies a preset condition in at least one third feature parameter combination as the target weight combination includes: If each third characteristic parameter in the third characteristic parameter combination is within the corresponding characteristic parameter range, the candidate weight combination corresponding to the third characteristic parameter combination is used as the target weight combination.

7. The method according to any one of claims 1 to 6, characterized in that The objective function is used to replace the loss function of the initial fault assessment model.

8. The method according to any one of claims 1 to 6, characterized in that The method for determining the initial fault assessment model includes: determining at least one candidate model; determining a Pareto optimal surface of each candidate model in the at least one candidate model based on the at least one business function; The candidate model corresponding to the Pareto optimal surface that meets the preset conditions in at least one Pareto optimal surface is used as the initial fault assessment model.

9. A computer-readable storage medium, characterized in that A program is stored thereon, and when the program is executed by a processor, the fault assessment method according to any one of claims 1 to 8 is implemented.

10. An electrical device, characterized in that: include: A memory, a processor, and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the fault assessment method according to any one of claims 1 to 8 is implemented.

11. A fault assessment device, characterized in that: The device comprises: An acquisition module is used to obtain the operating data of the object to be evaluated; An extraction module, configured to extract features from the operating data to obtain feature values ​​to be evaluated that are related to the fault of the object to be evaluated; An analysis module is configured to input the characteristic value to be evaluated into a fault assessment model, perform a fault assessment based on the characteristic value to be evaluated and a first characteristic parameter combination of the fault assessment model, and output a fault assessment result; wherein the fault assessment model is constructed based on an initial fault assessment model and the first characteristic parameter combination, the first characteristic parameter combination is calibrated based on an objective function, and the objective function is based on a combination of at least one business function and a target weight that matches at least one business requirement.

Images