CN-122023021-A - Loss part damage assessment processing method, device and equipment for vehicle and storage medium

CN122023021ACN 122023021 ACN122023021 ACN 122023021ACN-122023021-A

Abstract

The application provides a loss part assessment processing method, device, equipment and storage medium of a vehicle, wherein the method comprises the steps of responding to input operation at an assessment end when any vehicle needs to be assessed, generating vehicle information of the vehicle and short for loss part; the method comprises the steps of inputting short names of lost parts into a pre-built accessory retrieval system for retrieval processing to obtain corresponding one or more standard accessory names, generating model prompt words according to preset matching instructions and short names of the lost parts, inputting the model prompt words and the standard accessory names into a trained accessory matching model for matching processing to obtain target standard accessory names corresponding to the lost parts, inputting vehicle information and the target standard accessory names into a lost part mapping model for coding mapping processing to obtain system codes of the lost parts, and determining an damage assessment result of the lost parts according to the system codes, so that efficiency of loss part damage assessment processing of vehicles is improved.

Inventors

JIN LIANGLIANG
YANG YAGANG
HE DELIN
GAO TIAN

Assignees

中国人民财产保险股份有限公司

Dates

Publication Date: 20260512
Application Date: 20251229

Claims (10)

1. A lost part damage assessment processing method for a vehicle, which is applied to computer equipment and comprises the following steps: when any vehicle needs to be subjected to damage assessment, responding to input operation at a damage assessment end, and generating vehicle information of the vehicle and short for lost parts; Inputting the short name of the lost part into a pre-constructed accessory retrieval system for retrieval processing to obtain one or more corresponding standard accessory names; generating model prompt words according to preset matching instructions and short names of the lost pieces; Inputting the model prompt words and the names of the standard accessories into a trained accessory matching model for matching treatment so as to obtain the names of the target standard accessories corresponding to the lost pieces; Inputting the vehicle information and the target standard accessory name into a lost part mapping model for coding mapping processing so as to obtain a system code of the lost part; And determining the loss assessment result of the loss part according to the system code.
2. The method of claim 1, wherein the process of constructing the accessory retrieval system comprises: standard accessory names of all accessories corresponding to all vehicles are obtained; inputting standard accessory names of all accessories into a preset semantic model for semantic feature extraction processing to obtain multidimensional assembly semantic elements corresponding to all accessories; Storing the multidimensional assembly semantic elements of each accessory to a preset accessory retrieval library; After the storage of the preset accessory search library is completed, constructing a neighbor search index according to the preset accessory search library by a preset similarity method to obtain an accessory search system.
3. The method of claim 2, wherein the training process of the accessory-matching model comprises: Acquiring historical loss part loss assessment data, and extracting a plurality of initial loss part mapping pairs from the historical loss part loss assessment data, wherein each initial loss part mapping pair is used for indicating the mapping relation between the abbreviation of the corresponding loss part and the standard accessory name; cleaning the plurality of initial lost part mapping pairs to obtain cleaned lost part mapping pairs; loading a pre-trained fitting matching model; And performing fine tuning on the pre-trained fitting matching model through a preset fine tuning method and each cleaned lost part mapping pair so as to obtain a trained fitting matching model.
4. A method according to claim 3, wherein the cleaning of the plurality of initial loss element map pairs to obtain cleaned loss element map pairs comprises: filtering the plurality of initial loss element mapping pairs to obtain filtered loss element mapping pairs; performing outlier detection processing on each filtered lost part mapping pair to obtain each detected lost part mapping pair; And calibrating the detected loss element mapping pairs to obtain calibrated loss element mapping pairs.
5. The method of claim 4, wherein said fine tuning the pre-trained fitting matching model to obtain a trained fitting matching model by a pre-set fine tuning method and each cleaned pair of loss element mappings comprises: Freezing model parameters of a first preset proportion in the pre-trained fitting matching model through the preset fine tuning method; After the model parameters are frozen, injecting a low-rank matrix with a second preset proportion into an attention layer and a feedforward neural network layer in the pre-trained fitting matching model; Performing fine tuning training on the low-rank matrix according to the cleaned loss mapping pair to obtain a trained low-rank matrix; and integrating the trained low-rank matrix into the pre-trained accessory matching model to obtain a trained accessory matching model.
6. The method of claim 1, wherein the vehicle information is a vehicle identification code; Correspondingly, the step of inputting the vehicle information and the target standard accessory name into a loss part mapping model for coding mapping processing so as to obtain a system code of the loss part comprises the following steps: inputting the vehicle information and the target standard accessory name into a loss mapping model to execute the following steps: Acquiring an analysis rule of a vehicle identification code, and determining brand information, train information and year information corresponding to the vehicle according to the analysis rule and the vehicle identification code; And carrying out coding mapping processing according to the brand information, the train information, the annual style information and the target standard accessory name so as to obtain the system code of the lost part.
7. The method according to any one of claims 1 to 6, further comprising, after said inputting the vehicle information and the target standard accessory name into a loss mapping model for performing a code mapping process to obtain a systematic code of the loss: periodically, acquiring new model information or new part coding rules; And updating the new model information or the new part coding rule to the lost part mapping model through an incremental learning mechanism.
8. A lost motion estimation processing apparatus for a vehicle, characterized by being applied to a computer device, comprising: The first generation module is used for responding to input operation at the damage assessment end when any vehicle needs damage assessment, and generating vehicle information of the vehicle and short for lost parts; The searching module is used for inputting the short name of the lost part into a pre-built accessory searching system for searching processing so as to obtain one or more corresponding standard accessory names; the second generation module is used for generating model prompt words according to a preset matching instruction and the short name of the lost part; The first input module is used for inputting the model prompt words and the names of all standard accessories into a trained accessory matching model to carry out matching processing so as to obtain the names of the target standard accessories corresponding to the lost piece; The second input module is used for inputting the vehicle information and the target standard accessory name into a lost part mapping model to carry out coding mapping processing so as to obtain a system code of the lost part; and the determining module is used for determining the loss assessment result of the loss piece according to the system code.
9. A computer device comprises at least one processor and a memory; The memory stores computer-executable instructions; The at least one processor executing computer-executable instructions stored in the memory causes the at least one processor to perform the lost-part damage-assessment processing method of a vehicle as claimed in any one of claims 1 to 7.
10. A computer storage medium having stored therein computer-executable instructions which, when executed by a processor, implement the lost-part damage-assessment method of a vehicle according to any one of claims 1 to 7.

Description

Loss part damage assessment processing method, device and equipment for vehicle and storage medium Technical Field The present application relates to the field of vehicle damage assessment technologies, and in particular, to a method, an apparatus, a device, and a storage medium for processing damage of a damaged part of a vehicle. Background Loss of a lost part is a process of specially evaluating damaged parts in an accident, the accident loss and the natural abrasion are required to be distinguished, and the principle of maintenance or replacement is generally followed. And the damage assessment staff confirms a replacement or repair scheme and cost through on-site investigation, damage detection and maintenance data calculation, and finally forms a damage assessment report as a basis for claim settlement. In the prior art, according to a quotation provided by a repair shop, a loss evaluation person inputs a loss name through computer equipment, and selects a corresponding standard accessory from a system through keyword prompt and personal experience to generate an evaluation result. However, in the prior art, a loss assessment person inputs the name of the lost part through computer equipment, and selects a loss assessment mode of a corresponding standard accessory from a system through keyword prompt and personal experience, and the loss assessment mode is too dependent on manpower, so that the loss assessment processing efficiency of the vehicle is reduced. Disclosure of Invention The application provides a lost part damage assessment processing method, device, equipment and storage medium for a vehicle, which are used for solving the problems that damage assessment staff inputs a lost part name through computer equipment, and a damage assessment mode of selecting a corresponding standard accessory from a system through keyword prompt and personal experience is too dependent on manpower, so that the efficiency of the lost part damage assessment processing of the vehicle is reduced. In a first aspect, the present application provides a method for processing loss of a lost part of a vehicle, applied to a computer device, including: when any vehicle needs to be subjected to damage assessment, responding to input operation at a damage assessment end, and generating vehicle information of the vehicle and short for lost parts; Inputting the short name of the lost part into a pre-constructed accessory retrieval system for retrieval processing to obtain one or more corresponding standard accessory names; generating model prompt words according to preset matching instructions and short names of the lost pieces; Inputting the model prompt words and the names of the standard accessories into a trained accessory matching model for matching treatment so as to obtain the names of the target standard accessories corresponding to the lost pieces; Inputting the vehicle information and the target standard accessory name into a lost part mapping model for coding mapping processing so as to obtain a system code of the lost part; And determining the loss assessment result of the loss part according to the system code. In one possible design, the construction process of the accessory retrieval system comprises the steps of obtaining standard accessory names of all accessories corresponding to all vehicles, inputting the standard accessory names of all accessories into a preset semantic model to conduct semantic feature extraction processing to obtain multi-dimensional assembly semantic elements corresponding to all the accessories, storing the multi-dimensional assembly semantic elements of all the accessories into a preset accessory retrieval library, and constructing neighbor retrieval indexes according to the preset accessory retrieval library through a preset similarity method after the storage of the preset accessory retrieval library is completed to obtain the accessory retrieval system. In one possible design, the training process of the accessory matching model comprises the steps of obtaining historical loss piece loss assessment data, extracting a plurality of initial loss piece mapping pairs from the historical loss piece loss assessment data, wherein each initial loss piece mapping pair is used for indicating a mapping relation between a corresponding loss piece and a standard accessory name, cleaning the plurality of initial loss piece mapping pairs to obtain each cleaned loss piece mapping pair, loading a pre-trained accessory matching model, and performing fine adjustment on the pre-trained accessory matching model through a preset fine adjustment method and each cleaned loss piece mapping pair to obtain the trained accessory matching model. In one possible design, the cleaning process is performed on the plurality of initial loss map pairs to obtain each cleaned loss map pair, and the cleaning process comprises the steps of performing filtering process on the plurality of initial loss map pairs to obtain e