Vehicle signal completion method and device, electronic equipment and readable storage medium

Abstract

This application provides a vehicle signal completion method, apparatus, electronic device, and computer-readable storage medium. The method includes: acquiring real-time operating data of a target vehicle, the real-time operating data including: a current fault code, a target signal to be completed, and at least one context-related signal related to the target signal; based on the current fault code, searching for at least one similar historical case from a fault case library of the same vehicle series; for each similar historical case, performing time-series alignment processing on the historical target signal segments in the similar historical case based on the context-related signal sequence centered on the missing time period of the target signal, to obtain aligned historical target signal segments; and generating a completion signal for the target vehicle based on all aligned historical target signal segments.

Description

Technical Field

[0001] This application relates to the field of information processing technology, and in particular to a vehicle signal completion method, apparatus, electronic device, and readable storage medium. Background Technology

[0002] With the rapid development of intelligent connected vehicles, the scale and complexity of data generated during vehicle operation have increased dramatically. This data is the core basis for achieving accurate vehicle status monitoring, fault diagnosis, and early warning.

[0003] However, in actual vehicle communication and data acquisition processes, data loss and anomalies caused by factors such as signal interference, sensor failure, and unstable network transmission are becoming increasingly prominent, seriously affecting the accuracy and reliability of subsequent data analysis and decision-making.

[0004] Therefore, how to efficiently and accurately complete missing or abnormal vehicle signals has become a key technical challenge in improving the level of intelligent vehicle diagnosis and remote monitoring. Summary of the Invention

[0005] The purpose of this application is to provide a vehicle signal completion method, apparatus, electronic device, and readable storage medium that can efficiently and accurately complete missing or abnormal vehicle signals.

[0006] In a first aspect, embodiments of this application provide a vehicle signal completion method, the method comprising: Acquire real-time operating data of the target vehicle. The real-time operating data includes: current fault codes, target signals to be completed, and at least one context-related signal associated with the target signal. Based on the current fault code, search for at least one similar historical case from the fault case database of the same vehicle series; wherein, the fault case database of the same vehicle series includes multiple historical cases, and each historical case includes: historical fault code, historical environmental data, historical target signal and corresponding historical context associated signal; For each similar historical case, based on the context-related signal sequence centered on the missing period of the target signal, the historical target signal segments in the similar historical cases are time-series aligned to obtain the aligned historical target signal segments. Based on all aligned historical target signal segments, a completion signal for the target vehicle is generated.

[0007] Secondly, embodiments of this application provide a vehicle signal completion device, which includes: The acquisition module is used to acquire the real-time operating data of the target vehicle. The real-time operating data includes: the current fault code, the target signal to be completed, and at least one context-related signal related to the target signal. The search module is used to search for at least one similar historical case from the fault case library of the same vehicle series based on the current fault code; wherein, the fault case library of the same vehicle series includes multiple historical cases, and each historical case includes: historical fault code, historical environmental data, historical target signal and corresponding historical context associated signal; The alignment module is used to perform temporal alignment processing on historical target signal segments in similar historical cases based on the context-related signal sequence centered on the missing time period of the target signal, so as to obtain aligned historical target signal segments. The generation module is used to generate a complete signal for the target vehicle based on all aligned historical target signal segments.

[0008] Thirdly, embodiments of this application provide an electronic device including a processor and a memory, the memory storing programs or instructions executable on the processor, the programs or instructions, when executed by the processor, implementing the steps of the method described in the first aspect.

[0009] Fourthly, embodiments of this application provide a readable storage medium on which a program or instructions are stored, which, when executed by a processor, implement the steps of the method described in the first aspect.

[0010] Fifthly, embodiments of this application provide a chip, the chip including a processor and a communication interface, the communication interface being coupled to the processor, the processor being used to run programs or instructions to implement the method as described in the first aspect.

[0011] In a sixth aspect, embodiments of this application provide a computer program product stored in a storage medium, which is executed by at least one processor to implement the method described in the first aspect.

[0012] In the embodiments of this application, real-time operating data of the target vehicle is acquired. This real-time operating data includes: the current fault code, the target signal to be completed, and at least one context-related signal associated with the target signal. Based on the current fault code, at least one similar historical case is searched from the fault case library of the same vehicle series. The fault case library of the same vehicle series includes multiple historical cases, each of which includes: historical fault code, historical environmental data, historical target signal, and corresponding historical context-related signal. This allows the completion process to utilize the collective experience accumulated by a large number of vehicles of the same series under the same fault, ensuring that even in scenarios where the target vehicle's own historical data is sparse or severely missing, there is still sufficient and highly relevant reference data, thereby improving the feasibility and basic accuracy of completion under data-scarce conditions. For each similar historical case, the historical target signal segment in the similar historical case is time-series aligned based on the context-related signal sequence centered on the missing period of the target signal, resulting in an aligned historical target signal segment. This fully considers the contextual association of vehicle state changes in the time dimension, effectively avoiding signal abrupt changes or phase misalignments caused by directly using historical data due to differences in driving behavior and operating rhythm. Therefore, the completed signal can achieve a smooth transition at the boundaries before and after the missing time period, ensuring the temporal continuity of the repair result. Based on all aligned historical target signal segments, a completed signal for the target vehicle is generated. It can comprehensively consider information provided by multiple similar cases. By fusing multiple aligned signal segments, noise or specific fluctuations that may exist in individual historical cases can be naturally smoothed out, thereby extracting more universal and reliable signal change patterns. This makes the output completed signal more likely to approximate the vehicle's true state during the missing time period, achieving high-precision, highly continuous, and physically reliable vehicle signal completion. Attached Figure Description

[0013] Figure 1 This is a flowchart of a vehicle signal completion method provided in an embodiment of this application; Figure 2 This is a schematic diagram of a vehicle signal completion method provided in an embodiment of this application; Figure 3 This is a structural diagram of the vehicle signal completion device provided in the embodiments of this application; Figure 4 This is a structural diagram of the electronic device provided in the embodiments of this application. Detailed Implementation

[0014] To make the technical problems, technical solutions, and beneficial effects solved by this application clearer, the following detailed description is provided in conjunction with embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0015] The vehicle signal completion provided in this application embodiment can be applied to at least the following application scenarios, which will be described below.

[0016] In traditional remote fault diagnosis, when a vehicle reports a fault code, the diagnostic center often only receives intermittent or completely lost associated signals. This makes it impossible for technicians to accurately determine whether the fault is due to transient interference, poor wiring contact, or complete sensor failure. Technicians can only rely on limited information and experience to make inferences, which may lead to misdiagnosis, unnecessary component replacement, or incomplete repairs.

[0017] For autonomous vehicles, the safety of the system is extremely dependent on the continuity and reliability of sensor data. For example, if the inertial measurement unit (IMU), which is responsible for providing the vehicle's yaw rate, experiences a millisecond-level signal loss, it may cause the vehicle control system to misjudge its own attitude, leading to danger.

[0018] In large-scale fleet operations such as logistics and taxis, managers rely on continuous vehicle data monitoring to assess engine health and driving behavior for fuel consumption management and preventative maintenance. However, communication interruptions and data packet loss frequently occur in remote areas or complex electromagnetic environments, creating monitoring blind spots.

[0019] The technical terms used in the embodiments of this application will be introduced below.

[0020] Real-time operational data: refers to the set of data acquired in real-time or uploaded near real-time from the target vehicle's onboard sensors, controllers, and communication network at the moment the vehicle signal completion task is triggered. This data set is the logical input for performing signal completion and specifically includes: Current fault code: This refers to a standardized fault identification code generated and reported in real time by the vehicle's on-board diagnostic system when it detects an abnormality in a specific system or component. This code uniquely corresponds to a specific fault type. For example, P0335 represents "crankshaft position sensor A circuit fault," and its occurrence is the initial condition that triggers the signal completion process.

[0021] The target signal to be supplemented refers to the specific vehicle status signal that is missing, distorted, or abnormal in the current time period due to sensor failure, communication interference, or data packet loss, and needs to be repaired and filled. This signal is the core object of the supplementation operation and is usually a time-series continuous variable with a clear physical meaning, such as engine speed, vehicle speed, and battery cell voltage.

[0022] Context-related signals: These refer to one or more normally functioning sensor signals or controller internal state variables that are highly correlated with the target signal to be completed in terms of vehicle system dynamics, control logic, or statistical characteristics at the completion time. These signals collectively constitute the environmental information for understanding how the target signal should change, and their role is to provide a basis for inferring the value of the target signal.

[0023] In response to the problems in related technologies, embodiments of this application provide a vehicle signal completion method, apparatus, electronic device, and computer-readable storage medium, which can solve the problem of difficulty in efficiently and accurately completing vehicle signals in related technologies.

[0024] The vehicle signal completion method provided in this application will be described in detail below with reference to the accompanying drawings, through specific embodiments and application scenarios.

[0025] This application provides a vehicle signal completion method, including: Acquire real-time operating data of the target vehicle. The real-time operating data includes: current fault codes, target signals to be completed, and at least one context-related signal associated with the target signal. Based on the current fault code, search for at least one similar historical case from the fault case database of the same vehicle series; wherein, the fault case database of the same vehicle series includes multiple historical cases, and each historical case includes: historical fault code, historical environmental data, historical target signal and corresponding historical context associated signal; For each similar historical case, based on the context-related signal sequence centered on the missing period of the target signal, the historical target signal segments in the similar historical cases are time-series aligned to obtain the aligned historical target signal segments. Based on all aligned historical target signal segments, a completion signal for the target vehicle is generated.

[0026] In the embodiments of this application, real-time operating data of the target vehicle is acquired. This real-time operating data includes: the current fault code, the target signal to be completed, and at least one context-related signal associated with the target signal. Based on the current fault code, at least one similar historical case is searched from the fault case library of the same vehicle series. The fault case library of the same vehicle series includes multiple historical cases, each of which includes: historical fault code, historical environmental data, historical target signal, and corresponding historical context-related signal. This allows the completion process to utilize the collective experience accumulated by a large number of vehicles of the same series under the same fault, ensuring that even in scenarios where the target vehicle's own historical data is sparse or severely missing, there is still sufficient and highly relevant reference data, thereby improving the feasibility and basic accuracy of completion under data-scarce conditions. For each similar historical case, the historical target signal segment in the similar historical case is time-series aligned based on the context-related signal sequence centered on the missing period of the target signal, resulting in an aligned historical target signal segment. This fully considers the contextual association of vehicle state changes in the time dimension, effectively avoiding signal abrupt changes or phase misalignments caused by directly using historical data due to differences in driving behavior and operating rhythm. Therefore, the completed signal can achieve a smooth transition at the boundaries before and after the missing time period, ensuring the temporal continuity of the repair result. Based on all aligned historical target signal segments, a completed signal for the target vehicle is generated. It can comprehensively consider information provided by multiple similar cases. By fusing multiple aligned signal segments, noise or specific fluctuations that may exist in individual historical cases can be naturally smoothed out, thereby extracting more universal and reliable signal change patterns. This makes the output completed signal more likely to approximate the vehicle's true state during the missing time period, achieving high-precision, highly continuous, and physically reliable vehicle signal completion.

[0027] Example 1 This application provides a vehicle signal completion method. Please refer to [link / reference]. Figure 1 This includes the following steps: Step 110: Obtain the real-time operating data of the target vehicle, the real-time operating data including: the current fault code, the target signal to be completed, and at least one context-related signal related to the target signal; Real-time operational data serves as the fundamental input for triggering and executing the completion task. This data includes the current fault code, the target signal to be completed, and at least one context-related signal associated with the target signal. The current fault code is a standardized code detected and reported in real-time by the vehicle's on-board diagnostic system, indicating an abnormality in a specific system or component, such as the P0335 code indicating a crankshaft position sensor circuit malfunction. The target signal to be completed refers to a specific vehicle status signal that requires repair due to data loss, distortion, or transmission interruption at the time of the current fault, such as engine speed, battery voltage, or vehicle speed.

[0028] Context-related signals refer to other normal operating signals that are closely related to the state changes of the target signal in the vehicle system's operating logic, physical coupling relationship, or control strategy. For example, when the target signal is engine speed, the accelerator pedal opening signal that reflects the driver's power request and the intake manifold pressure signal that reflects the engine's intake state can be used as its context-related signals. Similarly, when the target signal is the voltage of a single cell of the power battery, signals such as the total battery voltage, battery pack temperature, and vehicle load current can be used as its context-related signals.

[0029] Step 120: Based on the current fault code, search for at least one similar historical case from the fault case library of the same vehicle series; wherein, the fault case library of the same vehicle series includes multiple historical cases, and each historical case includes: historical fault code, historical environmental data, historical target signal and corresponding historical context association signal; Based on the current fault code, at least one similar historical case is searched from the fault case database of the same vehicle series. The fault case database of the same vehicle series is a pre-built database stored in the cloud, which aggregates complete data packages of fault events recorded by numerous vehicles of the same model series during historical operation. Each historical case is a data package, containing historical fault codes, historical environmental data, historical target signals, and corresponding historical contextual signals. Historical fault codes and current fault codes share the same code system, used to identify the same fault type. Historical environmental data records the external and internal environmental conditions of the vehicle at the time of the fault, such as the ambient temperature, atmospheric pressure or altitude, and the vehicle's average speed range.

[0030] Historical target signals and their corresponding historical context signals completely record the original time-series data of relevant signals within a time window before and after the occurrence of the historical fault event. When a target vehicle experiences a specific fault resulting in signal loss, the cloud may have already stored a large amount of historical context data containing complete signals for the same model under the same fault codes. Preliminary retrieval using fault codes can quickly locate a group of potential reference cases with the same fault background. This lays the data foundation for subsequent refined matching and completion, allowing us to move beyond the limitations of historical data from a single vehicle and instead utilize the "experience" of a group of vehicles to address the current problem.

[0031] Step 130: For each of the similar historical cases, based on the context-related signal sequence centered on the missing time period of the target signal, perform time-series alignment processing on the historical target signal segments in the similar historical cases to obtain aligned historical target signal segments; For each similar historical case found, the historical target signal segment in the real-time operating data of the target vehicle, centered on the period when the target signal is missing, is used as a benchmark. This process performs time-series alignment on the historical target signal segment to obtain the aligned historical target signal segment. Although the historical case and the current vehicle reported the same fault code, due to subtle differences in driving behavior and specific operating conditions, the vehicle state change curves before and after the fault are not completely synchronized on the time scale.

[0032] For example, the current vehicle may have accelerated rapidly before the malfunction occurred, while a historical case might have shown gradual acceleration. This causes the throttle opening signal waveforms of both cases to be compressed or stretched on the time axis. Directly using historical target signal segments from historical cases for splicing or averaging would result in timing misalignment. Therefore, timing alignment processing is performed on historical target signal segments from similar historical cases to obtain aligned historical target signal segments. This allows historical target signal segments from different cases to be uniformly mapped onto a unified time reference centered on the missing time period of the current vehicle, eliminating timing differences caused by different operating conditions and enabling effective signal fusion subsequently.

[0033] Step 140: Generate a completion signal for the target vehicle based on all the aligned historical target signal segments.

[0034] Based on all historical target signal segments that have undergone time-alignment processing, a complete signal for the target vehicle is generated. Each aligned historical segment provides a reference trajectory showing how the target signal might change under similar faults and contexts. By combining multiple such reference trajectories, a complete signal curve that most likely reflects the current vehicle's true condition can be generated.

[0035] One approach is to perform a weighted average of all aligned segments, with weights determined by the morphological similarity between the context sequence of the source historical cases and the current baseline sequence. Another approach is to use these aligned segments as training samples, inputting them into a lightweight regression model that, conditioned on the current real-time context, outputs the final completed sequence. This fusion mechanism effectively smooths out potential randomness or noise in individual historical cases, extracts common features from multiple cases, and generates a more stable and reliable completed signal to fill in the gaps in the target vehicle's signal.

[0036] In some possible embodiments, step 120 may specifically include the following steps: From the fault case database of the same vehicle series, similar historical cases that meet the environmental matching conditions of the historical environmental data are selected; The environmental matching conditions include an ambient temperature range and an altitude threshold.

[0037] Based on the current fault codes, all potential historical cases with the same fault type are initially screened. Building upon this, environmental matching conditions are introduced to refine the screening of candidate cases, improving the similarity between the retrieved cases and the current operating environment of the target vehicle. This provides a higher-quality reference for subsequent signal completion. The environmental matching conditions refer to a set of preset rules or thresholds used to quantitatively assess whether the environmental state at the time of the historical case is similar to the current vehicle environmental state.

[0038] The environmental matching criteria specifically include two dimensions that can be applied independently or in combination: ambient temperature range and altitude threshold. The ambient temperature range refers to a closed interval centered on the current or fault-prone ambient temperature of the target vehicle, extending upwards and downwards by a certain range. For example, if the current ambient temperature of the target vehicle is 25 degrees Celsius, and the preset fluctuation range is ±5 degrees Celsius, then the ambient temperature range is 20 to 30 degrees Celsius. The altitude threshold is a value used to limit the highest altitude at which historical cases occurred. For example, it can be set to 1500 meters, meaning that only historical cases recorded in the environmental data with an altitude below this threshold are selected.

[0039] The performance and signal characteristics of many vehicle components are significantly affected by external environmental physical conditions. Ambient temperature directly affects engine intake air density, sensor response characteristics, and the operating state of electronic components. For example, in low-temperature environments, increased oil viscosity may cause a delay in the response of engine speed signals compared to normal temperatures. Altitude primarily affects atmospheric pressure, which in turn affects engine intake air volume, combustion efficiency, and the turbocharger's operating point, resulting in systematic deviations in the shape of engine operating parameter curves under the same fault code. Therefore, directly using all historical cases with the same fault code without considering environmental factors may introduce systematic noise due to environmental differences, reducing the applicability of the reference cases. This embodiment aims to select historical cases that are closest to the current vehicle state in key environmental dimensions from a large number of cases with the same fault code by adding environmental matching conditions.

[0040] For example, if a target vehicle malfunctions on an urban expressway with an ambient temperature of 28 degrees Celsius and an altitude of approximately 800 meters, by setting the ambient temperature range to 25-30 degrees Celsius and the altitude threshold to 1,000 meters, historical cases occurring in frigid regions of minus 10 degrees Celsius or at high altitudes of 3,000 meters can be effectively filtered out. Although these cases may have the same fault code, their signal patterns are less relevant due to the vastly different environments. During implementation, the system extracts the historical environmental data field from the historical case data package, compares the recorded temperature and altitude values ​​with the ambient temperature range and altitude threshold, and retains only historical cases that simultaneously meet both conditions as the final output of similar historical cases.

[0041] By introducing and applying environmental matching conditions—specifically, ambient temperature ranges and altitude thresholds—for secondary screening, retrieval accuracy was improved. This mechanism effectively controls environmental factors, a crucial interfering variable, ensuring that the retrieved similar historical cases not only match the current problem in fault type but also share a closer physical context. This provides a more reliable foundation for subsequent completion using historical signal fragments from these cases as references, reducing inherent deviations between historical data and current realities caused by environmental differences. Consequently, it enhances the accuracy and physical reliability of the final completed signal in representing the target vehicle's true state under specific environmental conditions.

[0042] In some possible embodiments, step 130 may specifically include the following steps: Step 210: Extract context-related signals from the real-time running data within the first time window centered on the missing period of the target signal to form a baseline context sequence; Step 220: Extract historical context association signals and historical target signals from similar historical cases, centered on the historical failure time and within a second time window of equal length to the first time window, to form a reference context sequence and a signal segment to be aligned, respectively. Step 230: Calculate the target matching path between the baseline context sequence and the reference context sequence; Step 240: Based on the temporal correspondence defined by the target matching path, map the signal segment to be aligned onto the time axis of the reference context sequence to obtain the aligned historical target signal segment.

[0043] From the real-time operating data of the target vehicle, all context-related signals within a first time window of fixed length, centered on the missing period of the target signal, are extracted. These signals are then sorted by timestamp and combined to form a baseline context sequence representing the specific situation when the current vehicle problem occurs. For example, if the missing period of the target signal is a specific 10-second interval, the first time window can be set to extend 5 seconds before and after the missing period, for a total length of 20 seconds, and sampling points of all context-related signals such as throttle opening and intake pressure within this window are extracted.

[0044] From similar historical cases currently being processed, data is extracted from a second time window centered on the recorded historical fault occurrence time of that case, with a duration identical to the first time window. From this window's data, historical context-related signals and historical target signals are separated; the former forms a reference context sequence, and the latter forms a signal segment to be aligned. The reference context sequence and the baseline context sequence correspond completely in signal type and time length for comparison.

[0045] By constructing a distance matrix to measure the differences between points in the baseline context sequence and the reference context sequence, and finding a path with the minimum cumulative distance from the start point to the end point of the matrix, this path is called the target matching path. The target matching path explicitly defines which one or more time points in the reference context sequence each time point in the baseline context sequence corresponds to, thus characterizing the optimal morphological correspondence between the two waveforms.

[0046] For example, even if the accelerator pedal is pressed rapidly in the current vehicle case but slowly in historical cases, the system can find the best-matching waveform shape between the two, allowing the reference sequence to be stretched or compressed on the time axis to fit the rhythm of the baseline sequence. Specifically, a dynamic time warping algorithm can be used to calculate the target matching path between the baseline context sequence and the reference context sequence. The dynamic time warping algorithm is a method that can effectively handle nonlinear scaling or speed differences between two time series on the time axis.

[0047] Based on the temporal correspondence defined by the target matching path, the signal segment to be aligned is mapped onto the time axis of the reference context sequence to obtain the aligned historical target signal segment. Specifically, the target matching path defines the mapping position of each sampling point in the reference context sequence and its corresponding signal segment to be aligned to the time axis of the reference context sequence. According to this mapping relationship, the values ​​of the signal segment to be aligned, i.e., the historical target signal, can be redistributed to the timestamps of the reference sequence. For non-one-to-one mappings, interpolation calculations are required to generate new values. After this operation, the historical target signal segment, which originally belonged to the historical case time axis, is transformed to the time axis synchronized with the current vehicle's reference context sequence, and its waveform phase and change rhythm are aligned with the current vehicle's context.

[0048] By accurately extracting the baseline and reference context sequences and utilizing a dynamic time warping algorithm to calculate the target matching path, this method can intelligently identify and compensate for temporal nonlinear differences between the current vehicle and historical cases caused by variations in driver operating habits, vehicle load, or traffic conditions. Finally, based on this path, historical target signal segments are time-mapped, ensuring that key feature points from different historical cases, such as peaks, valleys, and trend inflection points, are precisely aligned temporally with the context of the current vehicle's missing time period before fusion. This process fundamentally solves the waveform phase distortion and feature confusion problems inevitably caused by directly using unaligned historical signals for fusion, providing high-quality and temporally consistent input material for generating smooth, coherent, and physically correct completion signals. This is a crucial step in improving the accuracy of the entire completion method.

[0049] In some possible embodiments, step 140 may specifically include the following steps: Calculate the morphological similarity between the reference context sequence of the similar historical cases and the baseline context sequence to obtain the corresponding similarity score; Based on the similarity scores of each similar historical case, all aligned historical target signal segments are weighted and fused to generate the complete signal.

[0050] Morphological similarity specifically refers to the degree of agreement between two time series in terms of overall waveform shape, trend of change, and distribution of key feature points. It focuses on the matching of curve shapes, rather than just the absolute numerical difference at a certain moment. Calculating a similarity score quantifies how close the reference context provided by each historical case is to the current real-world vehicle context. One specific implementation method is to calculate the score by taking the reciprocal of the cumulative distance between the two sequences after dynamic time-normalized alignment; the smaller the distance, the higher the score, indicating greater morphological similarity.

[0051] Alternatively, the features of two sequences can be calculated at multiple scales, such as the cosine similarity between feature vectors extracted by wavelet transform. For example, if the current vehicle's reference context sequence shows a pattern of rapid acceleration followed by stable cruising, while a historical case's reference context sequence shows a pattern of slow acceleration, then the pattern similarity score between the two will be lower; conversely, if another historical case also shows similar rapid acceleration features, then its score will be higher.

[0052] Based on the similarity scores of each similar historical case, all aligned historical target signal segments are weighted and fused to generate the final completed signal. Weighted fusion refers to a method of generating an output signal by linearly or nonlinearly combining input signals according to their weights. In this step, each aligned historical target signal segment is an input signal, and its corresponding weight is determined by the similarity score of its source case. A specific fusion implementation method is to first normalize all similarity scores so that their sum is one; these normalized values ​​are used as the fusion weights for each segment.

[0053] On a unified, aligned timeline, the values ​​of all historical target signal segments at each time point are weighted and summed according to their corresponding weights to obtain the completed signal value at that time point, ultimately forming a complete completed signal curve. For example, if there are three similar historical cases with normalized weights of 0.5, 0.3, and 0.2 respectively, then at a certain moment in the completed signal, its value is equal to the value of the first case signal at that moment multiplied by 0.5, plus the corresponding value of the second case multiplied by 0.3, plus the corresponding value of the third case multiplied by 0.2. Another implementation method is that the weights are not only used for numerical weighting but also for selective fusion, for example, averaging only a few high-quality segments with weights higher than a certain threshold to exclude interference from low-similarity cases.

[0054] By calculating morphological similarity scores, the system can automatically assess and quantify the relevance of each historical reference case to the current context, thus providing an objective and differentiated trust index for subsequent fusion. Weighted fusion based on these scores enables intelligent and comprehensive utilization of multi-source historical information, ensuring that information from historical cases highly similar to the current context dominates the final completion result, while the influence of information from cases significantly different from the current context is weakened. This mechanism effectively extracts common change patterns most likely reflecting the true situation from multiple potentially inconsistent reference trajectories, while suppressing accidental noise or specific biases that may be present in individual cases. Therefore, the final completed signal not only fills in data gaps but also exhibits better smoothness and coherence, and has a higher probability of approximating the true physical state that the target vehicle should have exhibited during the missing time period.

[0055] In some possible embodiments, a predefined numerical boundary is obtained for the target signal; The values ​​in the candidate completion signals that exceed the numerical boundary are corrected to the nearest boundary value within the boundary, so as to obtain and output the final completion result.

[0056] A numerical boundary is a closed interval defined by vehicle design parameters, physical principles, or safety specifications. It defines the reasonable range of values ​​that a target signal is allowed to exhibit under any possible operating conditions, typically including a minimum and a maximum value. For example, for an engine speed signal, its numerical boundary can be set as the interval between the minimum and maximum permissible speeds; for a battery voltage signal, its boundary can be set as the interval between the safe discharge cut-off voltage and the charging cut-off voltage specified by the battery manufacturer. These boundaries can be obtained by querying a pre-set vehicle parameter database or by calling the corresponding boundary calculation model based on the signal type.

[0057] After defining the numerical boundaries, the candidate completion signals are verified and corrected point by point. Specifically, the value of each data point in the candidate completion signal sequence is sequentially traversed and compared with the defined numerical boundaries. If the value falls within the range specified by the boundaries, it is retained.

[0058] If the value exceeds the boundary range, it is corrected to the "nearest boundary value" within that boundary interval. The nearest boundary value refers to the value that is pulled back to the nearest boundary limit. For example, if the engine speed boundary is [600, 6500] rpm (unit: revolutions per minute), and a point in the candidate completion signal has a value of 6700 rpm, since it exceeds the maximum value of 6500, this point is corrected to 6500 rpm; similarly, if a point has a value of 550 rpm, which is lower than the minimum value of 600, it is corrected to 600 rpm. This process traverses the entire candidate completion signal sequence, ensuring that all points are processed, and finally outputs a corrected signal in which all values ​​are strictly within the predefined reasonable range, as the final completion result.

[0059] By introducing numerical boundary constraints based on prior knowledge, a global filtering and pruning process is performed on the completion results in a direct and reliable manner. This forces any values ​​exceeding the actual operating capabilities of the vehicle system back into the feasible region, fundamentally eliminating the possibility of generating physically unreliable signals. This not only makes the final completed output directly applicable to diagnosis or control in engineering but also greatly enhances the robustness and reliability of the entire completion method when faced with complex, sparse, or conflicting historical data.

[0060] In some possible embodiments, step 140 may specifically include the following steps: Based on all the aligned historical target signal segments, a candidate completion signal for the target vehicle is generated; The candidate completion signal is verified according to the physical constraint rules to identify abnormal data segments; the physical constraint rules are at least one predefined constraint rule for the target signal. For the abnormal data segment, corrections are made based on the boundary values ​​or change trends corresponding to the physical constraint rules to obtain the complete signal for the target vehicle.

[0061] Based on all aligned historical target signal segments, a preliminary candidate completion signal for the target vehicle is generated through fusion calculation. This candidate completion signal is an intermediate result directly derived from historical data fusion, and its reliability requires further verification. The generation method can be weighted averaging, median filtering, or other signal fusion algorithms.

[0062] Based on predefined physical constraint rules for the target signal, the candidate completion signal is systematically verified to identify any potentially anomalous data segments. Physical constraint rules are one or more rules abstracted from fundamental physical laws such as vehicle system dynamics, component mechanical limits, or energy conservation, used to determine the rationality of a signal sequence or its segments. Anomalous data segments refer to continuous or discrete data intervals in the candidate completion signal that violate one or more physical constraint rules.

[0063] For example, a physical constraint rule for engine speed signals might stipulate that "when the clutch is engaged and the gear is not in neutral, the engine speed decreasing as the throttle opening increases" is an impossible event; another rule might stipulate that "the rate of increase of engine speed per unit time must not exceed the threshold corresponding to the engine's maximum acceleration capability." The system will scan candidate signals according to these rules and mark all segments that violate the rules as abnormal data segments.

[0064] After identifying anomalous data segments, these segments are corrected based on the permissible boundary values ​​or reasonable trends of change corresponding to the physical constraints they violate. The purpose of correction is not simple deletion, but rather to adjust the anomalous segments to a reasonable range based on the physical principles inherent in the rules. If the rule corresponds to a specific boundary value, the anomalous data is corrected to within that boundary; if the rule corresponds to a trend, the signal curve may be smoothly redrawn based on that limit. For example, for an anomalously steep speed increase curve that violates the maximum speed change rate rule, the system will limit the slope of that segment based on the maximum permissible change rate and recalculate a gently rising curve to replace the original anomalous segment. After correcting all anomalous data segments, a physically reliable final completion signal for the target vehicle is obtained.

[0065] By integrating statistical patterns from historical data, a physical rule verification step reviews and constrains the statistical results. It effectively identifies and corrects completion errors that might arise from relying solely on data-driven approaches and violate fundamental physical laws, such as generating impossible instantaneous energy mutations or signal sequences that violate causal relationships. This improves the reliability and usability of the completion results in harsh engineering environments. The final output completion signal not only fills in the missing information but also ensures that the described state change process is possible in the physical world, thus enabling the signal to be safely and effectively used for subsequent vehicle state analysis, fault diagnosis, or control decisions.

[0066] In some possible embodiments, the following steps may also be included before step 130: If the sampling frequency of the target signal to be completed is different from that of the context-related signal, then the low-frequency sampling signal is uniformly resampled to a preset high-frequency time reference by an interpolation algorithm.

[0067] If the target signal to be completed has a different sampling frequency than the context-related signal. Sampling frequency refers to the number of points per second that a continuous physical signal is sampled and discretized, measured in Hertz (Hz). In vehicle networks and data acquisition systems, the sampling frequencies of signals on different buses or provided by different controllers may vary depending on design requirements. For example, wheel speed signals on the controller area network bus may be sampled at 100 Hz, while certain engine temperature parameters obtained through the diagnostic interface may be updated only at 1 Hz.

[0068] When a sampling frequency difference is confirmed, the system uses an interpolation algorithm to resample the low-frequency sampled signal to a preset high-frequency time reference. An interpolation algorithm is a mathematical method for estimating the values ​​of unknown intermediate points based on known discrete data points. Resampling refers to the process of converting the original time-series signal from one discrete time grid to another. The preset high-frequency time reference is usually the one with the highest sampling frequency among all relevant signals, or a uniform and sufficiently high time resolution specifically set according to subsequent processing requirements.

[0069] Specifically, the highest sampling frequency among all signals to be processed can be identified, and its corresponding timestamp sequence can be used as the target time reference. For signals with sampling frequencies lower than this reference, an interpolation algorithm is used to calculate the value that should be present at each new time point on the target high-frequency time reference based on its original discrete sampling point sequence, thereby generating a new signal sequence with a sampling rate consistent with the high-frequency reference. For example, a linear interpolation algorithm can be used, which assumes that the signal values ​​between adjacent known sampling points change linearly, thus directly calculating the value of the intermediate point. Alternatively, a cubic spline interpolation algorithm can be used, which constructs a set of smooth cubic polynomial curves passing through all known sampling points, thereby providing smoother and more accurate interpolation results.

[0070] By unifying all signals involved in subsequent calculations to the same and sufficiently fine time scale, it is ensured that the data points within and between sequences are strictly synchronized and comparable on the time axis when extracting the baseline context sequence and reference context sequence in step 130. This eliminates comparison errors that may be introduced due to misalignment of the original sampling time points. In addition, upscaling the low-frequency signal to a high-frequency reference also avoids the loss of key waveform details due to insufficient information density, enabling subsequent algorithms such as dynamic time warping to more accurately capture and match the microscopic morphological features of the waveform.

[0071] Example 2: After completing the temporal alignment processing of all similar historical cases and obtaining a series of historical target signal segments strictly aligned with the current vehicle time reference, this embodiment executes the following process to generate a complete signal: A dynamic fusion weight is calculated for each aligned historical target signal segment. This weight is generated using a probability distribution model based on the similarity scores of all cases. The morphological similarity scores of all similar historical cases are treated as a set of evidence and converted into a probability distribution using a softmax function with an adjustable temperature parameter. The temperature parameter controls the concentration of the weight distribution: a higher temperature parameter makes the weight distribution more uniform, reflecting smooth fusion of multiple cases; a lower temperature parameter makes the weight highly concentrated on the individual cases with the highest similarity, reflecting strong trust in the optimal case. The weight obtained for each historical target signal segment is its corresponding probability value in this probability distribution.

[0072] After determining the dynamic weights of each segment, weighted fusion is performed on a unified completion timeline. For each target time point on the timeline, the signal values ​​of all aligned historical target signal segments at that time point are multiplied by their respective dynamic weights and then summed to obtain the completed signal value for that time point. After traversing all time points, a continuous completed signal curve is generated.

[0073] In the embodiments of this application, the generation quality of the completion signal is further optimized by introducing a dynamic weight fusion mechanism based on a probability model. Through probabilistic weight allocation, a more rigorous mathematical framework is provided for the integration of multi-source historical information, enhancing the objectivity and interpretability of the process. The adjustable temperature parameter provides flexibility for the fusion strategy, enabling it to adapt to different data confidence scenarios: when the case quality is generally high, a more balanced fusion can be used to smooth noise; when a certain case is significantly better than others, a more focused fusion can be used to adopt the best information, improving the adaptability to complex and heterogeneous historical datasets.

[0074] This application also provides a vehicle signal completion device 300, please refer to... Figure 3 ,include: The acquisition module 310 is used to acquire real-time operating data of the target vehicle, the real-time operating data including: current fault code, target signal to be completed, and at least one context association signal related to the target signal; The lookup module 320 is used to look up at least one similar historical case from the same vehicle series fault case library based on the current fault code; wherein, the same vehicle series fault case library includes multiple historical cases, and each historical case includes: historical fault code, historical environmental data, historical target signal and corresponding historical context association signal; Alignment module 330 is used to perform temporal alignment processing on the historical target signal segments in the similar historical cases based on the context-related signal sequence centered on the missing time period of the target signal for each similar historical case, so as to obtain the aligned historical target signal segments. The generation module 340 is used to generate a completion signal for the target vehicle based on all the aligned historical target signal segments.

[0075] In some possible embodiments, the lookup module 320 is specifically used for: From the fault case database of the same vehicle series, similar historical cases that meet the environmental matching conditions of the historical environmental data are selected; The environmental matching conditions include an ambient temperature range and an altitude threshold.

[0076] In some possible embodiments, the alignment module 330 is specifically used for: From the real-time running data, extract the context-related signals within a first time window centered on the missing period of the target signal to form a baseline context sequence; From the similar historical cases, extract the historical context association signal and historical target signal within the second time window centered on the historical fault time and of the same length as the first time window, and respectively form the reference context sequence and the signal segment to be aligned. Calculate the target matching path between the baseline context sequence and the reference context sequence; Based on the temporal correspondence defined by the target matching path, the signal segment to be aligned is mapped onto the time axis of the reference context sequence to obtain the aligned historical target signal segment.

[0077] In some possible embodiments, the generation module 340 is specifically used for: Calculate the morphological similarity between the reference context sequence of the similar historical cases and the baseline context sequence to obtain the corresponding similarity score; Based on the similarity scores of each similar historical case, all aligned historical target signal segments are weighted and fused to generate the complete signal.

[0078] In some possible embodiments, the generation module 340 is specifically used for: Based on all the aligned historical target signal segments, a candidate completion signal for the target vehicle is generated; The candidate completion signal is verified according to the physical constraint rules to identify abnormal data segments; the physical constraint rules are at least one predefined constraint rule for the target signal. For the abnormal data segment, corrections are made based on the boundary values ​​or change trends corresponding to the physical constraint rules to obtain the complete signal for the target vehicle.

[0079] In some possible embodiments, the vehicle signal completion device 300 may further include: The sampling module is used to resample the low-frequency sampling signal to a preset high-frequency time reference by means of an interpolation algorithm if the sampling frequency of the target signal to be completed is different from that of the context-related signal.

[0080] In the embodiments of this application, real-time operating data of the target vehicle is acquired. This real-time operating data includes: the current fault code, the target signal to be completed, and at least one context-related signal associated with the target signal. Based on the current fault code, at least one similar historical case is searched from the fault case library of the same vehicle series. The fault case library of the same vehicle series includes multiple historical cases, each of which includes: historical fault code, historical environmental data, historical target signal, and corresponding historical context-related signal. This allows the completion process to utilize the collective experience accumulated by a large number of vehicles of the same series under the same fault, ensuring that even in scenarios where the target vehicle's own historical data is sparse or severely missing, there is still sufficient and highly relevant reference data, thereby improving the feasibility and basic accuracy of completion under data-scarce conditions. For each similar historical case, the historical target signal segment in the similar historical case is time-series aligned based on the context-related signal sequence centered on the missing period of the target signal, resulting in an aligned historical target signal segment. This fully considers the contextual association of vehicle state changes in the time dimension, effectively avoiding signal abrupt changes or phase misalignments caused by directly using historical data due to differences in driving behavior and operating rhythm. Therefore, the completed signal can achieve a smooth transition at the boundaries before and after the missing time period, ensuring the temporal continuity of the repair result. Based on all aligned historical target signal segments, a completed signal for the target vehicle is generated. It can comprehensively consider information provided by multiple similar cases. By fusing multiple aligned signal segments, noise or specific fluctuations that may exist in individual historical cases can be naturally smoothed out, thereby extracting more universal and reliable signal change patterns. This makes the output completed signal more likely to approximate the vehicle's true state during the missing time period, achieving high-precision, highly continuous, and physically reliable vehicle signal completion.

[0081] This application also provides an electronic device 60, please refer to... Figure 4 It includes a processor 610 and a memory 620, wherein the memory 610 is used to store computer programs; and the processor 620 is used to execute the programs stored in the memory 610 to implement the methods described in any embodiment of this application.

[0082] This application also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the method described in any embodiment of this application.

[0083] In this application, "multiple" refers to two or more.

[0084] In this application, unless otherwise expressly defined, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection between two components. Those skilled in the art can understand the specific meaning of the above terms in this application based on the specific circumstances.

[0085] The terms “first,” “second,” “third,” “fourth,” etc., in this application (if present) are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence.

[0086] In this application, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, in this application, the character " / " generally indicates that the preceding and following related objects have an "or" relationship.

[0087] Unless otherwise specified, all steps in this application may be performed sequentially or randomly. For example, if the method includes steps A and B, it means that the method may include steps A and B performed sequentially, or it may include steps B and A performed sequentially. For example, if the method may also include step C, it means that step C may be added to the method in any order. For example, the method may include steps A, B, and C, or it may include steps A, C, and B, or it may include steps C, A, and B, etc.

[0088] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A vehicle signal completion method, characterized in that, The method includes: Acquire real-time operating data of the target vehicle, the real-time operating data including: current fault code, target signal to be completed, and at least one context-related signal related to the target signal; Based on the current fault code, at least one similar historical case is searched from the fault case library of the same vehicle series; wherein, the fault case library of the same vehicle series includes multiple historical cases, and each historical case includes: historical fault code, historical environmental data, historical target signal and corresponding historical context association signal; For each of the similar historical cases, based on the context-related signal sequence centered on the missing time period of the target signal, the historical target signal segments in the similar historical cases are time-series aligned to obtain aligned historical target signal segments; Based on all the aligned historical target signal segments, a completion signal for the target vehicle is generated.

2. The method according to claim 1, characterized in that, The step of searching for at least one similar historical case from the fault case database of the same vehicle series based on the current fault code includes: From the fault case database of the same vehicle series, similar historical cases that meet the environmental matching conditions of the historical environmental data are selected; The environmental matching conditions include an ambient temperature range and an altitude threshold.

3. The method according to claim 1, characterized in that, The step of performing time-series alignment processing on historical target signal segments in the similar historical cases to obtain aligned historical target signal segments includes: From the real-time running data, extract the context-related signals within a first time window centered on the missing period of the target signal to form a baseline context sequence; From the similar historical cases, extract the historical context association signal and historical target signal within the second time window centered on the historical fault time and of the same length as the first time window, and respectively form the reference context sequence and the signal segment to be aligned. Calculate the target matching path between the baseline context sequence and the reference context sequence; Based on the temporal correspondence defined by the target matching path, the signal segment to be aligned is mapped onto the time axis of the reference context sequence to obtain the aligned historical target signal segment.

4. The method according to claim 3, characterized in that, The step of generating a completion signal for the target vehicle based on all the aligned historical target signal segments includes: Calculate the morphological similarity between the reference context sequence of the similar historical cases and the baseline context sequence to obtain the corresponding similarity score; Based on the similarity scores of each similar historical case, all aligned historical target signal segments are weighted and fused to generate the complete signal.

5. The method according to claim 1, characterized in that, The step of generating a complete signal for the target vehicle based on all the aligned historical target signal segments includes: Based on all the aligned historical target signal segments, a candidate completion signal for the target vehicle is generated; The candidate completion signal is verified according to the physical constraint rules to identify abnormal data segments; the physical constraint rules are at least one predefined constraint rule for the target signal. For the abnormal data segment, corrections are made based on the boundary values ​​or change trends corresponding to the physical constraint rules to obtain the complete signal for the target vehicle.

6. The method according to claim 1, characterized in that, Before performing time-series alignment processing on the historical target signal segments in the similar historical cases to obtain the aligned historical target signal segments, the method further includes: If the sampling frequency of the target signal to be completed is different from that of the context-related signal, then the low-frequency sampling signal is uniformly resampled to a preset high-frequency time reference by an interpolation algorithm.

7. A vehicle signal completion device, characterized in that, include: The acquisition module is used to acquire real-time operating data of the target vehicle. The real-time operating data includes: current fault code, target signal to be completed, and at least one context-related signal related to the target signal. The search module is used to search for at least one similar historical case from the same vehicle series fault case library based on the current fault code; wherein, the same vehicle series fault case library includes multiple historical cases, and each historical case includes: historical fault code, historical environmental data, historical target signal and corresponding historical context association signal; The alignment module is used to perform temporal alignment processing on the historical target signal segments in the similar historical cases based on the context-related signal sequence centered on the missing time period of the target signal, so as to obtain the aligned historical target signal segments. A generation module is used to generate a completion signal for the target vehicle based on all the aligned historical target signal segments.

8. An electronic device, characterized in that, Including processor and memory, among which, Memory, used to store computer programs; A processor for executing a program stored in memory to implement the method described in any one of claims 1-6.

9. A vehicle, characterized in that, It includes the electronic device as described in claim 10.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the method described in any one of claims 1-6.

Images