Method and system for motor vehicle inspection key parameter patrol inspection based on file comparison

Abstract

The application discloses a motor vehicle inspection key parameter inspection method and system based on file comparison, relates to the technical field of motor vehicle inspection, and can more scientifically measure the fluctuation rationality of the same vehicle or vehicle group in different periods of inspection data by analyzing the information entropy difference value of historical files and current files, combining bin modeling and probability distribution, realizing quantitative perception of detection consistency deviation, and effectively avoiding misjudgment caused by factors such as equipment error or manual input. By constructing parameter entropy difference and vehicle entropy fluctuation index, parameters with stable distribution law in the historical period but sudden change in the current inspection can be identified, suspected manipulation traces or abnormal fluctuations of equipment are tracked and prompted, processing suggestions are timely pushed, and stronger early warning and pointing capabilities are achieved. Based on file comparison and historical record mining, full-process modeling analysis is realized, and the application does not depend on external additional hardware or sensor modification, and can be seamlessly connected to the current motor vehicle inspection data management platform.

Description

Technical Field

[0001] This invention relates to the field of motor vehicle inspection technology, specifically to a method and system for inspecting key parameters of motor vehicle inspection based on document comparison. Background Technology

[0002] With the strengthening of vehicle annual inspection, environmental protection testing, and safety performance testing systems, testing agencies across the country have successively implemented digital data archiving, forming a continuous inspection record chain based on vehicles. In-depth analysis of this cross-cycle inspection data can help identify potential fluctuations, abnormal trends, or potential intervention behaviors hidden in the inspection results, becoming a key component of current motor vehicle quality supervision and intelligent inspection ranking.

[0003] In current vehicle inspection systems, parameter determination largely relies on static upper and lower threshold comparison rules, which determine whether the test data falls within a fixed range. However, this method ignores the distribution and evolution characteristics of parameters for the same vehicle throughout its historical inspections, making it difficult to reflect the trend changes and behavioral deviations of parameters. Even if the current test result is within the static threshold, a significant deviation from the vehicle's historical parameter distribution may still indicate performance degradation, detection bias, or abnormal results.

[0004] The root cause of these problems lies in the existing system's lack of statistical modeling capabilities for the behavior patterns of inspection parameters. It fails to utilize historical data to construct a baseline for the parameter behavior of each vehicle, thus making it impossible to identify the extent to which the current value perturbs the original behavior pattern. Once a vehicle experiences a sudden parameter change, even if it remains within the compliance range, the system struggles to provide a clear warning. Furthermore, human intervention during the detection process often manifests as "no abnormal values, but abnormal distribution," concealing actual risks. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method and system for inspecting key parameters of motor vehicle inspection based on document comparison, which solves the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solution: a vehicle inspection key parameter inspection system based on document comparison, including an inspection document structure parsing module, a parameter history sequence construction module, a parameter distribution entropy calculation module, an entropy difference screening and fluctuation index construction module, and an abnormal inspection pointing output module; The inspection document structure parsing module receives the current inspection document and historical inspection record documents, parses them into a standard structure table, and constructs a structured parameter set F; The parameter history sequence construction module is based on the structured parameter set F. It constructs a cross-inspection cycle isoparameter sequence for each motor vehicle, removes missing values ​​and arranges them in ascending order of time to construct the time series S of vehicle parameters. The parameter distribution entropy calculation module performs probability distribution modeling on the time series S of vehicle parameters, and uses equidistant binning and frequency method to calculate and obtain the historical information entropy Hihist, the current value of which is incorporated into the entropy Hicurr, and the parameter entropy difference ΔH. The entropy difference screening and fluctuation index construction module identifies parameters whose entropy difference ΔH exceeds the natural fluctuation range and constructs the vehicle entropy fluctuation index Ψwave. The anomaly inspection output module compares the vehicle entropy fluctuation index Ψwave with the set entropy wave threshold Twav, obtains the anomaly parameter set Pabn, and generates a specific processing strategy.

[0007] Preferably, the document structure parsing module includes a field mapping identification and parameter standardization unit and a data cleaning and normalization unit; The field mapping identification and parameter standardization unit receives motor vehicle inspection reports from different sources in the current inspection file and historical inspection record files, identifies the field structure in the report, including engine idle speed fluctuation pv, actual air-fuel ratio pa, noise stability coefficient po, and braking delay time ps; and fits it into a parameter name set P={pv, pa, po, ps}. The parameter name set P={pv,pa,po,ps} is mapped to a unified field parameter name p using a mapping function, and the original value and timestamp are recorded to construct a structured parameter tuple f=(pi,vi,tsi). In the formula, pi represents the parameter name of the i-th parameter (including pv, pa, po, ps), vi represents the original parameter value of the i-th parameter, and tsi represents the timestamp of the i-th parameter.

[0008] Preferably, the data cleaning and normalization unit performs validity checks on the obtained structured parameter tuple f = (pi, vi, tsi), including removing out-of-limit values ​​and correcting abnormal characters, to construct a valid dataset fa; The method for handling out-of-limit values ​​is as follows: For each type of parameter, the engineering boundary intervals are pre-configured, and the intervals are derived from the measurement range of the testing equipment, the operating condition range of the vehicle, and the upper and lower boundaries of historical statistical distributions; for example: Engine idle speed fluctuation: only changes within a small range; Air-fuel ratio: fluctuates narrowly around the stoichiometric ratio; Noise stability: finite fluctuation value; Braking delay time: only in the millisecond range; The above range is stored in the system in the form of parameter - minimum allowed value - maximum allowed value; Then, the structured parameter tuples are traversed one by one; the original parameter value in each structured parameter tuple f is compared with the physical range of the corresponding parameter; if the original parameter value is lower than the minimum allowable value or higher than the maximum allowable value, it is determined to be an out-of-limit value; and the records determined to be out of limit are deleted or marked as invalid and removed. The method for correcting abnormal characters is as follows: The first step is field type checking: determine whether the numeric field contains non-numeric characters or illegal symbols; The second step is character cleanup: extracting content containing unit characters or symbols, keeping the numerical part, and removing the characters "ms", "dB", and "%". The third step is to identify missing data: spaces, "-", "N / A", and "null" are uniformly identified as missing data. The fourth step is to handle missing values: either delete the missing data or impute it with values ​​from adjacent time points. Step 5, format rewriting: unify all numerical values ​​into floating-point format and write them; The obtained valid dataset fa is normalized using the max-min normalization method to obtain the structured parameter set F.

[0009] Preferably, the parameter history sequence construction module includes a parameter clustering and missing value screening unit and a time series ascending order reconstruction and fluctuation correction unit; The parameter clustering and missing value screening unit first classifies the structured parameter set F according to the VIN number to ensure that each group of data contains only the inspection records of the same vehicle; in the data of each vehicle, it is classified a second time according to the parameter name p, and each group of classification contains only the timestamps of different inspections with the same parameter name p. For the structured parameter set F after classification, delete the data without numerical values ​​and remove duplicate records under the same timestamp to obtain the original time series structure Srawi={(vi(t1),t1),(vi(t2),t2),...,(vi(tk),tk)}, where vi(tk) represents the original parameter value of the i-th parameter at time tk; The time series ascending order reconstruction and fluctuation correction unit sorts the original time series structure Srawi of each parameter in ascending order according to the timestamp ts, and obtains the standard vector form vSi=[vi(ts1),vi(ts2),...,vi(tsk)], where ts1<ts2<...<tsk; In the formula, vi(tsk) represents the original parameter value of the i-th parameter at timestamp tsk; The standard vector form vSi is subjected to fluctuation correction by the extreme slope mutation determination method, and the abnormal jump values ​​are smoothed to obtain the time series S of vehicle parameter P. The formula for determining extreme slope mutations is as follows: ; In the formula, vi(tj) represents the detected value of parameter pi in the j-th time segment, (tj+1) represents the timestamp of the (j+1)-th test, vi(tj+1) represents the detected value of parameter pi in the (j+1)-th time segment, and Δi(j) represents the slope of the change of the i-th parameter in the j-th time segment. If the absolute value of the slope Δi(j) of the i-th parameter in the j-th time segment is greater than the mean ± 3 times the standard deviation, it is marked as an anomalous jump.

[0010] Preferably, the parameter distribution entropy calculation module includes a binning modeling and historical distribution probability generation unit and a current value integration and entropy difference calculation unit; The binning modeling and historical distribution probability generation unit discretizes the time series S into a distribution form, sets the number of bins m, and obtains the bin width wi and the bin interval Bj; the commonly used value range for the number of bins m is 8 to 20, such as m=1; Find the minimum value minvi and the maximum value maxvi in ​​the time series S; minvi = min(S), maxvi = max(S); The width wi of each bin is obtained by dividing the difference between the maximum value maxvi and the minimum value minvi by the number of bins m. The binning interval Bj = [minvi + (j-1)wi, minvi + (j)wi), j = 1, 2, ..., m; Perform binning operation on all original parameter values ​​vi∈S, count the number of samples nj that fall into each bin interval Bj; calculate the frequency of occurrence for each bin as the probability pij; The probability pij is obtained by the ratio of the number of samples nj falling into each bin interval Bj to the number of samples in the time series.

[0011] Preferably, the current value integration and entropy difference calculation unit matches the current value Dv of the currently detected motor vehicle parameter with the bin position in the historical values, falls into the corresponding bin interval Bj, and obtains the current sample number Nnj; the current sample number Nnj is equal to the sample number nj+1; Calculate the new probability npij after the current value Dv of the motor vehicle parameter falls into the bin interval Bj; By analyzing the probability pij and the new probability npij, we obtain the historical information entropy hisH and the current information entropy DH. The historical information entropy hisH is obtained as follows: traverse each bin: starting from the 1st bin, process sequentially up to the mth bin; extract the probability value of each bin: for the jth bin, obtain its probability of occurrence in historical samples; multiply the probability value by its own natural logarithm; sum the products of all bins; add a negative sign to obtain the information entropy; finally, take the negative of the sum to obtain the historical information entropy hisH of the parameter in historical samples. The current information entropy DH is obtained as follows: based on the original bin probabilities, add the current detection value and update the probability of each bin; re-traverse all bins and extract the current probability value; multiply the current probability value by the natural logarithm; then, sum the products of all bins and take the negative sign; thus, the current information entropy DH is obtained. By analyzing the historical information entropy hisH and the current information entropy DH, the parameter entropy difference ΔH is obtained; The parameter entropy difference ΔH is obtained by the difference between the current information entropy DH and the historical information entropy hisH.

[0012] Preferably, the entropy difference screening and fluctuation index construction module includes a parameter entropy difference normalization processing and anomaly screening unit and a vehicle entropy fluctuation index calculation unit; The parameter entropy difference normalization and anomaly screening unit summarizes the parameter entropy difference ΔH of all detected vehicles to obtain the entropy difference set FH={ΔH1, ΔH2, ..., ΔHN}; where N represents the total number of detected vehicles. The parameter entropy difference ΔH for each vehicle is normalized as follows: Zi = ΔHi / σΔHi; where ΔHi represents the parameter entropy difference ΔH of the i-th parameter, σΔHi represents the parameter standard deviation of the i-th parameter, and Zi represents the normalized parameter entropy difference of the i-th parameter. The parameter entropy difference Zi of the i-th parameter after normalization is compared with the preset entropy difference threshold Tz to filter out the parameter set Nab that exceeds the natural fluctuation. The filtering method is as follows: Nab={pi∣∣Zi∣>Tz}, where the entropy difference threshold Tz is 2.0.

[0013] Preferably, the vehicle entropy fluctuation index calculation unit integrates the parameter entropy difference zi of all the normalized i-th parameter in the parameter set Nab that exceeds the natural fluctuation, and calculates and obtains the vehicle entropy fluctuation index Ψwave. The vehicle entropy fluctuation index Ψwave is obtained using the following formula: ; In the formula, M represents the total number of parameters for which the entropy difference of the current vehicle is calculated in this detection.

[0014] Preferably, the anomaly inspection output module includes an inspection object determination and anomaly parameter set extraction unit and an inspection output generation and processing suggestion push unit; The inspection object determination and abnormal parameter set extraction unit compares the acquired vehicle entropy fluctuation index Ψwave with the set vehicle fluctuation threshold Twav to obtain abnormal parameters; The vehicle fluctuation threshold Twav is obtained using the following formula: Twav = μΨwave + kx × σΨwave; In the formula, μΨwave represents the mean of the vehicle entropy fluctuation index, σΨwave represents the standard deviation of the vehicle entropy fluctuation index, and kx represents the adjustment coefficient. The abnormal parameters are obtained as follows: When the vehicle entropy fluctuation index Ψwave ≤ vehicle fluctuation threshold Twav, it indicates that the parameters are normal and no marking is made. When the vehicle entropy fluctuation index Ψwave > the vehicle fluctuation threshold Twav, it indicates that the parameter is abnormal and is marked. Fit all the parameters of the marked anomalies to obtain the anomaly parameter set Pabn; The inspection output generation and processing suggestion push unit generates a structured inspection pointer data structure Ovin=(Ψwave, Pabn, inspection level, processing suggestion) based on the results of the abnormal parameter set Pabn. Inspection levels and handling recommendations can be obtained through the following methods: When the vehicle fluctuation threshold Twav+1 > the vehicle entropy fluctuation index Ψwave > the vehicle fluctuation threshold Twav, it indicates the first inspection level, and the handling suggestion is: it is recommended to perform a retest of the braking system response. When the vehicle fluctuation threshold Twav+3 ≥ the vehicle entropy fluctuation index Ψwave ≥ the vehicle fluctuation threshold Twav+1, it indicates the second inspection level, and the handling suggestion is: it is recommended to randomly check the structural integrity or sound insulation components. When the vehicle entropy fluctuation index Ψwave > the vehicle fluctuation threshold Twav+3, it indicates the third inspection level, and the recommended action is to conduct engine bench re-inspection and status tracking.

[0015] The method for inspecting key parameters of motor vehicle inspection based on document comparison includes the following steps: Step 1: The inspection file structure parsing module receives the current inspection file and the historical inspection record file, parses them into a standard structure table, and constructs a structured parameter set F; Step 2: The parameter history sequence construction module uses the structured parameter set F as a basis to construct a cross-inspection cycle isoparameter sequence for each motor vehicle, removes missing values ​​and arranges them in ascending order of time to construct the time series S of vehicle parameters. Step 3: The parameter distribution entropy calculation module performs probability distribution modeling on the time series S of vehicle parameters, and uses equidistant binning and frequency method to calculate and obtain the historical information entropy Hihist, the current value of which is incorporated into the entropy Hicurr, and the parameter entropy difference ΔH. Step 4: The entropy difference screening and fluctuation index construction module identifies parameters whose entropy difference ΔH exceeds the natural fluctuation range and constructs the vehicle entropy fluctuation index Ψwave. Step 5: The anomaly inspection output module compares the vehicle entropy fluctuation index Ψwave with the set entropy wave threshold Twav, obtains the anomaly parameter set Pabn, and generates a specific processing strategy.

[0016] This invention provides a method and system for inspecting key parameters of motor vehicle inspection based on document comparison, which has the following beneficial effects: (1) When the system is running, by analyzing the information entropy difference between historical files and current files, combined with binning modeling and probability distribution, it can more scientifically measure the “fluctuation rationality” of the inspection data of the same vehicle or model group at different times, realize the quantitative perception of the detection consistency deviation, and effectively avoid misjudgment or omission caused by equipment error or human input.

[0017] (2) By constructing the parameter entropy difference ΔH and the vehicle entropy fluctuation index Ψwave, parameters that have a stable distribution pattern in the historical cycle but abruptly change in the current test can be identified. Tracking prompts can be generated for suspected manipulation traces or abnormal equipment fluctuations, and timely processing suggestions can be pushed. It has stronger "early warning" and "direction" capabilities. Through VIN number clustering and time series ascending order reconstruction mechanism, the system can automatically establish a unified grouping model for vehicle historical parameters. Standardization and normalization processing are performed in many aspects such as parameter name, field structure, and numerical format, so that test data from different sources and at different times are comparable, thereby supporting stable, long-term trend analysis and data fusion.

[0018] (3) By using the standard structure table parsing and field mapping mechanism, inspection records from different institutions or times are processed in a unified manner, ensuring that parameters from different sources can be compared horizontally and tracked vertically. Instead of being limited to single-point value judgments in the current inspection file, a cross-cycle continuous sequence of each motor vehicle is constructed based on historical records, so that the parameter change trend can be fully reflected.

[0019] (4) By using the calculation and unified index expression of the vehicle entropy fluctuation index, the results of multi-parameter disturbances can be comprehensively summarized, the overall abnormality of the vehicle can be quantitatively expressed, and the key vehicles can be automatically screened by comparing with the system threshold, thereby improving the scientific nature and priority of the inspection ranking. Since the system relies on the overall variation characteristics of the parameter behavior structure rather than absolute values, it has the ability to respond to behaviors such as human optimization and parameter edge detection, providing data support for combating false detection and helping to improve the authenticity and credibility of vehicle inspection results. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating the inspection system for key parameters of motor vehicle inspection based on document comparison, as described in this invention. Figure 2 This is a schematic diagram of the steps of the inspection method for key parameters of motor vehicle inspection based on document comparison according to the present invention. Figure 3 This is a schematic diagram of the process for obtaining and judging the vehicle entropy fluctuation index according to the present invention. Figure 4 This is a schematic diagram of the time series processing flow of the present invention. Detailed Implementation

[0021] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0022] Example 1 This invention provides a vehicle inspection key parameter inspection system based on document comparison. Please refer to [link / reference]. Figures 1 to 4 It includes a file structure parsing module, a parameter history sequence construction module, a parameter distribution entropy calculation module, an entropy difference screening and fluctuation index construction module, and an anomaly inspection output module; The inspection document structure parsing module receives the current inspection document and historical inspection record documents, parses them into a standard structure table, and constructs a structured parameter set F; The parameter history sequence construction module is based on the structured parameter set F. It constructs a cross-inspection cycle isoparameter sequence for each motor vehicle, removes missing values ​​and arranges them in ascending order of time to construct the time series S of vehicle parameters. The parameter distribution entropy calculation module performs probability distribution modeling on the time series S of vehicle parameters, and uses equidistant binning and frequency method to calculate and obtain the historical information entropy Hihist, the current value of which is incorporated into the entropy Hicurr, and the parameter entropy difference ΔH. The entropy difference screening and fluctuation index construction module identifies parameters whose entropy difference ΔH exceeds the natural fluctuation range and constructs the vehicle entropy fluctuation index Ψwave. The anomaly inspection output module compares the vehicle entropy fluctuation index Ψwave with the set entropy wave threshold Twav, obtains the anomaly parameter set Pabn, and generates a specific processing strategy.

[0023] In this embodiment, instead of relying on the traditional single-point judgment mode of whether parameters exceed their upper or lower limits, it establishes a time-series correlation between multiple historical test data to depict the evolution trajectory of parameter behavior from a vertical dimension. This allows it to identify test objects that, although not exceeding the limits, have shown abnormal trends or abrupt changes. By modeling the distribution characteristics of vehicle parameters within historical test cycles and introducing the change in information entropy as a disturbance assessment index, it reflects the overall volatility and distribution disturbance of parameters, thus overcoming the blind spot of traditional methods that cannot identify "normal values ​​but abnormal structures".

[0024] By calculating and unifying the vehicle entropy fluctuation index, the system can comprehensively summarize the results of multi-parameter disturbances, quantitatively express the overall degree of vehicle anomalies, and automatically screen key vehicles by comparing them with system thresholds, thereby improving the scientific nature and rationality of inspection ranking. Because the system relies on the overall variation characteristics of parameter behavior structure rather than absolute values, it has the ability to respond to human optimization and parameter manipulation, providing data support for combating fraudulent inspections and helping to improve the authenticity and credibility of vehicle inspection results.

[0025] This system achieves full-process modeling and analysis based on file comparison and historical record mining. It does not rely on external additional hardware or sensor modifications and can be seamlessly integrated into the current motor vehicle inspection data management platform, making it easy to promote and apply.

[0026] Example 2 This embodiment is an explanation based on Embodiment 1. Please refer to it. Figure 1 and Figure 4 Specifically: the document structure parsing module includes a field mapping identification and parameter standardization unit and a data cleaning and normalization unit; The field mapping identification and parameter standardization unit receives motor vehicle inspection reports from different sources in the current inspection file and historical inspection record files, identifies the field structure in the report, including engine idle speed fluctuation pv, actual air-fuel ratio pa, noise stability coefficient po, and braking delay time ps; and fits it into a parameter name set P={pv, pa, po, ps}. The parameter name set P={pv,pa,po,ps} is mapped to a unified field parameter name p using a mapping function, and the original value and timestamp are recorded to construct a structured parameter tuple f=(pi,vi,tsi). In the formula, pi represents the parameter name of the i-th parameter, and pi∈p, vi represents the original parameter value of the i-th parameter, and tsi represents the timestamp of the i-th parameter.

[0027] The data cleaning and normalization unit performs validity checks on the obtained structured parameter tuple f = (pi, vi, tsi), including removing out-of-limit values ​​and correcting abnormal characters, and constructs a valid dataset fa; The method for removing out-of-limit values ​​is as follows: For each type of parameter, the engineering boundary intervals that appear are pre-configured. The intervals are derived from the range of the testing equipment, the range of the vehicle's operating conditions, and the upper and lower limits of the historical statistical distribution. Then, the structured parameter tuples are traversed one by one; the original parameter value in each structured parameter tuple f is compared with the physical range of the corresponding parameter; if the original parameter value is lower than the minimum allowable value or higher than the maximum allowable value, it is determined to be an out-of-limit value; and the records determined to be out of limit are deleted or marked as invalid and removed. The method for correcting abnormal characters is as follows: The first step is field type checking: determine whether the numeric field contains non-numeric characters or illegal symbols; The second step is character cleanup: extracting content containing unit characters or symbols, keeping the numerical part, and removing the characters "ms", "dB", and "%". The third step is to identify missing data: spaces, "-", "N / A", and "null" are uniformly identified as missing data. The fourth step is to handle missing values: either delete the missing data or impute it with values ​​from adjacent time points. Step 5, format rewriting: unify all numerical values ​​into floating-point format and write them; The obtained valid dataset fa is normalized using the max-min normalization method to obtain the structured parameter set F.

[0028] In this embodiment, by using field mapping identification and parameter standardization units, key fields from inspection reports from different institutions, formats, or time periods can be automatically identified and uniformly fitted into a standardized set of parameter names. This avoids data mismatch or incomparability caused by inconsistent naming or mixed units, improving the reusability and horizontal alignment capabilities of historical data. By constructing structured parameter tuples, parameter values, parameter names, and corresponding timestamps are organically bound, enabling the parameter behavior of the same vehicle at different inspection time points to be effectively reconstructed and tracked, meeting the data structure requirements of subsequent time series modeling, trend identification, and other analysis modules.

[0029] By setting engineering boundary value ranges, abnormal data that clearly exceeds physical or operational possibilities is eliminated, effectively avoiding extreme value contamination caused by sensor drift, input errors, or data crosstalk, and improving the stability and interpretability of parameter sequences. The abnormal character correction mechanism can automatically handle data fields with non-standard formats, such as strings containing units, illegal symbols, and missing identifiers, giving the system stronger data cleaning and self-healing capabilities and reducing analysis interruptions or misjudgments caused by non-standard data.

[0030] By using maximal-minimum normalization, the original parameter values ​​are mapped to a uniform scale range, which helps to eliminate the influence of differences in the dimensions of different parameters and provides a unified input basis for subsequent information entropy modeling and disturbance assessment.

[0031] From field identification and anomaly removal to unified format and scaling transformation, a full-process, standardized, and scalable data processing mechanism has been built to ensure that the input data quality meets the "continuity, integrity, and standardization" input requirements of the subsequent entropy difference analysis module, thereby improving the stability and analytical accuracy of the entire system.

[0032] Example 3 This embodiment is an explanation based on Embodiment 2. Please refer to it. Figure 1 Specifically: the parameter history sequence construction module includes a parameter clustering and missing value screening unit and a time series ascending order reconstruction and fluctuation correction unit; The parameter clustering and missing value screening unit first classifies the structured parameter set F according to the VIN number to ensure that each group of data contains only the inspection records of the same vehicle; in the data of each vehicle, it is classified a second time according to the parameter name p, and each group of classification contains only the timestamps of different inspections with the same parameter name p. For the structured parameter set F after classification, delete the data without numerical values ​​and remove duplicate records under the same timestamp to obtain the original time series structure Srawi={(vi(t1),t1),(vi(t2),t2),...,(vi(tk),tk)}, where vi(tk) represents the original parameter value of the i-th parameter at time tk; The time series ascending order reconstruction and fluctuation correction unit sorts the original time series structure Srawi of each parameter in ascending order according to the timestamp ts, and obtains the standard vector form vSi=[vi(ts1),vi(ts2),...,vi(tsk)], where ts1<ts2<...<tsk; In the formula, vi(tsk) represents the original parameter value of the i-th parameter at timestamp tsk; The standard vector form vSi is subjected to fluctuation correction by the extreme slope mutation determination method, and the abnormal jump values ​​are smoothed to obtain the time series S of vehicle parameter P. The formula for determining extreme slope mutations is as follows: ; In the formula, vi(tj) represents the detected value of parameter pi in the j-th time segment, (tj+1) represents the timestamp of the (j+1)-th test, vi(tj+1) represents the detected value of parameter pi in the (j+1)-th time segment, and Δi(j) represents the slope of the change of the i-th parameter in the j-th time segment. If the absolute value of the slope Δi(j) of the i-th parameter in the j-th time segment is greater than the mean ± 3 times the standard deviation, it is marked as an anomalous jump.

[0033] The parameter distribution entropy calculation module includes a binning modeling and historical distribution probability generation unit and a current value integration and entropy difference calculation unit; The binning modeling and historical distribution probability generation unit discretizes the time series S into a distribution form, sets the number of bins m, and obtains the bin width wi and the bin interval Bj; Find the minimum value minvi and the maximum value maxvi in ​​the time series S; minvi = min(S), maxvi = max(S); The width wi of each bin is obtained by dividing the difference between the maximum value maxvi and the minimum value minvi by the number of bins m. The binning interval Bj = [minvi + (j-1)wi, minvi + (j)wi), j = 1, 2, ..., m; Perform binning operation on all original parameter values ​​vi∈S, count the number of samples nj that fall into each bin interval Bj; calculate the frequency of occurrence for each bin as the probability pij; The probability pij is obtained by the ratio of the number of samples nj falling into each bin interval Bj to the number of samples in the time series.

[0034] The current value integration and entropy difference calculation unit maps the current value Dv of the detected motor vehicle parameter to the bin position in the historical values, places it into the corresponding bin interval Bj, and obtains the current sample number Nnj; the current sample number Nnj is equal to the sample number nj+1; Calculate the new probability npij after the current value Dv of the motor vehicle parameter falls into the bin interval Bj; By analyzing the probability pij and the new probability npij, we obtain the historical information entropy hisH and the current information entropy DH. The historical information entropy hisH is obtained as follows: traverse each bin: starting from the 1st bin, process sequentially up to the mth bin; extract the probability value of each bin: for the jth bin, obtain its probability of occurrence in historical samples; multiply the probability value by its own natural logarithm; sum the products of all bins; add a negative sign to obtain the information entropy; finally, take the negative of the sum to obtain the historical information entropy hisH of the parameter in historical samples. The current information entropy DH is obtained as follows: based on the original bin probabilities, add the current detection value and update the probability of each bin; re-traverse all bins and extract the current probability value; multiply the current probability value by the natural logarithm; then, sum the products of all bins and take the negative sign; thus, the current information entropy DH is obtained. By analyzing the historical information entropy hisH and the current information entropy DH, the parameter entropy difference ΔH is obtained; The parameter entropy difference ΔH is obtained by the difference between the current information entropy DH and the historical information entropy hisH.

[0035] In this embodiment, a VIN numbering grouping mechanism is introduced to ensure that the same time series corresponds to only a single vehicle, effectively avoiding the misleading effects of mixed data from different vehicles and establishing a traceable and trackable single-vehicle parameter behavior trajectory. Re-clustering is performed according to parameter dimensions, so that each type of parameter forms an independent sequence under different detection cycles for the same vehicle, providing an accurate data foundation for trend modeling, fluctuation identification, and anomaly detection. Through operations such as missing value checking, duplicate removal, and time-order reconstruction, the data integrity and temporal consistency of the time series are significantly improved, preventing data contamination from affecting downstream judgment results.

[0036] The extreme slope mutation judgment strategy corrects and smooths parameter points with abnormal jumps, helping to suppress the interference of sporadic errors or unstable data on the overall behavioral trend, thereby improving the stability and reliability of parameter fluctuation analysis. Compared with traditional absolute value comparison or deviation identification, this embodiment introduces equidistant binning modeling and frequency statistics to construct a probability distribution model, so that the historical behavior of each type of parameter can be transformed into structural features from a statistical perspective. By using the current detection value as an external perturbation, embedding it into the historical distribution and dynamically updating the distribution probability, the difference in entropy values ​​before and after the perturbation is quantified, thereby forming a parameter entropy difference ΔH index with trend sensitivity.

[0037] Example 4 This embodiment is an explanation based on Embodiment 3. Please refer to it. Figure 1 and Figure 3 Specifically: the entropy difference screening and fluctuation index construction module includes a parameter entropy difference normalization processing and anomaly screening unit and a vehicle entropy fluctuation index calculation unit; The parameter entropy difference normalization and anomaly screening unit summarizes the parameter entropy difference ΔH of all detected vehicles to obtain the entropy difference set FH={ΔH1, ΔH2, ..., ΔHN}; where N represents the total number of detected vehicles. The parameter entropy difference ΔH for each vehicle is normalized as follows: Zi = ΔHi / σΔHi; where ΔHi represents the parameter entropy difference ΔH of the i-th parameter, σΔHi represents the parameter standard deviation of the i-th parameter, and Zi represents the normalized parameter entropy difference of the i-th parameter. The parameter entropy difference Zi of the i-th parameter after normalization is compared with the preset entropy difference threshold Tz to filter out the parameter set Nab that exceeds the natural fluctuation. The filtering method is as follows: Nab={pi∣∣Zi∣>Tz}, where the entropy difference threshold Tz is 2.0.

[0038] The vehicle entropy fluctuation index calculation unit integrates the parameter entropy difference zi of all the normalized i-th parameter in the parameter set Nab that exceeds the natural fluctuation, and calculates and obtains the vehicle entropy fluctuation index Ψwave. The vehicle entropy fluctuation index Ψwave is obtained using the following formula: ; In the formula, M represents the total number of parameters for which the entropy difference of the current vehicle is calculated in this detection.

[0039] The anomaly inspection output module includes an inspection object determination and anomaly parameter set extraction unit and an inspection output generation and processing suggestion push unit. The inspection object determination and abnormal parameter set extraction unit compares the acquired vehicle entropy fluctuation index Ψwave with the set vehicle fluctuation threshold Twav to obtain abnormal parameters; The abnormal parameters are obtained as follows: When the vehicle entropy fluctuation index Ψwave ≤ vehicle fluctuation threshold Twav, it indicates that the parameters are normal and no marking is made. When the vehicle entropy fluctuation index Ψwave > the vehicle fluctuation threshold Twav, it indicates that the parameter is abnormal and is marked. Fit all the parameters of the marked anomalies to obtain the anomaly parameter set Pabn; The inspection output generation and processing suggestion push unit generates a structured inspection pointer data structure Ovin=(Ψwave, Pabn, inspection level, processing suggestion) based on the results of the abnormal parameter set Pabn. Inspection levels and handling recommendations can be obtained through the following methods: When the vehicle fluctuation threshold Twav+1 > the vehicle entropy fluctuation index Ψwave > the vehicle fluctuation threshold Twav, it indicates the first inspection level, and the handling suggestion is: it is recommended to perform a retest of the braking system response. When the vehicle fluctuation threshold Twav+3 ≥ the vehicle entropy fluctuation index Ψwave ≥ the vehicle fluctuation threshold Twav+1, it indicates the second inspection level, and the handling suggestion is: it is recommended to randomly check the structural integrity or sound insulation components. When the vehicle entropy fluctuation index Ψwave > the vehicle fluctuation threshold Twav+3, it indicates the third inspection level, and the recommended action is to conduct engine bench re-inspection and status tracking.

[0040] In this embodiment, for detection parameters with different physical dimensions and varying magnitudes, this module employs standard deviation normalization to ensure that all parameters are compared on the same numerical scale. This avoids high-value parameters masking low-value fluctuations, significantly improving the consistency and fairness of fluctuation identification among parameters. Through a unified normalization threshold judgment mechanism, a natural fluctuation range is established, forming a parameter screening strategy that can be uniformly executed for different vehicles and different inspection rounds, reducing subjective bias introduced by manually set boundaries.

[0041] Instead of relying on abnormal changes in a single parameter, it integrates the comprehensive performance of multiple normalized parameters to form a unified indicator for the evolution of the entire vehicle's state, enabling the calculable, sortable, and hierarchical expression of complex fluctuation trends. The vehicle entropy fluctuation index has a strong trend expression capability and can sensitively capture the deviation trend of the current inspected vehicle from its own historical behavior in the statistical structure, thereby supporting a precise anomaly screening trigger mechanism.

[0042] Vehicles are classified into multiple inspection levels based on the numerical range of the vehicle entropy fluctuation index, enabling the system to formulate different levels of processing suggestions based on the degree of fluctuation, thereby enhancing the pertinence and stratification of inspection resource allocation.

[0043] This module generates a structured inspection output structure, including key values, anomaly parameter sets, indication levels, and suggested actions, supporting users in quickly identifying risk objects and key intervention areas in massive data environments. Through a structured anomaly identification process and a unified output format, this system can independently complete the entire process from data analysis to strategy formulation, significantly enhancing the system's closed-loop processing capabilities and reducing reliance on manual intervention.

[0044] Example 5 For a method of inspecting key parameters of motor vehicle inspection based on document comparison, please refer to [reference needed]. Figure 2 Specifically, it includes the following steps: Step 1: The inspection file structure parsing module receives the current inspection file and the historical inspection record file, parses them into a standard structure table, and constructs a structured parameter set F; Step 2: The parameter history sequence construction module uses the structured parameter set F as a basis to construct a cross-inspection cycle isoparameter sequence for each motor vehicle, removes missing values ​​and arranges them in ascending order of time to construct the time series S of vehicle parameters. Step 3: The parameter distribution entropy calculation module performs probability distribution modeling on the time series S of vehicle parameters, and uses equidistant binning and frequency method to calculate and obtain the historical information entropy Hihist, the current value of which is incorporated into the entropy Hicurr, and the parameter entropy difference ΔH. Step 4: The entropy difference screening and fluctuation index construction module identifies parameters whose entropy difference ΔH exceeds the natural fluctuation range and constructs the vehicle entropy fluctuation index Ψwave. Step 5: The anomaly inspection output module compares the vehicle entropy fluctuation index Ψwave with the set entropy wave threshold Twav, obtains the anomaly parameter set Pabn, and generates a specific processing strategy.

[0045] In this embodiment, a standard structure table parsing and field mapping mechanism is used to uniformly process inspection records from different institutions or time periods, ensuring that parameters from different sources can be compared horizontally and tracked vertically. It is no longer limited to judging single-point values ​​in the current inspection file, but instead constructs a cross-cycle continuous sequence for each vehicle based on historical records, allowing for a comprehensive reflection of parameter change trends.

[0046] This method does not rely on fixed thresholds for judgment. Instead, it calculates the statistical distribution characteristics of parameter values ​​through equidistant binning and frequency estimation methods to quantify the stability of historical changes. The latest parameter values ​​are dynamically integrated into the historical distribution structure, and the degree of disturbance to the original statistical structure is calculated, forming a quantifiable entropy difference ΔH. This identifies minor anomalies that are "within the range but unreasonable." By normalizing the parameter entropy difference and introducing a unified natural fluctuation threshold screening mechanism, the system ensures that it automatically identifies key parameters with significant statistical deviations.

[0047] Instead of judging single parameters in isolation, the system aggregates the entropy difference of multiple parameters to form a unified fluctuation expression index, serving as the core feature quantity of abnormal vehicle behavior. The system automatically extracts abnormal parameters and accordingly pushes inspection levels and differentiated intervention strategies, supporting different processing levels such as targeted retesting, spot checks, or comprehensive review. This effectively reduces manual intervention and subjective judgment, enhances the system's structured understanding and response capabilities to complex detection data, and forms a closed-loop intelligent inspection system.

[0048] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A vehicle inspection key parameter inspection system based on document comparison, characterized in that: It includes a file structure parsing module, a parameter history sequence construction module, a parameter distribution entropy calculation module, an entropy difference screening and fluctuation index construction module, and an anomaly inspection output module; The inspection document structure parsing module receives the current inspection document and historical inspection record documents, parses them into a standard structure table, and constructs a structured parameter set F; The parameter history sequence construction module is based on the structured parameter set F. It constructs a cross-inspection cycle isoparameter sequence for each motor vehicle, removes missing values ​​and arranges them in ascending order of time to construct the time series S of vehicle parameters. The parameter distribution entropy calculation module performs probability distribution modeling on the time series S of vehicle parameters, and uses equidistant binning and frequency method to calculate and obtain the historical information entropy Hihist, the current value of which is incorporated into the entropy Hicurr, and the parameter entropy difference ΔH. The entropy difference screening and fluctuation index construction module identifies parameters whose entropy difference ΔH exceeds the natural fluctuation range and constructs the vehicle entropy fluctuation index Ψwave. The anomaly inspection output module compares the vehicle entropy fluctuation index Ψwave with the set entropy wave threshold Twav, obtains the anomaly parameter set Pabn, and generates a specific processing strategy.

2. The vehicle inspection key parameter inspection system based on document comparison according to claim 1, characterized in that: The document structure parsing module includes a field mapping identification and parameter standardization unit and a data cleaning and normalization unit; The field mapping recognition and parameter standardization unit receives motor vehicle inspection reports from different sources in the current inspection file and historical inspection record file, and identifies the field structure in the report, including engine idle speed fluctuation pv, actual air-fuel ratio pa, noise stability coefficient po, and braking delay time ps; And fit it to a parameter name set P={pv, pa, po, ps}; The parameter name set P={pv,pa,po,ps} is mapped to a unified field parameter name p using a mapping function, and the original value and timestamp are recorded to construct a structured parameter tuple f=(pi,vi,tsi). In the formula, pi represents the parameter name of the i-th parameter, and pi∈p, vi represents the original parameter value of the i-th parameter, and tsi represents the timestamp of the i-th parameter.

3. The vehicle inspection key parameter inspection system based on document comparison according to claim 2, characterized in that: The data cleaning and normalization unit performs validity checks on the obtained structured parameter tuple f = (pi, vi, tsi), including removing out-of-limit values ​​and correcting abnormal characters, and constructs a valid dataset fa; The method for removing out-of-limit values ​​is as follows: For each type of parameter, the engineering boundary intervals that appear are pre-configured. The intervals are derived from the range of the testing equipment, the range of the vehicle's operating conditions, and the upper and lower limits of the historical statistical distribution. Then, the structured parameter tuples are traversed one by one; the original parameter value in each structured parameter tuple f is compared with the physical range of the corresponding parameter; if the original parameter value is lower than the minimum allowable value or higher than the maximum allowable value, it is determined to be an out-of-limit value. Records deemed to have exceeded the limit will be deleted or marked as invalid and removed. The method for correcting abnormal characters is as follows: The first step is field type checking: determine whether the numeric field contains non-numeric characters or illegal symbols; The second step is character cleanup: extracting content containing unit characters or symbols, keeping the numerical part, and removing the characters "ms", "dB", and "%". The third step is to identify missing data: all "space", "-", "N / A", and "null" are uniformly identified as missing data. The fourth step is to handle missing values: either delete the missing data or impute it with values ​​from adjacent time points. Step 5, format rewriting: unify all numerical values ​​into floating-point format and write them; The obtained valid dataset fa is normalized using the max-min normalization method to obtain the structured parameter set F.

4. The vehicle inspection key parameter inspection system based on document comparison according to claim 3, characterized in that: The parameter history sequence construction module includes a parameter clustering and missing value screening unit and a time series ascending order reconstruction and fluctuation correction unit; The parameter clustering and missing value screening unit first classifies the structured parameter set F according to the VIN number to ensure that each group of data contains only the inspection records of the same vehicle; in the data of each vehicle, it is classified a second time according to the parameter name p, and each group of classification contains only the timestamps of different inspections with the same parameter name p. For the structured parameter set F after classification, delete the data without numerical values ​​and remove duplicate records under the same timestamp to obtain the original time series structure Srawi={(vi(t1),t1),(vi(t2),t2),...,(vi(tk),tk)}, where vi(tk) represents the original parameter value of the i-th parameter at time tk; The time series ascending order reconstruction and fluctuation correction unit sorts the original time series structure Srawi of each parameter in ascending order according to the timestamp ts, and obtains the standard vector form vSi=[vi(ts1),vi(ts2),...,vi(tsk)], where ts1<ts2<...<tsk; In the formula, vi(tsk) represents the original parameter value of the i-th parameter at timestamp tsk; The standard vector form vSi is subjected to fluctuation correction by the extreme slope mutation determination method, and the abnormal jump values ​​are smoothed to obtain the time series S of vehicle parameter P. The formula for determining extreme slope mutations is as follows: ； In the formula, vi(tj) represents the detected value of parameter pi in the j-th time segment, (tj+1) represents the timestamp of the (j+1)-th test, vi(tj+1) represents the detected value of parameter pi in the (j+1)-th time segment, and Δi(j) represents the slope of the change of the i-th parameter in the j-th time segment. If the absolute value of the slope Δi(j) of the i-th parameter in the j-th time segment is greater than the mean ± 3 times the standard deviation, it is marked as an anomalous jump.

5. The vehicle inspection key parameter inspection system based on document comparison according to claim 4, characterized in that: The parameter distribution entropy calculation module includes a binning modeling and historical distribution probability generation unit and a current value integration and entropy difference calculation unit; The binning modeling and historical distribution probability generation unit discretizes the time series S into a distribution form, sets the number of bins m, and obtains the bin width wi and the bin interval Bj; Find the minimum value minvi and the maximum value maxvi in ​​the time series S; minvi=min(S), maxvi=max(S); The width wi of each bin is obtained by dividing the difference between the maximum value maxvi and the minimum value minvi by the number of bins m. The binning interval Bj = [minvi + (j-1)wi, minvi + (j)wi), j = 1, 2, ..., m; Perform binning operation on all original parameter values ​​vi∈S, count the number of samples nj that fall into each bin interval Bj; calculate the frequency of occurrence for each bin as the probability pij; The probability pij is obtained by the ratio of the number of samples nj falling into each bin interval Bj to the number of samples in the time series.

6. The vehicle inspection key parameter inspection system based on document comparison according to claim 5, characterized in that: The current value integration and entropy difference calculation unit maps the current value Dv of the detected motor vehicle parameter to the bin position in the historical values, places it into the corresponding bin interval Bj, and obtains the current sample number Nnj; the current sample number Nnj is equal to the sample number nj+1; Calculate the new probability npij after the current value Dv of the motor vehicle parameter falls into the bin interval Bj; By analyzing the probability pij and the new probability npij, we obtain the historical information entropy hisH and the current information entropy DH. The historical information entropy hisH is obtained as follows: traverse each bin: starting from the 1st bin, process sequentially up to the mth bin; extract the probability value of each bin: for the jth bin, obtain its probability of occurrence in historical samples; multiply the probability value by its own natural logarithm; sum the products of all bins; add a negative sign to obtain the information entropy; finally, take the negative of the sum to obtain the historical information entropy hisH of the parameter in historical samples. The current information entropy DH is obtained as follows: based on the original bin probabilities, add the current detection value and update the probability of each bin; re-traverse all bins and extract the current probability value; multiply the current probability value by the natural logarithm; then, sum the products of all bins and take the negative sign; thus, the current information entropy DH is obtained. By analyzing the historical information entropy hisH and the current information entropy DH, the parameter entropy difference ΔH is obtained; The parameter entropy difference ΔH is obtained by the difference between the current information entropy DH and the historical information entropy hisH.

7. The vehicle inspection key parameter inspection system based on document comparison according to claim 6, characterized in that: The entropy difference screening and fluctuation index construction module includes a parameter entropy difference normalization processing and anomaly screening unit and a vehicle entropy fluctuation index calculation unit. The parameter entropy difference normalization and anomaly screening unit summarizes the parameter entropy difference ΔH of all detected vehicles to obtain the entropy difference set FH={ΔH1, ΔH2, ..., ΔHN}; where N represents the total number of detected vehicles. The parameter entropy difference ΔH for each vehicle is normalized as follows: Zi = ΔHi / σΔHi; where ΔHi represents the parameter entropy difference ΔH of the i-th parameter, σΔHi represents the parameter standard deviation of the i-th parameter, and Zi represents the normalized parameter entropy difference of the i-th parameter. The parameter entropy difference Zi of the i-th parameter after normalization is compared with the preset entropy difference threshold Tz to filter out the parameter set Nab that exceeds the natural fluctuation. The filtering method is as follows: Nab={pi∣∣Zi∣>Tz}, where the entropy difference threshold Tz is 2.

0.

8. The vehicle inspection key parameter inspection system based on document comparison according to claim 7, characterized in that: The vehicle entropy fluctuation index calculation unit integrates the parameter entropy difference zi of all the normalized i-th parameter in the parameter set Nab that exceeds the natural fluctuation, and calculates and obtains the vehicle entropy fluctuation index Ψwave. The vehicle entropy fluctuation index Ψwave is obtained using the following formula: ； In the formula, M represents the total number of parameters for which the entropy difference of the current vehicle is calculated in this detection.

9. The vehicle inspection key parameter inspection system based on document comparison according to claim 8, characterized in that: The anomaly inspection output module includes an inspection object determination and anomaly parameter set extraction unit and an inspection output generation and processing suggestion push unit. The inspection object determination and abnormal parameter set extraction unit compares the acquired vehicle entropy fluctuation index Ψwave with the set vehicle fluctuation threshold Twav to obtain abnormal parameters; The abnormal parameters are obtained as follows: When the vehicle entropy fluctuation index Ψwave ≤ vehicle fluctuation threshold Twav, it indicates that the parameters are normal and no marking is made. When the vehicle entropy fluctuation index Ψwave > the vehicle fluctuation threshold Twav, it indicates that the parameter is abnormal and is marked. Fit all the parameters of the marked anomalies to obtain the anomaly parameter set Pabn; The inspection output generation and processing suggestion push unit generates a structured inspection pointer data structure Ovin=(Ψwave, Pabn, inspection level, processing suggestion) based on the results of the abnormal parameter set Pabn. Inspection levels and handling recommendations can be obtained through the following methods: When the vehicle fluctuation threshold Twav+1 > the vehicle entropy fluctuation index Ψwave > the vehicle fluctuation threshold Twav, it indicates the first inspection level, and the handling suggestion is: it is recommended to perform a retest of the braking system response. When the vehicle fluctuation threshold Twav+3 ≥ the vehicle entropy fluctuation index Ψwave ≥ the vehicle fluctuation threshold Twav+1, it indicates the second inspection level, and the handling suggestion is: it is recommended to randomly check the structural integrity or sound insulation components. When the vehicle entropy fluctuation index Ψwave > the vehicle fluctuation threshold Twav+3, it indicates the third inspection level, and the recommended action is to conduct engine bench re-inspection and status tracking.

10. A method for inspecting key parameters of motor vehicle inspection based on document comparison, applied to the motor vehicle inspection key parameter inspection system based on document comparison as described in any one of claims 1 to 9, characterized in that: Includes the following steps: Step 1: The inspection file structure parsing module receives the current inspection file and the historical inspection record file, parses them into a standard structure table, and constructs a structured parameter set F; Step 2: The parameter history sequence construction module uses the structured parameter set F as a basis to construct a cross-inspection cycle isoparameter sequence for each motor vehicle, removes missing values ​​and arranges them in ascending order of time to construct the time series S of vehicle parameters. Step 3: The parameter distribution entropy calculation module performs probability distribution modeling on the time series S of vehicle parameters, and uses equidistant binning and frequency method to calculate and obtain the historical information entropy Hihist, the current value of which is incorporated into the entropy Hicurr, and the parameter entropy difference ΔH. Step 4: The entropy difference screening and fluctuation index construction module identifies parameters whose entropy difference ΔH exceeds the natural fluctuation range and constructs the vehicle entropy fluctuation index Ψwave. Step 5: The anomaly inspection output module compares the vehicle entropy fluctuation index Ψwave with the set entropy wave threshold Twav, obtains the anomaly parameter set Pabn, and generates a specific processing strategy.

Images