Automobile wire harness core wire fracture detection method fusing ultrasonic wave and electromagnetic induction

Abstract

The invention relates to the technical field of automobile wire harness detection, and discloses an automobile wire harness core wire fracture detection method fusing ultrasonic waves and electromagnetic induction. The method comprises the following steps: acquiring ultrasonic wave and electromagnetic induction monitoring data of a wire harness through a sensor to form two types of data sets; traversing the data set to perform stability analysis, determining two types of stability factors, and searching a target detection step length in a detection configuration space by taking the two types of stability factors as indexes. Meanwhile, historical detection data is utilized, and experience layers are divided according to the detection parameter difference and the strategy confidence degree to form a multi-layer experience pool; an intelligent detection strategy is determined according to the target fracture state, and an experience detection strategy is screened in combination with the confidence coefficient and the matching degree. The two strategies are dynamically adjusted through a double-channel mechanism, a target detection strategy is determined, an adjusting instruction is generated to detect the wire harness to obtain the actual fracture state, an experience layer is updated according to the state difference, and experience pool data are perfected. The method integrates the advantages of multiple technologies, and improves the detection adaptability and reliability.

Description

Technical Field

[0001] This invention relates to the field of automotive wiring harness testing technology, specifically a method for detecting core wire breakage in automotive wiring harnesses that integrates ultrasonic waves and electromagnetic induction. Background Technology

[0002] As the main network component of automotive circuits, automotive wiring harnesses play a crucial role in power transmission and signal transmission. The integrity of their core wires directly affects the overall safety and stability of the vehicle's operation. With the continuous improvement of automotive electronics, wiring harness structures are becoming increasingly complex, leading to a higher probability of core wire breakage faults. These faults are often characterized by their high degree of concealment and difficulty in localization, posing numerous challenges to testing.

[0003] Currently, commonly used methods for detecting core wire breakage in automotive wiring harnesses include manual visual inspection, resistance measurement, and single-sensor detection. Manual visual inspection relies on the experience and observation skills of the operator, which is not only inefficient but also makes it difficult to detect breakage defects inside the core wire or under the insulation layer, easily leading to missed or false detections. While resistance measurement can determine the presence of breakage by detecting changes in the wiring harness's resistance, this method requires disconnecting the wiring harness from other components, making it cumbersome and unable to accurately locate the breakage, thus limiting its applicability to complex wiring harness systems.

[0004] In single-sensor detection methods, ultrasonic testing technology can identify structural anomalies within the core wire by utilizing the differences in the propagation characteristics of ultrasonic waves in different media. However, limited by the attenuation characteristics of ultrasonic waves, it is easily affected by external interference when detecting long or multi-core wire bundles, leading to insufficient stability of the detection data and difficulty in accurately distinguishing between normal fluctuations and breakage signals. Electromagnetic induction detection technology determines the conductivity of the core wire by detecting changes in the electromagnetic field. However, this technology is quite sensitive to the material and thickness of the insulation layer on the core wire surface. When the insulation layer is aged or damaged, it will interfere with the electromagnetic induction signal, resulting in a decrease in detection accuracy.

[0005] Most existing testing methods lack effective utilization of historical testing data and cannot adaptively adjust testing strategies according to different fracture states. They suffer from insufficient flexibility and adaptability when facing diverse fracture fault types. Furthermore, their responsiveness to real-time data changes during testing is weak, making it difficult to dynamically correct testing parameters. This results in the reliability and timeliness of test results failing to meet the high standards of the automotive industry for wiring harness testing. These problems make it difficult for existing testing methods to efficiently and accurately complete the core wire fracture detection task in automotive wiring harnesses in practical applications. A new testing method is needed that integrates the advantages of multiple testing technologies and improves testing stability and adaptability. Summary of the Invention

[0006] The purpose of this invention is to provide a method for detecting core wire breakage in automotive wiring harnesses that integrates ultrasonic waves and electromagnetic induction, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides a method for detecting core wire breakage in automotive wiring harnesses that integrates ultrasonic waves and electromagnetic induction, the method comprising:

[0008] By collecting ultrasonic monitoring data and electromagnetic induction monitoring data of automotive wiring harnesses through sensors, ultrasonic monitoring data sets and electromagnetic induction monitoring data sets are obtained.

[0009] The ultrasonic monitoring data set and the electromagnetic induction monitoring data set are traversed to perform data stability analysis and determine the ultrasonic stability factor and the electromagnetic induction stability factor.

[0010] Using the ultrasonic stability factor and electromagnetic induction stability factor as indexes, a centralized search is performed in the detection configuration space to determine the target detection step size;

[0011] Based on the differences in historical detection parameters and the confidence level of the strategy, the historical detection strategy is divided into multiple experience layers to form a multi-layered experience pool.

[0012] Based on the set target fracture state, an intelligent detection strategy is determined under the target fracture state, and the multi-layer experience pool is filtered based on the strategy confidence and target matching degree to obtain an experience detection strategy.

[0013] The intelligent detection strategy and the empirical detection strategy are dynamically adjusted using a dual-pathway mechanism to cope with changes in real-time monitoring data, thereby determining the target detection strategy.

[0014] The vehicle wiring harness is detected based on adjustment instructions to obtain the actual breakage state, wherein the adjustment instructions are generated based on the target detection strategy and the current monitoring data;

[0015] The factors that cause the difference between the actual fracture state and the target fracture state are determined, and the content of the experience layer is updated based on the factors that cause the difference to improve the multi-layer experience pool data.

[0016] Preferably, the step of acquiring ultrasonic monitoring data and electromagnetic induction monitoring data of the automotive wiring harness through sensors to obtain ultrasonic monitoring data sets and electromagnetic induction monitoring data sets includes:

[0017] Ultrasonic sensors are used to acquire ultrasonic reflection signals from automotive wiring harnesses, forming an ultrasonic monitoring data set.

[0018] Electromagnetic induction sensors are used to acquire electromagnetic field change signals of automotive wiring harnesses, forming an electromagnetic induction monitoring data set.

[0019] Preferably, the step of traversing the ultrasonic monitoring data set and the electromagnetic induction monitoring data set to perform data stability analysis and determine the ultrasonic stability factor and the electromagnetic induction stability factor includes:

[0020] The variance of the ultrasonic monitoring data set is calculated to obtain the ultrasonic stability factor;

[0021] The variance of the electromagnetic induction monitoring data set is calculated to obtain the electromagnetic induction stability factor.

[0022] Preferably, the step of determining the target detection step size by performing a centralized search in the detection configuration space using the ultrasonic stability factor and the electromagnetic induction stability factor as indexes includes:

[0023] A three-dimensional space is pre-constructed, wherein the x-axis of the three-dimensional space is the ultrasonic stability factor, the y-axis is the electromagnetic induction stability factor, and the z-axis is the detection step size;

[0024] The ultrasonic stability factor and electromagnetic induction stability factor are input into the three-dimensional space, and the sample points in the detection configuration space are matched by a centralized search algorithm to determine the target detection step size.

[0025] Preferably, the step of dividing the historical detection strategy into multiple experience layers based on the differences in historical detection parameters and the strategy confidence using historical detection data to form a multi-layered experience pool includes:

[0026] Analyze historical detection data to identify the gap between the confidence level of each historical detection strategy and the historical detection parameters;

[0027] Based on the confidence level and the gap, the historical detection strategy is divided into an expert experience layer, a key information experience layer, and a general information experience layer to form a multi-layered experience pool.

[0028] Preferably, the step of determining an intelligent detection strategy under a set target fracture state, and filtering the multi-layer experience pool based on strategy confidence and target matching degree to obtain an experience detection strategy, includes:

[0029] The target fracture state is input into the intelligent algorithm to generate an intelligent detection strategy;

[0030] Samples are drawn from the multi-layered experience pool based on the strategy confidence, and then the samples are filtered based on the target matching degree to obtain the experience detection strategy.

[0031] Preferably, the step of dynamically adjusting the intelligent detection strategy and the experience-based detection strategy using a dual-pathway mechanism to respond to changes in real-time monitoring data, thereby determining the target detection strategy, includes:

[0032] Real-time monitoring of ultrasonic and electromagnetic induction data of automotive wiring harnesses;

[0033] When data anomalies are detected, an experience-based detection strategy is activated for dynamic adjustment; otherwise, an intelligent detection strategy is executed to determine the target detection strategy.

[0034] Preferably, the step of detecting the automotive wiring harness based on adjustment commands to obtain the actual breakage state includes:

[0035] An adjustment command is generated based on the target detection strategy and the current ultrasonic monitoring data set and electromagnetic induction monitoring data set.

[0036] The adjustment command is used to drive the detection device to perform wire harness detection and obtain the actual breakage state.

[0037] Preferably, the factors for determining the difference between the actual fracture state and the target fracture state include:

[0038] Compare the difference between the actual fracture state and the target fracture state;

[0039] Based on the difference values, the fracture difference stages are identified, and the difference factors are extracted.

[0040] Preferably, updating the content of the experience layer based on the difference factors to improve the multi-layer experience pool data includes:

[0041] Develop adjustment strategies based on the aforementioned differences;

[0042] The adjustment strategy is updated to the corresponding experience layer in the multi-layer experience pool.

[0043] Compared with the prior art, the present invention has the following beneficial effects:

[0044] Through multi-dimensional data collection and intelligent processing, it effectively makes up for the shortcomings of traditional detection methods and shows significant advantages in many aspects such as detection stability, strategy adaptability, and result reliability.

[0045] In the data acquisition and processing stage, this method simultaneously acquires ultrasonic monitoring data and electromagnetic induction monitoring data. By traversing both types of data sets to perform stability analysis and determine the corresponding stability factors, it can fully leverage the complementary characteristics of the two detection technologies. The combination of ultrasonic detection technology's ability to identify internal structural anomalies in the core wire and electromagnetic induction detection technology's sensitivity to the core wire's conductivity status reduces the impact of external interference when using a single sensor. Furthermore, the introduction of stability factors provides accurate data for optimizing subsequent detection strategies, avoiding misjudgments caused by fluctuations in single data points, and significantly improving the effectiveness and stability of the detection data.

[0046] In terms of detection strategy optimization, this method innovatively constructs a multi-layered experience pool. By analyzing historical detection data, it divides the experience layer according to the differences in historical detection parameters and strategy confidence, thus achieving a systematic integration of past detection experience. Based on the target fracture state, an intelligent detection strategy is determined, and experience-based detection strategies are selected by combining strategy confidence and target matching degree. This fully utilizes effective information from historical data, making the detection strategy more aligned with actual detection needs. Compared to the fixed detection patterns in traditional methods, this strategy generation approach, combining intelligent algorithms and historical experience, allows for flexible adjustments to the detection approach for different fracture states, significantly improving the adaptability and targeting of the detection strategy.

[0047] The application of the dual-path mechanism further enhances the dynamic adjustment capability of the detection process. By dynamically adjusting the intelligent detection strategy and the experience-based detection strategy, it can respond to changes in monitoring data in real time, promptly correct detection parameters and steps, and avoid detection deviations caused by fluctuations in real-time data. This dynamic adjustment capability enables the method to maintain stable detection performance even in complex and ever-changing detection environments, such as wire harness insulation aging and external electromagnetic interference, effectively meeting the detection needs of different scenarios.

[0048] In the detection execution and experience update stages, adjustment instructions are generated based on the target detection strategy and current monitoring data. This ensures that the detection operation accurately meets actual needs, reduces unnecessary steps, and improves detection efficiency. Simultaneously, by comparing the differences between the actual fracture state and the target fracture state, the multi-layered experience pool is updated, continuously enriching the data content within the pool. This allows the experience-based detection strategy to be continuously optimized with the accumulation of real-world application scenarios, forming a virtuous cycle of "detection-feedback-update." This experience accumulation and iteration mechanism not only improves the efficiency and accuracy of subsequent detection strategy generation but also enables the method to gradually adapt to more diverse fracture fault types in long-term application, further expanding its applicability. Attached Figure Description

[0049] Figure 1 This is a schematic diagram illustrating the working principle of the automotive wiring harness core wire breakage detection method that integrates ultrasonic waves and electromagnetic induction as described in this invention.

[0050] Figure 2 A flowchart for collecting ultrasonic and electromagnetic induction monitoring data;

[0051] Figure 3 A flowchart for determining the target detection step size;

[0052] Figure 4 A flowchart for constructing a multi-layered experience pool. Detailed Implementation

[0053] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0054] Please see Figure 1 This invention provides a method for detecting core wire breakage in automotive wiring harnesses that integrates ultrasonic waves and electromagnetic induction. The method includes:

[0055] By integrating ultrasonic and electromagnetic induction technologies, precise detection of core wire fractures in automotive wiring harnesses is achieved. Sensors collect ultrasonic and electromagnetic induction monitoring data of the wiring harness, forming corresponding datasets. Subsequently, stability analysis is performed on these datasets to calculate ultrasonic and electromagnetic induction stability factors. Using these stability factors as indices, a centralized search is conducted within a predefined detection configuration space to determine the target detection step size. Simultaneously, based on historical detection data, historical detection strategies are divided into multiple experience layers according to the differences in historical detection parameters and strategy confidence, constructing a multi-layered experience pool. For a given target fracture state, an intelligent detection strategy is generated, and experience detection strategies are selected from the multi-layered experience pool based on strategy confidence and target matching degree. A dual-pathway mechanism is used to dynamically adjust the intelligent detection strategy and experience detection strategies to cope with changes in real-time monitoring data, thereby determining the final target detection strategy. Based on the target detection strategy and current monitoring data, adjustment commands are generated to drive the detection equipment to detect the automotive wiring harness and obtain the actual fracture state. By comparing the difference between the actual fracture state and the target fracture state, discrepancy factors are identified, and the experience layers in the multi-layered experience pool are updated based on these factors to optimize the subsequent detection process.

[0056] Example 1: See Figure 2In automotive wiring harness core wire breakage detection, data acquisition is the primary and fundamental step, relying on the coordinated operation of physical sensors and signal processing systems. Ultrasonic sensors are strategically placed at several key nodes of the wiring harness under test, such as junctions, connector interfaces, or known fatigue-prone areas. The selection of these sensors must consider their frequency response characteristics; typically, piezoelectric transducers with center frequencies in the range of several hundred kHz to several MHz are chosen to suit the acoustic characteristics of common automotive wiring harness materials. The sensors operate in pulse-echo mode, periodically emitting brief ultrasonic pulses and subsequently receiving signals reflected from the internal structure of the wiring harness (including the core wire, insulation layer, and any defective interfaces). The received raw analog signal is extremely weak and noisy, so it is immediately amplified by a preamplifier and then passed through a bandpass filter to suppress out-of-band interference. The conditioned signal is then sampled by a high-speed analog-to-digital converter and converted into a digital signal stream. Each sampling point contains a timestamp, amplitude information, and possibly phase information. These data points are arranged in chronological order and together constitute the ultrasonic monitoring dataset. This dataset is a dynamically updated sequence, and its size depends on the detection duration and sampling rate.

[0057] Electromagnetic induction monitoring is performed simultaneously. Electromagnetic induction sensors, typically in the form of Rogowski coils or Hall effect sensors, are arranged non-contactly around the wiring harness. Their core function is to sense the alternating magnetic field generated by changes in current within the harness. When a working current flows through the harness, a corresponding magnetic field is generated around it; if a core wire breaks, the current path at that point is blocked, causing a distortion in the magnetic field distribution. The sensor captures these subtle magnetic field changes and converts them into a proportional voltage signal. This signal also faces noise challenges, primarily from electromagnetic interference from other electronic devices within the vehicle. Therefore, signal conditioning circuitry typically includes shielding measures, differential amplification, and power frequency notch filtering to improve the signal-to-noise ratio. The conditioned analog magnetic field signal is converted into a digital sequence by an ADC, forming an electromagnetic induction monitoring dataset. Each data point in this dataset records the absolute value of the magnetic field strength at a specific moment, or the change relative to a reference value.

[0058] After obtaining these two independent but time-synchronized datasets, the system began data stability analysis. The purpose of this analysis was to quantify the fluctuations of the two datasets over a short period, providing a basis for subsequent adaptive adjustments to the detection parameters. For the ultrasonic monitoring dataset, stability analysis focused on the volatility of its amplitude sequence. The system extracted a continuous data window from the dataset and calculated the statistical variance of all amplitude data within that window. The variance was calculated as follows: first, the arithmetic mean of all amplitude values ​​within the window was calculated and used as a benchmark; then, the deviation of each data point from this mean was calculated, and all deviations were squared; finally, the average of these squared values ​​was calculated. The result is the ultrasonic stability factor. A lower factor value indicates that the intensity of the ultrasonic reflected signal is relatively stable and fluctuates little within that time window, suggesting that the state of the harness structure may not have changed drastically; conversely, a higher factor value indicates that the reflected signal is unstable, possibly due to external vibration interference, changes in sensor coupling state, or the development of potential defects.

[0059] The same variance calculation process is performed on the electromagnetic induction monitoring dataset in complete parallel. The system selects magnetic field strength data points within the same time span, performs identical mathematical operations, and ultimately obtains the electromagnetic induction stability factor. This factor characterizes the stability of the electromagnetic field signal. Drastic fluctuations in the magnetic field signal may originate from sudden changes in load current, interference from a strong external magnetic field, or phenomena such as arc discharge caused by changes in contact resistance due to wire breakage. These two stability factors, one derived from acoustic physics and the other from electromagnetic physics, together constitute a set of core indicators describing the reliability and signal quality of the current monitoring environment. They will be transmitted to the downstream decision-making module to guide fine-tuning of detection behavior. The entire implementation process emphasizes synchronous data acquisition, independent processing, and comprehensive evaluation of indicators.

[0060] Example 2: See Figure 3 In determining the target detection step size, the system relies on a pre-constructed mathematical model based on a large amount of historical experimental data. This model is conceptualized as a three-dimensional virtual space, built upon correlated data accumulated from countless past detection runs. Within this space, each dimension represents a key parameter affecting detection accuracy. The horizontal x-axis is defined as the ultrasonic stability factor, its value range determined based on the minimum and maximum values ​​of this factor in historical data. The perpendicular y-axis is defined as the electromagnetic induction stability factor, its value range also based on historical observations. The vertical z-axis represents the detection step size, an output parameter that needs to be dynamically determined; its value represents the interval or period of spatial movement or temporal sampling of the detection equipment.

[0061] This three-dimensional space is not empty; it is filled with a large number of discrete sample points. Each sample point is a data triple containing an ultrasonic stability factor value, an electromagnetic induction stability factor value, and a detection step size value that has been proven in historical practice to best match this set of stability conditions. These sample points constitute an empirical "configuration space," which is the core carrier for transforming historical experience into operational knowledge. The sample points come from various sources, including calibration experiments under controlled laboratory conditions, long-term monitoring data of known good wire harnesses, and optimal parameters recorded when testing wire harnesses containing artificially created defects.

[0062] After the system calculates the current ultrasonic stability factor and electromagnetic induction stability factor from real-time monitoring data, it inputs these two values ​​as a query point into the pre-constructed three-dimensional configuration space. The next task is to find the nearest sample point to the current query point in the dense sample point cloud and adopt its corresponding detection step size value as the target value for this detection. This process is accomplished by an algorithm called clustered search. This algorithm first calculates the Euclidean distance between the query point and every sample point in the configuration space. In three-dimensional space, this distance is the square root of the sum of the squares of the differences between the query point and the sample point in the x-axis, y-axis, and z-axis directions. Since the z-axis value (i.e., the detection step size) is itself the target to be solved, in the distance calculation, the algorithm actually performs a preliminary proximity assessment on a two-dimensional plane composed of the x-axis and y-axis, that is, it only considers the differences in the two-dimensional coordinates composed of the ultrasonic and electromagnetic induction stability factors.

[0063] After calculating the distances between all sample points and the query point on the two-dimensional plane, the algorithm selects a number of candidate sample points with the smallest distances. This number is a pre-set parameter. The algorithm then further examines the detection step size associated with these candidate sample points. A common strategy is to directly use the z-axis value of the nearest neighbor sample point as the final target detection step size. Another more robust strategy is to take a weighted average or median of the detection step size values ​​of these multiple neighboring sample points to smooth out possible outliers, thereby outputting a more reliable target detection step size value.

[0064] This determined target detection step size is then passed to the downstream detection control module, its physical meaning being to guide the scanning behavior of the detection equipment. For example, if the stability factor indicates that the current signal noise is high and fluctuations are strong, the search algorithm may match a larger detection step size, which means that the physical interval of the sensor's movement on the harness will increase, or the sampling period will become longer. This is a trade-off strategy: when the signal quality is poor, the sampling density is reduced to avoid collecting too much noise-contaminated data while ensuring detection efficiency. Conversely, if the stability factor indicates that the signal is very clear and stable, the algorithm may match a smaller detection step size, allowing for higher resolution fine scanning to capture the smallest potential defect features. The essence of the entire process is to automatically transform the real-time monitored signal quality features into optimal detection action parameters through an empirical knowledge base, achieving adaptive adjustment of the detection process.

[0065] Example 3: See Figure 4 In constructing a multi-layered experience pool, the system needs to conduct in-depth analysis of accumulated historical detection data to form a strategy hierarchy with different levels of confidence and practicality. Historical detection data originates from records of all previously executed detection tasks. Each record contains complete contextual information: the strategy parameters used during detection (such as sensor configuration, detection frequency, and analysis algorithm thresholds), the raw monitoring data collected at that time, the final determination of the fracture state, and the actual fracture state verified afterward (if known). The system first performs quantitative calculations on two core indicators for each historical detection strategy: strategy confidence and the difference between historical detection parameters.

[0066] Policy confidence (using symbols) The historical strategy (represented by the sliding time window) is a comprehensive evaluation metric used to measure the reliability and consistency of results when implemented in the past. Its calculation does not simply rely on the success or failure of a single detection, but is based on the performance of multiple detections within a sliding time window. The calculation formula is as follows: ,in: This indicates the number of times the detection results match the subsequent verification results in the history of the strategy's execution; This is the total number of times the strategy was used; This represents the original result value output by the strategy during the i-th detection; This represents the true value of the post-verification corresponding to the i-th detection; It is the upper limit of the range of the result value, used for normalization; and These are weighting coefficients, satisfying... The success rate and outcome bias are assigned different levels of importance. This calculation process quantifies the accuracy and stability of the strategy.

[0067] Differences in historical detection parameters (using symbols) The parameter set (represented by the parameter set) measures the deviation of a strategy from its ideal state during application. It is calculated by comparing historical strategy parameters with a "golden set" derived from extensive statistical data. For each configurable parameter, its relative deviation from the golden set is calculated, and finally, a weighted average of the deviations for all parameters is taken to obtain the result. In obtaining each historical strategy and Then, the system stratifies the data according to preset threshold rules. (Setting) and As the upper and lower thresholds of confidence, set This serves as a high threshold for the parameter difference. The expert experience layer includes those layers that simultaneously satisfy... and The strategies are considered well-validated and have near-optimal parameter settings. The key information experience layer contains strategies that satisfy… and These strategies have good reliability but may not be optimal in certain scenarios. The general information experience layer contains all... or The strategies employed may have questionable reliability or parameter rationality, but they may still contain valuable information. These three layers together constitute a structured, multi-layered pool of experience.

[0068] When the system needs to generate a detection strategy for a target fracture state, its operation is two-track. On one hand, the target fracture state (e.g., specifying microcracks with a length of 0.5 mm to 2 mm to be detected) is input into an intelligent algorithm module. This module is typically a trained machine learning model, such as a deep neural network. The model receives a vectorized representation of the target state (such as crack type, length range, expected location, etc.) and, through its complex internal nonlinear transformations, directly outputs a set of recommended detection strategy parameters, i.e., the intelligent detection strategy. This strategy is generated based on general patterns learned by the model from massive training data. On the other hand, the system retrieves relevant experience in parallel from a multi-layered experience pool. The retrieval process consists of two steps: first, based on the strategy confidence... A preliminary screening is performed. The system sets a minimum confidence threshold and extracts samples only from the expert and key levels. Policy samples exceeding this threshold ensure the basic reliability of the considered experience. These candidate samples are then finely screened, measured based on target matching degree. Target matching degree is calculated by transcribing the cosine similarity between the feature vectors of the candidate policy's historical application scenarios and the feature vector of the current target state. Finally, the policy sample with the highest matching degree is selected as the experience detection policy.

[0069] At this point, the system possesses both an intelligent detection strategy generated by a data-driven model, designed to solve general problems, and an experience-based detection strategy derived from successful historical practices and highly adapted to current objectives. These two strategies will be fed into a subsequent dual-pathway mechanism for final decision-making and fusion. The entire implementation process embodies a method of systematizing, quantifying, and hierarchically layering historical experience, and combining it with modern intelligent algorithms to address complex detection challenges.

[0070] Example 4: In implementing the dynamic strategy adjustment and detection execution based on the dual-path mechanism, the system relies on a continuously running real-time data monitoring and decision-making pipeline. The core function of this pipeline is to process the continuous stream of ultrasonic and electromagnetic induction monitoring data, determine which detection strategy to adopt, and ultimately generate executable instructions to drive the physical detection equipment.

[0071] The real-time data monitoring module reads the latest data packets from deployed ultrasonic and electromagnetic induction sensors at a fixed sampling period (e.g., 100 times per second). Each data packet contains a short sequence of signal samples (e.g., 10 milliseconds). This raw data is immediately fed into a lightweight preprocessing and feature extraction stage. For ultrasonic data, this stage quickly calculates the average amplitude, zero-crossing rate, and peak count within the time window. For electromagnetic induction data, it calculates the mean and standard deviation of the magnetic field strength within the window, as well as the difference from the previous window. These calculated feature values ​​constitute a lightweight summary of the signal state at the current moment.

[0072] These feature summaries are fed in real-time into an anomaly detection submodule, which maintains a dynamically updated "normal range" baseline. This baseline is established through statistical learning of feature data from the initial period of system startup (e.g., the first 5 minutes). Anomaly detection is performed by comparing real-time feature values ​​with the upper and lower bounds of the baseline. For example, if the average amplitude of an ultrasonic signal exceeds twice the standard deviation of the baseline average for three consecutive sampling cycles, and the differential value of electromagnetic induction simultaneously shows a significant peak, the anomaly detection module will determine that the current data is "abnormal." This multi-feature, continuous anomaly determination logic helps reduce false alarms and avoids unnecessary strategy switching triggered by transient interference.

[0073] The core decision-making logic of the dual-path mechanism is built upon the aforementioned anomaly detection results. The system always maintains two available strategy paths simultaneously: one is the "intelligent detection strategy" generated by the intelligent algorithm, and the other is the "experience detection strategy" selected from a multi-layered experience pool. When the system is running normally and no anomaly flag is triggered, the decision logic defaults to selecting and outputting the parameter set of the intelligent detection strategy. These parameters are directly passed to the strategy execution engine.

[0074] Once the anomaly detection module confirms an abnormal state, the decision logic immediately switches. It suspends the use of the intelligent strategy and instead activates the parameter set of the experience-based detection strategy, sending it to the strategy execution engine. This switch is almost instantaneous, designed to leverage effective methods from historical experience for handling similar anomalies, offering a faster response time than rerunning the intelligent algorithm for computation. Notably, the anomaly state is continuously monitored. When the system determines that real-time data characteristics have returned to the normal range and persisted for a period of time, the decision logic automatically switches back to the intelligent detection strategy, thus restoring the detection mode based on the general model. This dynamic adjustment mechanism allows the system to flexibly respond to various changes in signal conditions during operation.

[0075] The strategy parameter set received by the strategy execution engine is a blueprint for a series of control commands. The target detection strategy is ultimately determined and encapsulated into specific, executable "adjustment instructions." The generation of these instructions is a translation process that combines the strategy parameters with the latest snapshot of monitoring data. For example, one strategy parameter might require "increasing the ultrasonic sensor's transmission power by 20%," and the engine will calculate the specific value to be set based on the current actual power value; another parameter might require "reducing the scanning speed to 50% of the standard value," and the engine will calculate the new target speed value based on the current motor speed. All these specific, numerical commands are assembled into a structured instruction package.

[0076] This adjustment command packet is sent to the underlying detection equipment controller via a communication interface (such as CAN bus or Ethernet). The detection equipment is typically an automated platform integrating an ultrasonic probe, an electromagnetic induction coil, and a precision motion mechanism. The controller parses the command packet and executes the commands sequentially: adjusting various sensor settings (such as gain and frequency), and controlling a robotic arm or conveyor belt to move the probe at specified step sizes and speeds to scan the wire harness. Throughout the scanning process, the equipment synchronously acquires high-precision response data.

[0077] After scanning, the large amount of raw data collected is transmitted back to the system's data analysis core. Here, a specialized fracture identification algorithm (such as a wavelet transform-based feature extraction and classifier) ​​is used to process and analyze the data. This algorithm comprehensively analyzes the ultrasonic echo waveform and electromagnetic field distortion mode, ultimately outputting a diagnostic conclusion about the harness condition, i.e., the actual fracture state. This state may include qualitative judgments such as "no fracture," "suspected fracture," and "confirmed fracture," as well as quantitative information such as the location coordinates of the fracture point and the estimated value of the fracture degree. This result is stored and used for subsequent performance evaluation and experience pool updates, as shown in Table 1.

[0078] Table 1: Examples of real-time monitoring data streams and decision responses.

[0079] Example 5: After completing the inspection of the automotive wiring harness and obtaining the actual fracture state, the system enters a crucial analysis and self-optimization phase. The core task of this phase is to evaluate the performance of this inspection and use the evaluation results to optimize the system's internal knowledge base, i.e., the multi-layered experience pool, thereby improving the accuracy and efficiency of future inspections. The system first juxtaposes the output of this inspection: the actual fracture state, with the target fracture state initially set for this inspection task. The target fracture state is a set of expected result indicators predefined according to the inspection requirements before the inspection begins. It may include the fracture type (e.g., complete fracture, insulation layer damage, microcracks), the precise location of the fracture (e.g., coordinates at a specific distance from the connector), the physical dimensions of the fracture (e.g., crack length, depth), and the severity level of the fracture. The actual fracture state is the specific conclusion drawn after executing the inspection strategy and analyzing sensor data. The comparison process is quantitative, calculating the difference value for each indicator. For example, for fracture location, the difference value is the Euclidean distance between the actual and target coordinates; for crack length, the difference value is the absolute difference between the actual measurement and the target value; for qualitative indicators such as type, they are converted into comparable numerical differences through a predefined mapping table. All these individual difference values ​​are aggregated and a comprehensive difference value is calculated using a weighted algorithm that takes into account the relative importance of different indicators in the overall assessment.

[0080] After calculating the overall difference value, the system needs to interpret the meaning of this value, that is, to identify the difference stage. The difference stage is a qualitative classification used to describe the severity and nature of the deviation between the current detection result and the expected target. The system predefines several difference stages and their corresponding overall difference value threshold ranges. For example, when the overall difference value is below a very low lower threshold, it is classified as the "no significant difference" stage, indicating that the detection result is highly consistent with the target. When the overall difference value falls into an intermediate range, it may be classified as the "parametric bias" stage, suggesting that although a fracture was detected, there is a measurable discrepancy in the specific quantitative parameters. When the overall difference value exceeds a high upper threshold, it is classified as the "fundamental difference" stage, indicating that a fundamental misjudgment may have occurred, such as not detecting an existing fracture or falsely reporting a non-existent fracture. This stage classification helps in the initial localization of the problem.

[0081] Next comes the extraction of discrepancy factors, an attribution process designed to identify the root causes of the discrepancies. The system traces back the entire detection chain, analyzing each step from data acquisition to result output. It examines the quality of real-time monitoring data, checking whether the stability factors of ultrasonic and electromagnetic induction signals are at abnormal levels; it reviews the parameters of the target detection strategy used, analyzing whether it is the optimal choice for the current harness environment and target state; it also evaluates the output confidence of the signal processing algorithm and fracture identification model. Through this multi-faceted diagnosis, the system extracts specific factors that may lead to discrepancies. These factors may be technical, such as "the ultrasonic sensor is subjected to strong environmental noise interference at a specific frequency," or strategic, such as "the scanning step size used for microcrack targets is too coarse, leading to missed detections," or algorithm-related, such as "the fracture classification model is insufficient in extracting the features of a certain novel insulating material's damage."

[0082] Based on the extracted discrepancies, the system formulates corresponding adjustment strategies. These strategies are specific, actionable improvement plans designed to eliminate or mitigate the impact of the identified factors. If a factor is diagnosed as a data quality issue, the adjustment strategy might include "increasing the sensor calibration frequency under specific operating conditions" or "adding a filtering algorithm for specific noise to the signal preprocessing stage." If the factor is related to the detection strategy, the adjustment strategy might involve "adding a strategy entry emphasizing higher resolution scanning to the experience pool for a certain type of target fracture state" or "adjusting the abnormal threshold that triggers strategy switching in the dual-path mechanism." The formulation of these adjustment strategies is a reasoning process, with the core objective of preventing similar discrepancies from recurring in the future.

[0083] The system puts these adjustment strategies into practice, performing targeted updates to the multi-layered experience pool. The update process is not simply adding new strategies to the pool, but intelligently updating to the corresponding experience layer based on the severity and nature of the discrepancies. For fundamental strategy flaws revealed by "principle-based differences," the adjustment strategy may be directly added to the expert experience layer for future priority adoption. For minor optimizations corresponding to "parametric biases," the adjustment strategy may be used to correct the parameters of existing strategies in the key information experience layer. Sometimes, an adjustment strategy may mean lowering the confidence rating of an existing strategy, or even moving it from a higher layer to the general information experience layer. This entire update process transforms the multi-layered experience pool from a static historical record into an intelligent system capable of learning from each detection practice and dynamically evolving its knowledge, thereby continuously improving the overall performance of the detection system.

[0084] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0085] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for detecting core wire breakage in automotive wiring harnesses that integrates ultrasonic waves and electromagnetic induction, characterized in that, include: By collecting ultrasonic monitoring data and electromagnetic induction monitoring data of automotive wiring harnesses through sensors, ultrasonic monitoring data sets and electromagnetic induction monitoring data sets are obtained. The ultrasonic monitoring data set and the electromagnetic induction monitoring data set are traversed to perform data stability analysis and determine the ultrasonic stability factor and the electromagnetic induction stability factor. Using the ultrasonic stability factor and electromagnetic induction stability factor as indexes, a centralized search is performed in the detection configuration space to determine the target detection step size; Based on the differences in historical detection parameters and the confidence level of the strategy, the historical detection strategy is divided into multiple experience layers to form a multi-layered experience pool. Based on the set target fracture state, an intelligent detection strategy is determined under the target fracture state, and the multi-layer experience pool is filtered based on the strategy confidence and target matching degree to obtain an experience detection strategy. The intelligent detection strategy and the empirical detection strategy are dynamically adjusted using a dual-pathway mechanism to cope with changes in real-time monitoring data, thereby determining the target detection strategy. The vehicle wiring harness is detected based on adjustment instructions to obtain the actual breakage state, wherein the adjustment instructions are generated based on the target detection strategy and the current monitoring data; The factors that cause the difference between the actual fracture state and the target fracture state are determined, and the content of the experience layer is updated based on the factors that cause the difference to improve the multi-layer experience pool data.

2. The method for detecting core wire breakage in automotive wiring harnesses by integrating ultrasonic waves and electromagnetic induction as described in claim 1, characterized in that, The process involves collecting ultrasonic and electromagnetic induction monitoring data from automotive wiring harnesses using sensors to obtain ultrasonic monitoring data sets and electromagnetic induction monitoring data sets, including: Ultrasonic sensors are used to acquire ultrasonic reflection signals from automotive wiring harnesses, forming an ultrasonic monitoring data set. Electromagnetic induction sensors are used to acquire electromagnetic field change signals of automotive wiring harnesses, forming an electromagnetic induction monitoring data set.

3. The method for detecting core wire breakage in automotive wiring harnesses by integrating ultrasonic waves and electromagnetic induction according to claim 2, characterized in that, The process of traversing the ultrasonic monitoring data set and the electromagnetic induction monitoring data set to perform data stability analysis, and determining the ultrasonic stability factor and the electromagnetic induction stability factor, includes: The variance of the ultrasonic monitoring data set is calculated to obtain the ultrasonic stability factor; The variance of the electromagnetic induction monitoring data set is calculated to obtain the electromagnetic induction stability factor.

4. The method for detecting core wire breakage in automotive wiring harnesses by integrating ultrasonic waves and electromagnetic induction as described in claim 3, characterized in that, The step of determining the target detection step size by performing a centralized search in the detection configuration space using the ultrasonic stability factor and the electromagnetic induction stability factor as indexes includes: A three-dimensional space is pre-constructed, wherein the x-axis of the three-dimensional space is the ultrasonic stability factor, the y-axis is the electromagnetic induction stability factor, and the z-axis is the detection step size; The ultrasonic stability factor and electromagnetic induction stability factor are input into the three-dimensional space, and the sample points in the detection configuration space are matched by a centralized search algorithm to determine the target detection step size.

5. The method for detecting core wire breakage in automotive wiring harnesses by integrating ultrasonic waves and electromagnetic induction according to claim 4, characterized in that, The historical detection strategy is divided into multiple experience layers based on the differences in historical detection parameters and the strategy confidence level using historical detection data, forming a multi-layered experience pool, including: Analyze historical detection data to identify the gap between the confidence level of each historical detection strategy and the historical detection parameters; Based on the confidence level and the gap, the historical detection strategy is divided into an expert experience layer, a key information experience layer, and a general information experience layer to form a multi-layered experience pool.

6. The method for detecting core wire breakage in automotive wiring harnesses by integrating ultrasonic waves and electromagnetic induction as described in claim 5, characterized in that, The process of determining an intelligent detection strategy under a set target fracture state, and filtering the multi-layered experience pool based on strategy confidence and target matching to obtain an experience detection strategy, includes: The target fracture state is input into the intelligent algorithm to generate an intelligent detection strategy; Samples are drawn from the multi-layered experience pool based on the strategy confidence, and then the samples are filtered based on the target matching degree to obtain the experience detection strategy.

7. The method for detecting core wire breakage in automotive wiring harnesses by integrating ultrasonic waves and electromagnetic induction according to claim 6, characterized in that, The method of dynamically adjusting the intelligent detection strategy and the empirical detection strategy using a dual-pathway mechanism to cope with changes in real-time monitoring data, thereby determining the target detection strategy, includes: Real-time monitoring of ultrasonic and electromagnetic induction data of automotive wiring harnesses; When data anomalies are detected, an experience-based detection strategy is activated for dynamic adjustment; otherwise, an intelligent detection strategy is executed to determine the target detection strategy.

8. The method for detecting core wire breakage in automotive wiring harnesses by integrating ultrasonic waves and electromagnetic induction according to claim 7, characterized in that, The method of detecting the vehicle wiring harness based on adjustment commands to obtain the actual breakage state includes: An adjustment command is generated based on the target detection strategy and the current ultrasonic monitoring data set and electromagnetic induction monitoring data set. The adjustment command is used to drive the detection device to perform wire harness detection and obtain the actual breakage state.

9. The method for detecting core wire breakage in automotive wiring harnesses by integrating ultrasonic waves and electromagnetic induction according to claim 8, characterized in that, The factors determining the difference between the actual fracture state and the target fracture state include: Compare the difference between the actual fracture state and the target fracture state; Based on the difference values, the fracture difference stages are identified, and the difference factors are extracted.

10. The method for detecting core wire breakage in automotive wiring harnesses by integrating ultrasonic waves and electromagnetic induction according to claim 9, characterized in that, The process of updating the content of the experience layer based on the aforementioned difference factors to improve the multi-layer experience pool data includes: Develop adjustment strategies based on the aforementioned differences; The adjustment strategy is updated to the corresponding experience layer in the multi-layer experience pool.

Images