Household appliance wire harness electrical performance detection method and system

Abstract

The invention relates to the technical field of electrical performance detection, in particular to a household appliance wire harness electrical performance detection method and system. The method comprises the following steps: acquiring initial current and voltage data of a wire harness by using a sensor to obtain original electrical parameter matrix data; performing resistance modeling according to the original electrical parameter matrix data to obtain a resistance dynamic model; performing resistance characteristic analysis according to the resistance dynamic model to obtain resistance characteristic analysis data; and according to the resistance characteristic analysis data, carrying out resistance compliance determination to obtain compliance matrix data so as to carry out household appliance wire harness electrical performance detection auxiliary operation. The electrical performance of the household appliance wire harness is comprehensively detected and accurately evaluated, high technical innovation and practical application value are shown, and the crossing from simple static detection to intelligent dynamic analysis is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of electrical performance detection, and in particular to a method and system for detecting the electrical performance of household wiring harnesses. Background Art

[0002] With the rapid development of the home appliance industry and the continuous improvement of the intelligence of home appliance products, the performance quality of home appliance harnesses, as key components for electrical connection and power transmission in home appliances, is directly related to the operational stability and safety of the entire machine. Home appliance harnesses not only bear high-frequency current conduction, but also need to maintain long-term stable performance under various complex voltage fluctuations and environmental interference. Its importance is not only reflected in the normal operation of home appliances, but also directly affects the user's safety and the overall life of the product.

[0003] The operating environment of household appliances is diverse, and the wiring harness needs to face extreme conditions such as current mutation, voltage spike, frequent startup, high temperature and humidity. This complexity poses unprecedented challenges to the electrical performance of the wiring harness. Especially in modern household appliances, high efficiency energy saving and intelligent control have become the mainstream trend. As a bridge between information and energy, the performance of the wiring harness will directly affect the working efficiency and intelligence level of the equipment.

[0004] Traditional wiring harness testing methods mostly focus on single-dimensional static performance measurements, such as resistance value, current conduction stability, etc. These methods can only evaluate the basic performance of wiring harnesses under standard working conditions, and cannot reflect their true performance under dynamic loads, rapidly changing currents, or complex voltage fluctuations. Dynamic current and voltage changes are common phenomena in the operation of home wiring harnesses. This detection method that ignores dynamic characteristics leaves a huge blind spot in the comprehensive evaluation of products. Summary of the invention

[0005] In order to solve the above technical problems, the present invention proposes a method and system for detecting the electrical performance of household wiring harnesses to solve at least one of the above technical problems.

[0006] The present application provides a method for detecting electrical performance of a household wiring harness, comprising the following steps: Step S1: using sensors to collect initial current and voltage data of the wiring harness to obtain original electrical parameter matrix data; Step S2: Perform resistance modeling according to the original electrical parameter matrix data to obtain a resistance dynamic model; Step S3: performing resistance characteristic analysis according to the resistance dynamic model to obtain resistance characteristic analysis data; Step S4: Perform resistance compliance determination based on the resistance characteristic analysis data to obtain compliance matrix data to assist in the electrical performance testing of home wiring harnesses.

[0007] In the present invention, data is collected in real time through sensors and edge computing devices to ensure the real-time and integrity of the data. The modeling process captures the dynamic response behavior of the resistor under different current and voltage conditions, and can reflect the true performance of the wiring harness under actual working conditions. Through characteristic analysis, multi-dimensional characteristics can be extracted from the resistor model, including contact resistance, rate of change, stability, etc., to comprehensively evaluate the performance of the wiring harness. By comparing the characteristic analysis data with the standard value, it is possible to accurately determine whether the wiring harness meets the electrical performance requirements.

[0008] Preferably, step S1 specifically includes: Deploy a distributed sensor network to collect the voltage and current of the wiring harness and obtain the wiring harness state matrix data; Perform multi-dimensional fusion on the harness state matrix data to obtain multi-dimensional fusion matrix data; Perform dynamic current sampling based on multi-dimensional fusion matrix data to obtain dynamic current characteristic data; Perform voltage analysis on the multi-dimensional fusion matrix data to obtain voltage characteristic data; Performing multi-modal data calibration on the multi-dimensional fusion matrix data according to the dynamic current characteristic data and the voltage characteristic data to obtain calibration matrix data; Perform matrix generation on the calibration matrix data to obtain the original electrical parameter matrix data.

[0009] In the present invention, through multi-level data acquisition and processing, the original electrical parameter matrix data generated in step S1 contains dynamic and static information of the wiring harness current and voltage, providing a high-quality data basis for analysis. The distributed sensor network covers the key parts of the wiring harness, and can fully capture the current and voltage status of the wiring harness to avoid missing important information. The multi-dimensional fusion matrix data can reflect the global trend and local changes of the wiring harness performance. The sampling frequency is dynamically adjusted according to the current change rate to capture the details of the rapidly changing area while saving resources. The key characteristics of the voltage (such as peak value, voltage drop, stability, etc.) are extracted to provide rich characteristic information for multi-modal calibration. Multi-modal calibration aligns the dynamic current and voltage characteristic data to eliminate the time and amplitude deviations in the acquisition process. The original electrical parameter matrix data integrates the calibration results of current and voltage, and fully characterizes the electrical performance of the wiring harness.

[0010] Preferably, the dynamic current sampling is specifically: Performing regional change sampling according to the multi-dimensional fusion matrix data to obtain first dynamic current characteristic data; Perform multi-frequency sampling according to the multi-dimensional fusion matrix data to obtain second dynamic current characteristic data; Performing time-series correlation fusion on the first dynamic current characteristic data and the second dynamic current characteristic data to obtain dynamic current characteristic data; The specific details of time series correlation fusion are: Extract dynamic characteristics according to the first dynamic current characteristic data and the second dynamic current characteristic data to obtain first characteristic vector data and second characteristic vector data respectively; Performing hierarchical Bayesian network processing on the first characteristic vector data and the second characteristic vector data to obtain first hierarchical characteristic data and second hierarchical characteristic data respectively; Constructing a time series characteristic interaction relationship according to the first layer characteristic data and the second layer characteristic data to obtain interaction characteristic matrix data; Multi-layer characteristic fusion is performed on the first layer characteristic data and the second layer characteristic data according to the interactive characteristic matrix data to obtain dynamic current characteristic data.

[0011] The regional change sampling in the present invention dynamically adjusts the sampling strategy according to the rate and amplitude of the current change, which can effectively capture the regional characteristics with significant changes. Multi-frequency sampling combines high-frequency and low-frequency signal characteristics to achieve comprehensive capture of the global trend and local details of the current signal. Through the time series correlation fusion method, the time correlation between the characteristic data generated by different sampling strategies is mined to improve the timing consistency of the overall characteristic data. After extracting the dynamic characteristics, the first characteristic vector data and the second characteristic vector data are generated respectively, which can independently characterize the signal characteristics under different sampling strategies. The first layered characteristic data and the second layered characteristic data can respectively describe the high-level global characteristics and the low-level local characteristics through hierarchical processing, which improves the expression ability of the characteristic data. By constructing the characteristic interaction relationship, the dynamic correlation between the current characteristic data can be effectively identified. Through fusion, the generated dynamic current characteristic data can simultaneously reflect the local detail characteristics and the global trend characteristics.

[0012] Preferably, the regional change sampling is specifically: Perform time series segmentation according to the multi-dimensional fusion matrix data to obtain matrix time series segmentation data; Performing global current change rate calculation and local current change rate calculation on the matrix time series segmentation data to obtain global current change rate data and local current change rate data respectively; Performing regional classification on the multi-dimensional fusion matrix data according to the global current change rate data and the local current change rate data to obtain regional classification data; Allocate sampling rate according to regional classification data to obtain sampling rate matrix data; The current of the wiring harness is collected according to the sampling rate matrix data to obtain the wiring harness sampling change data; Perform multi-frequency data fusion based on the wire harness sampling change data to obtain preliminary reconstructed matrix data; Feature extraction is performed based on the preliminary reconstructed matrix data to obtain first dynamic current feature data.

[0013] The time segmentation in the present invention can divide the key time periods of the current signal, which is helpful to identify the local dynamic behavior of the current change. The global current change rate calculation provides the rate information of the overall change of the current signal, which is suitable for capturing the macro change trend. The local current change rate calculation can accurately locate the local characteristics (such as mutation points, fluctuation intervals). By combining the global and local change rate data for regional classification, the dynamic and stable areas of the current signal can be accurately divided. The regional classification data can intuitively reflect the dynamic characteristics of different areas of the harness and provide a basis for the allocation of sampling rates. By dynamically allocating the sampling rate, high-frequency sampling is used in high-dynamic areas and low-frequency sampling is used in low-dynamic areas, which greatly saves storage and computing resources. The harness sampling change data covers global and local characteristics and ensures the comprehensiveness of the signal. Multi-frequency data fusion integrates high-frequency and low-frequency signals, and the generated preliminary reconstruction matrix data can reflect the global trend and local details of the signal. Feature extraction can extract key dynamic current characteristics (such as peak value, change rate, stability, etc.) from the preliminary reconstruction matrix.

[0014] Preferably, the multi-frequency sampling is specifically: Perform spectrum graph convolution calculation based on multi-dimensional fusion matrix data to obtain matrix convolution layer data; According to the matrix convolution layer data, the key frequency band is identified to obtain the key frequency band data; Frequency band priority is divided according to key frequency band data to obtain frequency band priority data; Initialize the frequency allocation strategy according to the frequency band priority data to obtain the frequency allocation matrix data; Performing parallel multi-frequency signal sampling on the current of the wiring harness according to the frequency allocation matrix data to obtain parallel sampling matrix data; Extracting the frequency division domain signal characteristics of the parallel sampling matrix data to obtain the frequency division characteristic matrix data; Performing time domain characteristic fusion on the frequency division characteristic matrix data to obtain time domain characteristic fusion data; Multi-scale feature extraction is performed on the time domain characteristic fusion data to obtain second dynamic current characteristic data.

[0015] In the present invention, the spectrum graph is calculated by a convolutional neural network to extract key spectrum characteristics and realize in-depth analysis of the signal. The spectrum characteristic weights are used to quickly identify the frequency bands that most significantly affect the signal. Priority division ensures that high-priority frequency bands obtain more sampling resources and reduces redundant sampling of low-priority frequency bands. The frequency allocation strategy is dynamically initialized based on priority data to realize intelligent adjustment of signal sampling. Multi-frequency sampling can simultaneously collect signals in different frequency ranges to ensure the real-time sampling. Characteristics are extracted from signals in different frequency bands respectively, and the generated frequency division characteristic matrix data can more accurately describe the frequency domain behavior. Time domain characteristic fusion integrates frequency domain characteristics with time characteristics, and the generated data can simultaneously reflect the time and frequency characteristics of the signal. Multi-scale feature extraction can extract dynamic characteristics of different scales from time domain characteristic fusion data, providing richer characteristic information.

[0016] Preferably, the multimodal data calibration is specifically: According to the dynamic current characteristic data and the voltage characteristic data, the multi-dimensional fusion matrix data is subjected to inter-modal alignment learning to obtain the modal alignment weight matrix data; Perform characteristic time matching on the modal alignment weight matrix data to obtain characteristic time matching data; Perform multimodal feature completion on the feature time matching data to obtain completed feature matrix data; Performing modal normalization processing on the completed characteristic matrix data to obtain normalized calibration matrix data; Perform attention fusion on the normalized calibration matrix data to obtain the calibration matrix data.

[0017] The inter-modal alignment learning in the present invention can accurately capture the time series correlation between the dynamic current and voltage features by modeling the correlation between the two. Through feature time matching, the time axes of the dynamic current features and voltage features are aligned to ensure the consistency of the two characteristic data in the time dimension. Through multimodal feature completion, the missing data points that may exist in the signal acquisition process are filled, ensuring the integrity of the completed characteristic matrix data. Modal normalization processing unifies the dynamic current feature and voltage feature data into the same numerical range, avoiding the impact of feature amplitude differences on subsequent analysis. The attention mechanism can automatically focus on the key feature areas in the normalized calibration matrix, and improve the key feature expression ability of the fused calibration matrix data.

[0018] Preferably, step S2 specifically includes: Matrix feature extraction is performed according to the original electrical parameter matrix data to obtain electrical matrix feature data; Perform feature mapping modeling on the electrical matrix feature data to obtain an initial resistance model; Performing curve fitting on the initial resistance model to obtain a resistance model; The resistor model is iterated by parameter optimization to obtain a resistor dynamic model.

[0019] The feature extraction in the present invention can extract the key characteristics related to current, voltage and resistance from the original matrix data, reducing redundant information. The extracted characteristic data is converted into an initial model related to resistance through feature mapping, and a basic model of wiring harness resistance is constructed. By fitting the generated resistance model, the dynamic changes of resistance under different currents and voltages can be accurately described. The optimized resistance dynamic model can accurately reflect the dynamic response of resistance with time, load or environmental changes.

[0020] Preferably, step S3 specifically includes: Perform model input analysis according to the resistor dynamic model to obtain analytical input data; The characteristic space is divided according to the analytical input data and the resistance dynamic model to obtain the characteristic dimension data; Extract characteristic features according to characteristic dimension data to obtain characteristic feature data; Perform frequency domain analysis according to the characteristic feature data to obtain frequency domain characteristic data; Performing clustering calculation on the frequency domain characteristic data to obtain frequency domain characteristic clustering data; Multi-characteristic fusion analysis is performed based on frequency domain characteristic clustering data to obtain resistance characteristic analysis data.

[0021] In the present invention, the input data is parsed to structure the data of the dynamic model, which is convenient for hierarchical feature extraction. The characteristic space division decomposes the resistance characteristics into multiple independent dimensions (such as time, frequency, voltage, current, etc.), which is convenient for independent analysis of the characteristics of each dimension. The characteristic feature extraction can identify the key patterns in the resistance characteristics (such as mutation points, stability, etc.) and highlight the key information. The frequency domain analysis decomposes the resistance characteristics into the spectrum space, and can identify the characteristic behaviors of different frequency bands. The frequency domain characteristic data can highlight the patterns of different frequency bands (such as high-frequency changes, low-frequency stability), which is convenient for classification and clustering. Clustering calculation classifies similar frequency domain characteristics into the same category to help identify the main patterns of resistance characteristics. The characteristic fusion analysis integrates different frequency bands, time dimensions and spatial characteristics, and the generated resistance characteristic analysis data is global and complete.

[0022] Preferably, step S4 is specifically: Perform data compliance matching based on resistance characteristic analysis data to obtain matching matrix data; Perform threshold determination according to matching matrix data to obtain threshold determination data; Weight evaluation is performed based on the threshold judgment data to obtain compliance matrix data for auxiliary work in electrical performance testing of home wiring harnesses.

[0023] The compliance matching in the present invention can compare the resistance characteristic analysis data with the predefined standard database to quickly identify compliance and differences. The threshold judgment quantifies the resistance characteristic data as a result that exceeds or meets a specific threshold, providing a clear basis for judging the compliance of the characteristic data. The weight mechanism can highlight the role of key characteristics (such as peak change rate, etc.) in the judgment, ensuring the scientificity and reliability of the judgment results.

[0024] Preferably, the present application also provides a household wiring harness electrical performance detection system, which is used to perform the household wiring harness electrical performance detection method as described above, and the household wiring harness electrical performance detection system comprises: The original electrical parameter matrix acquisition module is used to collect the initial current and voltage data of the wiring harness using sensors to obtain the original electrical parameter matrix data; The original electrical parameter matrix building module is used to perform resistor modeling based on the original electrical parameter matrix data to obtain a resistor dynamic model; A resistance characteristic analysis module, used to perform resistance characteristic analysis according to a resistance dynamic model to obtain resistance characteristic analysis data; The resistance compliance determination module is used to perform resistance compliance determination based on resistance characteristic analysis data to obtain compliance matrix data for auxiliary work in electrical performance detection of household wiring harnesses.

[0025] The beneficial effect of the present invention is that the dynamic data of the harness current and voltage are collected by sensors to generate original electrical parameter matrix data, providing high-quality basic data for subsequent modeling and analysis. Multi-dimensional fusion technology is used to integrate multimodal signals (such as dynamic current characteristics and voltage characteristics) to ensure the comprehensiveness and consistency of the collected data. Dynamic current sampling technology can optimize the sampling frequency according to the current change characteristics, effectively capture the detailed changes of electrical signals, and avoid the generation of redundant data. Through resistance modeling, a resistance dynamic model is generated, which reflects the true performance of the harness resistance under different current and voltage conditions. Parameter optimization iteration improves the accuracy and adaptability of the model, so that the modeling results can cover a variety of usage scenarios and load conditions. Curve fitting processes nonlinear dynamic behavior and enhances the model's adaptability to complex signals. Feature space division and feature feature extraction analyze the resistance dynamic model from multiple dimensions and deeply explore key characteristics (such as stability, resistance change rate, etc.). Frequency domain analysis and clustering calculation further identify characteristic patterns and abnormal behaviors, providing support for the deep understanding of harness performance. The resistance characteristic analysis data generated by multi-feature fusion is characterized by comprehensiveness and strong expression ability, laying a scientific foundation for subsequent compliance determination. Data compliance matching automatically compares characteristic analysis data with industry standards to ensure the objectivity and accuracy of the judgment results. The weight evaluation mechanism dynamically adjusts the importance of each characteristic to make the judgment results more in line with actual application needs. The generated compliance matrix data provides quantitative support for the evaluation and improvement of wiring harness electrical performance, with traceability and operability. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting implementations made with reference to the following drawings: Figure 1 A flowchart showing a method for detecting electrical performance of a household wiring harness according to an embodiment of the present invention is provided; Figure 2 A flowchart showing a method for collecting an original electrical parameter matrix according to an embodiment is shown; Figure 3 A flowchart showing a method for constructing an original electrical parameter matrix according to an embodiment of the present invention is shown; Figure 4 A flowchart showing a method for analyzing resistance characteristics according to an embodiment of the present invention is provided; Figure 5 A flowchart of a resistor compliance determination method according to an embodiment is shown. DETAILED DESCRIPTION

[0027] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are 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 technicians in this field without creative work are within the scope of protection of the present invention.

[0028] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0029] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0030] The electrical performance test is carried out on the power supply harness of household electrical appliances. The existing working conditions are rated current 10A and working voltage 220V. The test goal is to evaluate the dynamic performance and compliance of the resistance of the harness under 10%-120% rated load. The test standard requires the resistance change rate under dynamic load to be <5% and the resistance stability to be >95%. High-precision current and voltage sensors are used to collect dynamic signals of the harness under different load conditions. The acquisition frequency is 10kHz, and the signal duration is 10 seconds to form the original electrical parameter matrix. After the signal noise is filtered, the original electrical parameter matrix is ​​formed.

[0031] Calculate the mean, variance, peak and other key characteristics of current and voltage: current mean: μI=10.1 A, voltage mean: μV=220.1 V, peak change rate: ΔIpeak=0.4 A / s. Initial resistance model formula: ; in, is the voltage parameter, The resistance change curve is calculated by the initial resistance model. The parameters are adjusted by optimization algorithms (such as particle swarm optimization) to generate a resistance dynamic model. , For the harness in time The instantaneous resistance value at is the amplitude of dynamic resistance change, is a dynamically disturbed sinusoidal wave function, describing the periodic fluctuation of resistance value over time. is the disturbance frequency, which indicates how fast the resistance changes dynamically.

[0032] Extract key characteristics from dynamic resistance data, including stability and rate of change. Stability refers to the magnitude of resistance fluctuations, while rate of change measures the rate at which resistance changes over time. Perform spectrum analysis on dynamic resistance data to extract the main frequency components. Main frequency =2π corresponds to a change period of 1s. The main frequency reflects the periodicity of the dynamic change of the resistance and provides a basis for distinguishing between the high dynamic range and the low dynamic range.

[0033] Based on the spectrum and characteristic data, the dynamic resistance data is divided into high dynamic interval and low dynamic interval. The division result can clearly show the dynamic behavior of the resistance characteristics in different time periods, providing grouping support for matching. The above characteristic data is matched with the predefined industry standards item by item. The stability requirement is ≥95%, and the change rate requirement is ≤5%. The matching result shows whether the dynamic resistance data meets the industry standards.

[0034] Calculate a comprehensive score based on the matching results. Calculate the score based on the percentage of the amplitude of the resistance fluctuation (e.g. 96%). Calculate the score based on the percentage of the resistance change rate (e.g. 4.8%). The comprehensive score quantifies the standardized characteristic values ​​for easy weight assignment and further analysis. Based on the comprehensive score, assign weights to different characteristics: Stability has a higher weight (because it affects the overall performance more significantly). Rate of change has a lower weight (because its fluctuation range is smaller). Generate a characteristic weight table to dynamically adjust the contribution of each characteristic to the final judgment. Combine the characteristic scores and weights to generate a compliance matrix. ,in is the compliance matrix data, is the stability compliance index, The compliance matrix clearly shows the degree of match between the dynamic resistance characteristics and industry standards, providing a basis for judgment.

[0035] Extract the stability and change rate of dynamic resistance. Spectrum analysis extracts the main frequency components, the main frequency =2π corresponds to a change period of 1s. Clustering calculation divides the data into two groups, high dynamic interval and low dynamic interval. Match the analysis characteristics with industry standards to obtain matching results, and calculate the comprehensive score for stability (95%) and change rate (<5%) based on the matching results. Combine the weights of different characteristics (weight calculation based on the comprehensive score) to generate a compliance matrix.

[0036] See also Figures 1 to 5 The present application provides a method for detecting electrical performance of household wiring harnesses, comprising the following steps: Step S1: using sensors to collect initial current and voltage data of the wiring harness to obtain original electrical parameter matrix data; Specifically, a sample of household wiring harness is connected to the test equipment. The current and voltage data of the wiring harness in the initial working state are collected by the current sensor and the voltage sensor respectively. Before collection, the sensor is calibrated by the standard voltage source and current source to ensure the accuracy of the measured data. The sensor collects the current and voltage signals of the wiring harness in real time according to the set sampling frequency (such as 1kHz). In order to ensure the integrity of the data, the sampling time should cover an entire cycle of normal operation of the wiring harness. The collected current and voltage signals are stored as time series data and are denoised (such as sliding average or discrete noise filtering) and expressed in matrix form: ; in is the original electrical parameter matrix data, For the The voltage at each sampling point, For the The voltage at each sampling point, For the The voltage at each sampling point, For the The current at each sampling point.

[0037] Step S2: Perform resistance modeling according to the original electrical parameter matrix data to obtain a resistance dynamic model; Specifically, the current and voltage matrix is ​​obtained, and the instantaneous resistance value of the wiring harness is calculated point by point by Ohm's law. For each sampling point, the corresponding instantaneous resistance is calculated ,in For the The voltage at each sampling point, For the According to the time series, all instantaneous resistance values ​​are Represented as a function model , which is used to reflect the change trend of resistance over time. During the dynamic modeling process, fitting processing will be performed, such as piecewise linear fitting or dynamic analysis based on specific conditional distribution, so that the model can better reflect the resistance change law of the actual harness. The output is regressed against the sampled data to ensure that the error is within an acceptable range (e.g., less than 5%).

[0038] Step S3: performing resistance characteristic analysis according to the resistance dynamic model to obtain resistance characteristic analysis data; Specifically, according to the dynamic model , further analyze the resistance characteristics. Trend analysis, calculate the average, maximum and minimum values ​​of resistance over time, and extract characteristic data. Stability analysis, calculate the rate of change of resistance, which is used to evaluate the fluctuation of harness resistance in different time periods. Large fluctuations may indicate potential failures. Compare with standard harness resistance characteristic data, calculate deviation, and mark it as abnormal if the deviation ratio exceeds the preset threshold (such as 10%).

[0039] Step S4: Perform resistance compliance determination based on the resistance characteristic analysis data to obtain compliance matrix data to assist in the electrical performance testing of home wiring harnesses.

[0040] Specifically, according to industry standards, thresholds are set for the average value, stability, and deviation rate of resistance. For example, the average resistance must be within the resistance standard ±10. The stability change rate must be less than a certain critical value (such as 0.1Ω / s). Each characteristic is judged one by one, and the judgment results are stored in the compliance matrix as 0 / 1. The compliance matrix data is combined with the harness identification to generate a test report for quality judgment and subsequent processing of home wiring harnesses.

[0041] Preferably, step S1 specifically includes: Step S11: deploying a distributed sensor network to collect the voltage and current of the wiring harness to obtain wiring harness state matrix data; Specifically, in order to cover different nodes of the harness, a distributed sensor network deployment is adopted. Each sensor node includes a voltage sensor and a current sensor, which are used to collect voltage and current data of the target harness in real time. According to the physical structure and function of the harness, key monitoring nodes are determined, such as joint positions, main loop areas, and branch line positions, and sensors are evenly deployed on these nodes, and the sensor positions are recorded. Each sensor collects voltage and current signals at a set sampling frequency (such as 1kHz) and records the timestamp. Each sensor sends data to the central data processing unit (CPU) through a wireless communication module (such as LoRa or Zigbee). After summarizing the voltage and current signals collected by each node, the harness state matrix is ​​constructed in the following way: the first column is voltage, the second column is current, the third column is the corresponding acquisition time, and the fourth column is the corresponding acquisition node.

[0042] Step S12: multi-dimensionally fuse the harness state matrix data to obtain multi-dimensional fused matrix data; Specifically, the core goal of multi-dimensional fusion is to integrate and align electrical data from different nodes to improve the relevance and expressiveness of the data. Since there are slight errors in the sampling time of each sensor, time synchronization is required to align the sampling points of all sensors to a unified time axis. The voltage and current data of adjacent nodes are fused by interpolation to calculate the equivalent voltage and current values ​​of the intermediate positions. For example, for the position between node 1 and node 2, the fused voltage can be expressed as: ,in is the voltage fusion data in the multi-dimensional fusion matrix data, is the spatial interpolation coefficient, which is dynamically adjusted according to the distance between nodes. is the first voltage data of node 1, is the first voltage data of node 2, Represents the spatial interpolation coefficient, which is dynamically adjusted according to the distance between nodes. Combining the geographical location, timestamp and sampling frequency of the nodes, the fused data is reorganized into a multi-dimensional fusion matrix, with the first column for voltage, the second column for current, and the third column for the corresponding acquisition time.

[0043] Step S13: Perform dynamic current sampling according to the multi-dimensional fusion matrix data to obtain dynamic current characteristic data; Specifically, dynamic current sampling is used to capture the changing characteristics of current in different time periods. The current data in the multi-dimensional fusion matrix is ​​segmented according to the preset time window (such as 10ms). In each time window, the maximum, minimum and mean values ​​of the current are calculated as the basic characteristic data of the dynamic current. The rate of change and fluctuation amplitude of the current are further calculated to describe the dynamic characteristics of the current.

[0044] Step S14: performing voltage analysis on the multi-dimensional fusion matrix data to obtain voltage characteristic data; Specifically, voltage analysis is mainly carried out for the voltage distribution characteristics of different nodes. The voltage difference between adjacent nodes is calculated to evaluate whether the conductivity of the wiring harness is uniform. The mean and standard deviation of each node voltage in a stable state are extracted to analyze whether there are obvious abnormal fluctuations. Time series analysis is performed on the voltage data to observe its changing trend and extract features such as peaks, troughs and zero crossings.

[0045] Specifically, more importantly, voltage time domain characteristics are extracted from multi-dimensional fusion matrix data to obtain voltage time domain characteristics data, wherein the voltage time domain characteristics data include voltage time domain mean data, voltage time domain change data, voltage time domain peak data and voltage time domain stability data; voltage frequency domain characteristics are analyzed based on the voltage time domain characteristics data to obtain voltage frequency domain characteristics data; the voltage frequency domain characteristics data are decomposed at a large scale and a small scale to obtain first preliminary voltage characteristic data and second preliminary voltage characteristic data; polynomial fitting is performed on the first preliminary voltage characteristic data to obtain voltage characteristic trend fitting data; long short-term memory network feature extraction is performed on the voltage characteristic trend fitting data to obtain first voltage characteristic data; wavelet packet decomposition is performed on the second preliminary voltage characteristic data to obtain small-scale multi-band voltage detail characteristic data; local change rate calculation is performed on the small-scale multi-band voltage detail characteristic data to obtain second voltage characteristic data; the first voltage characteristic data and the second voltage characteristic data are layered and fused to obtain voltage characteristic data.

[0046] Extract the time domain characteristics from the collected voltage data. Specifically, it includes mean data, which calculates the average level of the voltage signal in the time dimension and reflects the stable central value of the voltage in the overall state. Change data, by analyzing the fluctuation amplitude of the voltage in the time series, reflects the change trend of the voltage during operation. Peak data, extract the maximum and minimum values ​​in the voltage signal, and identify the peak characteristics of the voltage under sudden conditions. Stability data, analyze the discrete degree of voltage fluctuation, and judge the stability performance of the voltage by the data fluctuation range. Obtain voltage time domain characteristic data, including mean, change, peak and stability. These data can intuitively reflect the performance of the voltage in the overall operation process. Based on the extraction of time domain characteristics, the voltage signal is converted to the frequency domain, and the frequency domain characteristics of the voltage are extracted, including the main frequency component, and the dominant frequency component in the voltage signal is identified to reflect the main periodic characteristics of the voltage fluctuation. Frequency energy distribution, analyze the energy distribution of the voltage signal in different frequency bands, and determine the energy proportion of high frequency and low frequency. Frequency domain fluctuation characteristics, analyze the dynamic change characteristics of the voltage signal from the perspective of frequency, and identify abnormal high-frequency disturbances or low-frequency offsets. The voltage frequency domain characteristic data, including the main frequency, frequency energy distribution and fluctuation characteristics, are obtained to provide support for the dynamic analysis of the voltage signal. The global trend of the voltage signal is analyzed, and the voltage data is decomposed into the overall trend part to reflect the fluctuation law of the voltage over a long period of time. For example, the voltage may gradually drop slightly as the load increases, and this trend can be extracted by data smoothing and fitting. The voltage signal is analyzed in local detail, and the signal is decomposed into multiple high-frequency detail components to capture the dynamic fluctuation characteristics of the voltage in a short period of time. For example, the voltage may oscillate slightly at the moment of current startup, and this detail needs to be captured through fine decomposition. Large-scale voltage trend characteristic data and small-scale voltage detail characteristic data are obtained respectively. The voltage trend data obtained by large-scale decomposition is further analyzed for its overall change law and long-term fluctuation characteristics are extracted. For example, the overall change direction and trend point of the voltage are identified by smoothing the trend data. On the basis of small-scale decomposition, the local detail data is analyzed to identify the mutation point, rapid change area and oscillation characteristics of the voltage signal. For example, the rate of change of voltage in a short period of time can reflect the performance fluctuation of the wiring harness during startup or abnormal state. Large-scale voltage trend feature data and small-scale voltage detail feature data are extracted.

[0047] The first voltage characteristic data (large-scale trend) and the second voltage characteristic data (small-scale details) are fused and processed, and the proportion of global trend and local details in the overall characteristics is determined through dynamic weight allocation. The fused data can simultaneously reflect the overall operating trend and local dynamic details of the voltage, making the voltage characteristic data more comprehensive. The voltage characteristic data obtained comprehensively reflects the stability, volatility and deep-level dynamic structural characteristics of the voltage signal, providing high-quality data support for the electrical performance detection of home wiring harnesses.

[0048] Step S15: performing multi-modal data calibration on the multi-dimensional fusion matrix data according to the dynamic current characteristic data and the voltage characteristic data to obtain calibration matrix data; Specifically, the goal of multimodal calibration is to combine current and voltage features to correct the deviation data in the multidimensional fusion matrix. The dynamic current features and voltage features are associated with the values ​​of the original multidimensional fusion matrix at the corresponding time points to establish a preliminary association model. The association model is used to detect and correct abnormal data points, such as points where voltage and current do not match.

[0049] Step S16: performing matrix generation on the calibration matrix data to obtain original electrical parameter matrix data.

[0050] Specifically, the calibrated data is reintegrated and the original electrical parameter matrix is ​​generated. The calibrated voltage and current data are normalized to unify the different data ranges. With time as the row and nodes as the column, a standardized original electrical parameter matrix is ​​generated for analysis and processing.

[0051] Preferably, the dynamic current sampling is specifically: Performing regional change sampling according to the multi-dimensional fusion matrix data to obtain first dynamic current characteristic data; Specifically, the purpose of regional change sampling is to analyze the dynamic characteristics of current changes in the wiring harness in different physical areas. According to the physical length of the wiring harness and the sensor layout, the wiring harness is divided into several areas (such as area A, area B, area C, etc.), each of which contains several sampling points. For example, area A consists of nodes 1-5, and area B consists of nodes 6-10. In each area, the current values ​​of all nodes are averaged to represent the overall current state of the area. The rate of change and difference of the regional current are calculated to obtain the change characteristics. The difference is obtained by calculating the standard deviation of the current in the area. The current mean, rate of change and difference feature data of each area are organized to generate the first dynamic current feature data matrix, in which each row corresponds to the characteristics of a region.

[0052] Perform multi-frequency sampling according to the multi-dimensional fusion matrix data to obtain second dynamic current characteristic data; Specifically, multi-frequency sampling aims to extract the dynamic frequency characteristics of current through different sampling frequencies, set multiple sampling frequencies (such as 10Hz, 100Hz, 1kHz), and extract current data from the multidimensional fusion matrix respectively. Frequency feature extraction is performed for the data of each sampling frequency, and the following features are calculated. Mean and variance, calculate the mean and variance of the current signal at each frequency to reflect the overall characteristics of different frequencies; spectrum analysis, perform time-frequency domain analysis on the current signal, extract the main frequency components and their amplitudes, and record the frequency peak position; signal fluctuation characteristics, calculate the fluctuation range (maximum value-minimum value) at each frequency. Perform frequency comparison analysis on the data of each sampling frequency, compare the characteristics at each frequency, observe the changing trend of the current signal under different frequency sampling, and identify the abnormal frequency components in the signal. Feature data is constructed for the data of each sampling frequency, and the feature values ​​at multiple frequencies are integrated to form a second dynamic current feature data matrix.

[0053] Performing time-series correlation fusion on the first dynamic current characteristic data and the second dynamic current characteristic data to obtain dynamic current characteristic data; Specifically, the goal of time series correlation fusion is to combine regional features and multi-frequency features to generate dynamic current feature data, and align the first dynamic current feature data and the second dynamic current feature data according to the timestamps of the two sets of data. The current mean and variance in the multi-frequency feature data are mapped and analyzed with the mean and rate of change in the regional feature data. For example, the current mean in region A is correlated with the mean at a sampling frequency of 10 Hz, such as by using the covariance formula. The degree of matching between the current features and the frequency features in the region is evaluated by correlation calculation. Dynamic feature fusion, based on the results of time alignment, fuses the two sets of feature data in a weighted manner to obtain a dynamic current feature matrix.

[0054] The specific details of time series correlation fusion are: Extract dynamic characteristics according to the first dynamic current characteristic data and the second dynamic current characteristic data to obtain first characteristic vector data and second characteristic vector data respectively; Specifically, key characteristics such as current mean, change rate and difference characteristics are extracted from the first dynamic current characteristic data (regional change sampling characteristic matrix) to construct a regional characteristic vector. All regional characteristic vectors are arranged in chronological order to generate the first characteristic vector data. The current mean, variance and fluctuation characteristics at different sampling frequencies are extracted from the second dynamic current characteristic data (multi-frequency sampling characteristic matrix) to construct a frequency characteristic vector. All frequency characteristic vectors are arranged in chronological order to generate the second characteristic vector data. Each characteristic in the vector is normalized to eliminate dimensional differences.

[0055] Performing hierarchical Bayesian network processing on the first characteristic vector data and the second characteristic vector data to obtain first hierarchical characteristic data and second hierarchical characteristic data respectively; Specifically, a two-layer Bayesian network is constructed for each characteristic vector. The first layer represents characteristic nodes (such as mean, rate of change, etc.), and the second layer represents associated characteristics (such as the interaction between fluctuation characteristics and rate of change). Each node in the network represents a characteristic, and the edges between nodes represent the probabilistic dependency between characteristics. Based on historical data, the joint distribution between characteristics is statistically analyzed to construct a conditional probability table. For example, the joint distribution of the regional current mean and rate of change can be obtained by statistically analyzing the following probabilities: ,in is the joint distribution of the regional current mean and change rate, is the frequency of the regional current mean and change rate, is the total number of samples. For frequency characteristics, the probability dependence of the association between the mean and the fluctuation characteristics at different frequencies is calculated. Using the conditional probability inference method, the network state is updated according to the current input data, and the posterior probability of each characteristic is calculated. For example, after inputting the mean data, the posterior distribution of the change rate is inferred to generate the first hierarchical characteristic data and the second hierarchical characteristic data, which respectively reflect the hierarchical structure of the regional characteristics and the frequency characteristics.

[0056] Constructing a time series characteristic interaction relationship according to the first layer characteristic data and the second layer characteristic data to obtain interaction characteristic matrix data; Specifically, the first hierarchical characteristic data and the second hierarchical characteristic data are time-aligned to ensure that the two characteristic data are consistent at the same time point. The similarity and interaction between the regional characteristics and the frequency characteristics are cross-calculated. For example, the relationship between vectors is calculated using cosine similarity. The dynamic change pattern between the characteristics is further analyzed to extract the associated characteristics. For example, the association formula between the regional change rate and the frequency fluctuation characteristic is the rate of change of the regional current multiplied by the fluctuation amplitude of the current in the region. The rate of change of the regional current is the change of the regional average current over time within a specific time window, and the fluctuation amplitude of the current in the region is the maximum change of the current within the specified time window. The fluctuation amplitude of the current in the region is the maximum change of the current within the specified time window. Construct an interactive characteristic matrix to represent the interactive relationship between the first hierarchical characteristics and the second hierarchical characteristics. The row represents the interaction of the characteristic dimension (such as current fluctuation), including the strength of its association with itself, and the interactive relationship with other characteristic dimensions (such as voltage change, resistance fluctuation). The column represents the interaction of the characteristic dimension, including the strength of its relationship with other characteristic dimensions (for example, the impact of voltage change on current fluctuation in different regions).

[0057] Multi-layer characteristic fusion is performed on the first layer characteristic data and the second layer characteristic data according to the interactive characteristic matrix data to obtain dynamic current characteristic data.

[0058] Specifically, all elements in the interaction characteristic matrix are accumulated, which represents the total strength of the interaction relationship between all characteristics in the matrix. By accumulating the values ​​in the interaction matrix, the contribution of the first layer characteristic data in the overall interaction can be evaluated. The sum of the interaction characteristic values ​​is used as the weight of the first layer characteristic data. The larger the weight value, the more significant the contribution of the first layer characteristic to the overall interaction. The weight of the second layer characteristic data is calculated by normalizing the weight. The weight of the second layer characteristic is equal to 1-the weight of the first layer characteristic data, ensuring that the total weight remains consistent. The first and second layer characteristic data are weighted and fused. According to the weight, the first layer characteristic data is weighted and fused with the second layer characteristic data. The contribution of the first layer characteristic data is adjusted by the weight of the first layer characteristic data, and the contribution of the second layer characteristic data is adjusted by the weight of the second layer characteristic data. The fused data retains the main information of the two layers of characteristic data at the same time, and optimizes their performance in the overall dynamic characteristics through weights. The fused dynamic characteristic data is sorted according to the time series to ensure the consistency of the data on the time axis, including arranging the characteristic values ​​of each time point to form an ordered time series dynamic characteristic data matrix. The dynamic feature data matrix contains all the fused temporal features, specifically including the fusion results of the two hierarchical feature data, and integrates the temporal dynamic changes of the two.

[0059] Preferably, the regional change sampling is specifically: Perform time series segmentation according to the multi-dimensional fusion matrix data to obtain matrix time series segmentation data; Specifically, based on the time dimension, the multidimensional fusion matrix data is divided into several sub-matrices according to a fixed time window to capture the changing characteristics of the harness current in different time periods. A time window length (such as 100ms) is preset to ensure that there are enough sampling points in each window to calculate the rate of change. For the multidimensional fusion matrix, it is divided according to the range of the timestamp to generate sub-matrices, each of which contains the current, voltage and other information of all nodes in the time window.

[0060] Performing global current change rate calculation and local current change rate calculation on the matrix time series segmentation data to obtain global current change rate data and local current change rate data respectively; Specifically, for each time series submatrix, the average current change rate of all nodes is calculated, and the global current change rate of all time windows is combined into a time series. For each time series submatrix, the current change rate of each node in the window is calculated separately, and the local change rate of each node is recorded as local current change rate data according to the time window.

[0061] Performing regional classification on the multi-dimensional fusion matrix data according to the global current change rate data and the local current change rate data to obtain regional classification data; Specifically, the global change rate and local change rate of each time window are compared, and nodes with significant deviations in change rate are marked. For example, if the local change rate exceeds the threshold compared to the global change rate, the node is marked as a high change area. High change area: the node current change rate deviates significantly from the global average. Low change area: the node current change rate is close to the global average.

[0062] Allocate sampling rate according to regional classification data to obtain sampling rate matrix data; Specifically, high-variability areas need to be assigned a higher sampling rate to capture fast dynamic changes. Low-variability areas can be assigned a lower sampling rate to reduce redundant data. The sampling rate of high-variability areas is set to 1kHz, and the sampling rate of low-variability areas is set to 100Hz, for example.

[0063] The current of the wiring harness is collected according to the sampling rate matrix data to obtain the wiring harness sampling change data; Specifically, the sampling frequency of each node is adjusted in real time according to the setting of the sampling rate matrix. The current data collected by the high sampling rate node will cover a finer time resolution, and the data collected by the low sampling rate node will be sparser. The collected node current data is stored in chronological order to form the wiring harness sampling change data.

[0064] Perform multi-frequency data fusion based on the wire harness sampling change data to obtain preliminary reconstructed matrix data; Specifically, the data at different sampling rates are realigned to a unified time axis, and the gaps in the low sampling rate data are processed using interpolation. For the data of high sampling rate and low sampling rate nodes, the weighted average is calculated in combination with the time window to fuse the current characteristics at different frequencies. The fused data is organized into a preliminary reconstruction matrix.

[0065] Feature extraction is performed based on the preliminary reconstructed matrix data to obtain first dynamic current feature data.

[0066] Specifically, the mean, variance, change rate and fluctuation amplitude of each node current are calculated, the characteristic vector is extracted, and the characteristic vectors of all nodes are combined to generate a first dynamic current characteristic data matrix.

[0067] Preferably, the multi-frequency sampling is specifically: Perform spectrum graph convolution calculation based on multi-dimensional fusion matrix data to obtain matrix convolution layer data; Specifically, the current signal in the multi-dimensional fusion matrix data is fast Fourier transformed to obtain a spectrum diagram, in which each column corresponds to the frequency component of a node. The spectrum data of all nodes are stacked according to the frequency dimension to generate a spectrum matrix. The convolution kernel (such as a 3×3 window) is set to perform a convolution operation on the spectrum matrix to extract local frequency features. After the convolution operation, matrix convolution layer data is generated, which represents the local characteristics of the spectrum in different frequency intervals.

[0068] According to the matrix convolution layer data, the key frequency band is identified to obtain the key frequency band data; Specifically, the energy value of the convolution layer data in each frequency interval is calculated, and the frequency bands with the highest energy values ​​are selected as key frequency bands according to the energy value sorting. The screening criteria can be a certain proportion of the total energy (such as the frequency bands with the top 90% energy). The frequency range and energy value of the key frequency bands are recorded to generate a key frequency band data table.

[0069] Frequency band priority is divided according to key frequency band data to obtain frequency band priority data; Specifically, the priorities are sorted according to the energy values ​​of the key frequency bands, and the higher the energy, the higher the priority. The priorities can be divided into three levels: high, medium, and low, or more levels. The priority of each key frequency band is recorded as a matrix.

[0070] Initialize the frequency allocation strategy according to the frequency band priority data to obtain the frequency allocation matrix data; Specifically, a higher sampling frequency is assigned to high-priority bands to capture detailed features. A lower sampling frequency is assigned to medium and low-priority bands to save resources. A frequency allocation matrix is ​​constructed, where each row represents the sampling frequency of a band.

[0071] Performing parallel multi-frequency signal sampling on the current of the wiring harness according to the frequency allocation matrix data to obtain parallel sampling matrix data; Specifically, according to the frequency allocation matrix, sensors are set to sample data of different frequency bands in parallel. The sampling rate of high-priority frequency bands is higher, and the sampling rate of medium and low-priority frequency bands is lower. The sampling data of all frequency bands are arranged in chronological order to form a parallel sampling matrix.

[0072] Extracting the frequency division domain signal characteristics of the parallel sampling matrix data to obtain the frequency division characteristic matrix data; Specifically, for the data of each frequency band, frequency domain characteristics are extracted, such as amplitude, phase, spectrum peak, etc. The characteristic vectors of all frequency bands are summarized to generate a frequency division characteristic matrix.

[0073] Performing time domain characteristic fusion on the frequency division characteristic matrix data to obtain time domain characteristic fusion data; Specifically, the characteristics of different frequency bands are time-aligned to ensure that the data of all frequency bands are aligned at the same time point. The frequency division characteristics are weighted and fused according to the frequency band priority to generate a time domain characteristic fusion data sequence for multi-scale analysis.

[0074] Multi-scale feature extraction is performed on the time domain characteristic fusion data to obtain second dynamic current characteristic data.

[0075] Specifically, the time domain fusion data is decomposed at multiple scales, for example, by sliding statistics with different window sizes to extract characteristics at different time scales. Dynamic characteristics such as mean, variance, and rate of change at multiple scales are extracted to construct characteristic vectors. The characteristic vectors at all time scales are summarized to generate the second dynamic current characteristic data.

[0076] Preferably, the multimodal data calibration is specifically: According to the dynamic current characteristic data and the voltage characteristic data, the multi-dimensional fusion matrix data is subjected to inter-modal alignment learning to obtain the modal alignment weight matrix data; Specifically, the dynamic current feature data and voltage feature data come from different modes, and there are differences in time and space distribution. Through modal alignment learning, the data of the two modes are uniformly aligned. After the current features and voltage features are aligned in time, the feature similarity at each time point is calculated, and the modal alignment weight is assigned according to the similarity value. If a certain mode contributes more to the current feature, its weight increases accordingly.

[0077] Perform characteristic time matching on the modal alignment weight matrix data to obtain characteristic time matching data; Specifically, there is a sampling time deviation between the dynamic current and voltage features. Calculate the time offset of the two modal features ,in To indicate that the variable index that makes the function value reach the maximum value is returned, the time offset is returned , making the current characteristic and voltage characteristics The cross-correlation value reaches the maximum. For all possible time offsets The maximum value of the cross-correlation value of the current and voltage characteristics, is the index of the current and voltage characteristics at a certain moment in time, It is used to quantify the similarity or match between two time series, especially the correlation when considering different time offsets. To indicate the time point The current characteristic value on is a dynamic current characteristic sequence that changes with time. To indicate the time point The voltage characteristic value on Relative to a point in time The offset is a dynamic voltage characteristic sequence that changes over time. , adjust the time axis so that the two modal features are completely aligned in time. Rearrange the aligned features into a time series matrix.

[0078] Perform multimodal feature completion on the feature time matching data to obtain completed feature matrix data; Specifically, check for missing values ​​in the feature time matching data and mark incomplete timing points in the current or voltage features. For example, for the tth moment, if the current feature value is missing, it is marked as NaN. Based on the modal alignment weight matrix, interpolation is performed using the characteristics of another modality. , For at time point The supplementary characteristic value is calculated by combining the voltage characteristic and the current characteristic. For at time point The importance weight of voltage characteristics in the completion process, For at time point The voltage characteristic value on For at time point If a certain modal characteristic is missing, the weight of another modal characteristic is used to complete it. The completed multi-modal characteristic data is recombined into a matrix.

[0079] Performing modal normalization processing on the completed characteristic matrix data to obtain normalized calibration matrix data; Specifically, the dimension difference of the characteristic value is eliminated to ensure that the characteristic data is in the same numerical range, which is convenient for fusion processing. The minimum-maximum normalization is applied to each column of data, and the normalized characteristic value range is [0,1]. The normalized multimodal characteristic data is recorded as a normalized calibration matrix.

[0080] Perform attention fusion on the normalized calibration matrix data to obtain the calibration matrix data.

[0081] Specifically, the attention mechanism dynamically adjusts the weights of current and voltage characteristics at different time points to highlight key characteristics. Calculate the attention distribution at each time point: , ,in and Respectively represent the current and voltage characteristics at the time point The attention weight on is the current characteristic at the current time point t, is the voltage characteristic at the current time point t. According to the attention weight, the current or voltage before this moment is weighted and fused to obtain the current data and voltage data at the new time point, which are rearranged into a calibration matrix. For example, the current data at the third moment is weighted for attention calculation, and the current data of the first three points are multiplied by the attention weight and added and divided by 3 to obtain the new current attention weighted data, which is the new current data at the third moment.

[0082] Preferably, step S2 specifically includes: Step S21: extracting matrix features according to the original electrical parameter matrix data to obtain electrical matrix feature data; Specifically, the original electrical parameter matrix contains the sequence data of the current and voltage of the wiring harness changing over time. Through feature extraction, key data related to resistance (such as current mean, voltage mean, correlation between voltage and current, etc.) are mined. For the current and voltage data at each time point, the following characteristics are calculated, such as the average value, mutation point, and instantaneous value of resistance. The maximum, minimum, and standard deviation of the characteristics of current, voltage, and resistance are extracted. The time rate of change of current and voltage is calculated, and the mean and maximum fluctuation range of the rate of change are extracted. The above characteristics are organized into electrical matrix feature data.

[0083] Step S22: performing feature mapping modeling on the electrical matrix feature data to obtain an initial resistance model; Specifically, the extracted electrical features are used to construct a mapping relationship between resistance, current, and voltage to form an initial resistance model. Resistance R(t) is a nonlinear function of current I(t) and voltage V(t), and the expression can be expressed as: R(t)=f(I(t),V(t)). Linear or nonlinear regression methods are used to fit the features and establish a preliminary mapping relationship. For example, the relationship between resistance and features is fitted by the least squares method: Where a1, a2, …, a5 are model coefficients. Perform residual analysis on the fitting results to evaluate the fitting effect of the model. If the residual is large, it is necessary to adjust the feature mapping formula or introduce high-order features.

[0084] Step S23: performing curve fitting on the initial resistance model to obtain a resistance model; Specifically, curve fitting is performed on the output of the initial resistance model so that the model can more accurately describe the change pattern of resistance. The data output by the initial resistance model is matched with time, and multiple fitting functions (such as polynomial, exponential function or piecewise function) are tried for fitting. If the data is periodic, a sine function or Fourier series is used for fitting. The error (such as mean square error or absolute error) between the fitting curve and the actual data is calculated, and the fitting parameters are adjusted to optimize the fitting effect. The fitting function is used as the mathematical expression of the resistance model.

[0085] Step S24: performing parameter optimization iteration on the resistor model to obtain a resistor dynamic model.

[0086] Specifically, the parameters in the resistance model (such as fitting coefficients) are optimized so that the model can more accurately reflect dynamic changes. Model parameters are optimized using methods such as gradient descent or genetic algorithms: ,in After optimization calculation, the resistance model The updated value of the parameter, The resistance model before this iteration. The current value of the parameter, To adjust the proportional coefficient of each parameter update step, it is a positive value less than 1 (such as 0.01). is a function that measures the error between the resistance model prediction and the actual value. The resistor dynamic model The fitting parameters of the parameters, such as the The coefficient of the parameter or the The weight of each parameter is calculated. Under dynamic input conditions (such as different scenarios where current and voltage change over time), the model is tested to verify whether the dynamic resistance value output by it is consistent with the actual observed value. The expression of the resistance model is updated according to the optimized parameters to generate a resistance dynamic model. The resistance dynamic model can calculate the change value of the resistance in real time based on the input of current and voltage.

[0087] Preferably, step S3 specifically includes: Step S31: performing model input analysis according to the resistor dynamic model to obtain analysis input data; Specifically, the input variables of the resistor dynamic model are analyzed to extract the characteristics of core parameters such as current, voltage and time. The input parameters, such as current and voltage parameters at each moment, are analyzed from the resistor dynamic model. The input parameters are statistically analyzed by time period and key characteristics, such as average and change, are calculated. The input characteristics within each time window are recorded to generate an analytical input data table.

[0088] Step S32: dividing the characteristic space according to the analytical input data and the resistance dynamic model to obtain characteristic dimension data; Specifically, the characteristics of the analyzed input data are spatially divided to clarify the multi-dimensional relationship between current, voltage and resistance. A three-dimensional characteristic space is constructed with current, voltage and time as coordinate axes. The characteristic space is divided into multiple sub-regions according to the value range of current and voltage. For example, the current and voltage are segmented. Each sub-region represents a characteristic dimension. The boundary values ​​and time ranges of each sub-region are recorded.

[0089] Step S33: extracting characteristic features according to the characteristic dimension data to obtain characteristic feature data; Specifically, extract the core characteristics of the resistor in each characteristic dimension (such as mean value, change rate, etc.). For the resistor data in each characteristic dimension, calculate the following characteristics: resistance mean value, resistance change rate, and resistance fluctuation range. Record the calculation results as a characteristic feature data table.

[0090] Step S34: performing frequency domain analysis according to the characteristic feature data to obtain frequency domain characteristic data; Specifically, the frequency characteristics of the resistance characteristics changing over time are mined through frequency domain analysis. The resistance data of each characteristic dimension is Fourier transformed to obtain the frequency components. The main frequency components and their amplitudes are extracted, and the energy characteristics of the frequency distribution are recorded. The frequency domain characteristics of each dimension are recorded as frequency domain characteristic data.

[0091] Step S35: performing clustering calculation on the frequency domain characteristic data to obtain frequency domain characteristic clustering data; Specifically, the data of each dimension is clustered according to the frequency domain characteristics to identify the resistance characteristic grouping. The frequency component, amplitude and energy are selected as the cluster feature vectors, and the characteristic vectors are clustered using an unsupervised clustering method to obtain cluster labels. The cluster labels of each dimension are recorded as frequency domain characteristic clustering data.

[0092] Step S36: Perform multi-characteristic fusion analysis based on the frequency domain characteristic clustering data to obtain resistance characteristic analysis data.

[0093] Specifically, based on the clustering results, various frequency domain characteristics are fused to generate a unified resistance characteristic analysis result. The average characteristics in each cluster are calculated, and the common features of various frequency domain characteristics, such as average frequency, average amplitude, and energy range, are extracted. The fusion results of all categories are integrated to generate resistance characteristic analysis data.

[0094] Preferably, step S4 is specifically: Step S41: performing data compliance matching according to the resistance characteristic analysis data to obtain matching matrix data; Specifically, the resistance characteristic analysis data is matched with the standard resistance characteristic data to determine the deviation of the actual measured value. The reference characteristic data of the resistor (such as standard frequency, amplitude range, etc.) is extracted from the standard database. The measured characteristics are matched with the standard characteristics to calculate the deviation. A matching matrix is ​​generated to record the deviation of each characteristic dimension.

[0095] Step S42: Perform threshold determination according to the matching matrix data to obtain threshold determination data; Specifically, set a deviation threshold for each characteristic (for example: frequency deviation <1%, amplitude deviation <5%, energy deviation <10%). Define compliance conditions, and meet the above conditions to be qualified. For each row of matching matrix data, check whether the compliance conditions are met one by one, and generate a binary judgment result (1 means compliance, 0 means non-compliance). Record the judgment results of each characteristic dimension and generate threshold judgment data.

[0096] Step S43: Perform weight evaluation based on the threshold judgment data to obtain compliance matrix data to assist in the detection of electrical performance of household wiring harnesses.

[0097] Specifically, a comprehensive compliance score is generated by evaluating the importance of different features. A weight is assigned to each feature, and the weight can be based on the actual impact of the feature (for example, frequency has a higher weight than energy). The comprehensive compliance score of each feature dimension is calculated based on the threshold judgment data and the weight vector, and the compliance score range can be normalized to [0,1]. The comprehensive score of each feature dimension is recorded as a compliance matrix.

[0098] Preferably, the present application also provides a household wiring harness electrical performance detection system, which is used to perform the household wiring harness electrical performance detection method as described above, and the household wiring harness electrical performance detection system comprises: The original electrical parameter matrix acquisition module is used to collect the initial current and voltage data of the wiring harness using sensors to obtain the original electrical parameter matrix data; The original electrical parameter matrix building module is used to perform resistor modeling based on the original electrical parameter matrix data to obtain a resistor dynamic model; A resistance characteristic analysis module, used to perform resistance characteristic analysis according to a resistance dynamic model to obtain resistance characteristic analysis data; The resistance compliance determination module is used to perform resistance compliance determination based on resistance characteristic analysis data to obtain compliance matrix data for auxiliary work in electrical performance detection of household wiring harnesses.

[0099] Therefore, from any point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the attached application documents rather than the above description, and it is intended that all changes falling within the meaning and scope of equivalent elements of the application documents are included in the present invention.

[0100] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for detecting electrical performance of household wiring harnesses, characterized in that: The following steps are involved: Step S1: using sensors to collect initial current and voltage data of the wiring harness to obtain original electrical parameter matrix data; Step S2: Perform resistance modeling according to the original electrical parameter matrix data to obtain a resistance dynamic model; Step S3: performing resistance characteristic analysis according to the resistance dynamic model to obtain resistance characteristic analysis data; Step S4: Perform resistance compliance determination based on the resistance characteristic analysis data to obtain compliance matrix data to assist in the electrical performance testing of home wiring harnesses.

2. The method according to claim 1, characterized in that Step S1 is specifically as follows: Deploy a distributed sensor network to collect the voltage and current of the wiring harness and obtain the wiring harness state matrix data; Perform multi-dimensional fusion on the harness state matrix data to obtain multi-dimensional fusion matrix data; Perform dynamic current sampling based on multi-dimensional fusion matrix data to obtain dynamic current characteristic data; Perform voltage analysis on the multi-dimensional fusion matrix data to obtain voltage characteristic data; Performing multi-modal data calibration on the multi-dimensional fusion matrix data according to the dynamic current characteristic data and the voltage characteristic data to obtain calibration matrix data; Perform matrix generation on the calibration matrix data to obtain the original electrical parameter matrix data.

3. The method according to claim 2, characterized in that The dynamic current sampling is as follows: Performing regional change sampling according to the multi-dimensional fusion matrix data to obtain first dynamic current characteristic data; Perform multi-frequency sampling according to the multi-dimensional fusion matrix data to obtain second dynamic current characteristic data; Performing time-series correlation fusion on the first dynamic current characteristic data and the second dynamic current characteristic data to obtain dynamic current characteristic data; The specific details of time series correlation fusion are: Extract dynamic characteristics according to the first dynamic current characteristic data and the second dynamic current characteristic data to obtain first characteristic vector data and second characteristic vector data respectively; Performing hierarchical Bayesian network processing on the first characteristic vector data and the second characteristic vector data to obtain first hierarchical characteristic data and second hierarchical characteristic data respectively; Constructing a time series characteristic interaction relationship according to the first layer characteristic data and the second layer characteristic data to obtain interaction characteristic matrix data; Multi-layer characteristic fusion is performed on the first layer characteristic data and the second layer characteristic data according to the interactive characteristic matrix data to obtain dynamic current characteristic data.

4. The method according to claim 3, characterized in that The specific regional change sampling is as follows: Perform time series segmentation according to the multi-dimensional fusion matrix data to obtain matrix time series segmentation data; Performing global current change rate calculation and local current change rate calculation on the matrix time series segmentation data to obtain global current change rate data and local current change rate data respectively; Performing regional classification on the multi-dimensional fusion matrix data according to the global current change rate data and the local current change rate data to obtain regional classification data; Allocate sampling rate according to regional classification data to obtain sampling rate matrix data; The current of the wiring harness is collected according to the sampling rate matrix data to obtain the wiring harness sampling change data; Perform multi-frequency data fusion based on the wire harness sampling change data to obtain preliminary reconstructed matrix data; Feature extraction is performed based on the preliminary reconstructed matrix data to obtain first dynamic current feature data.

5. The method according to claim 3, characterized in that: The multi-frequency sampling is specifically: Perform spectrum graph convolution calculation based on multi-dimensional fusion matrix data to obtain matrix convolution layer data; According to the matrix convolution layer data, the key frequency band is identified to obtain the key frequency band data; Frequency band priority is divided according to key frequency band data to obtain frequency band priority data; Initialize the frequency allocation strategy according to the frequency band priority data to obtain the frequency allocation matrix data; Performing parallel multi-frequency signal sampling on the current of the wiring harness according to the frequency allocation matrix data to obtain parallel sampling matrix data; Extracting the frequency division domain signal characteristics of the parallel sampling matrix data to obtain the frequency division characteristic matrix data; Performing time domain characteristic fusion on the frequency division characteristic matrix data to obtain time domain characteristic fusion data; Multi-scale feature extraction is performed on the time domain characteristic fusion data to obtain second dynamic current characteristic data.

6. The method according to claim 2, characterized in that The multimodal data calibration is specifically as follows: According to the dynamic current characteristic data and the voltage characteristic data, the multi-dimensional fusion matrix data is subjected to inter-modal alignment learning to obtain the modal alignment weight matrix data; Perform characteristic time matching on the modal alignment weight matrix data to obtain characteristic time matching data; Perform multimodal feature completion on the feature time matching data to obtain completed feature matrix data; Performing modal normalization processing on the completed characteristic matrix data to obtain normalized calibration matrix data; Perform attention fusion on the normalized calibration matrix data to obtain the calibration matrix data.

7. The method according to claim 1, characterized in that Step S2 is specifically as follows: Matrix feature extraction is performed according to the original electrical parameter matrix data to obtain electrical matrix feature data; Perform feature mapping modeling on the electrical matrix feature data to obtain an initial resistance model; Performing curve fitting on the initial resistance model to obtain a resistance model; The resistor model is iterated by parameter optimization to obtain a resistor dynamic model.

8. The method according to claim 1, characterized in that Step S3 is specifically as follows: Perform model input analysis according to the resistor dynamic model to obtain analytical input data; The characteristic space is divided according to the analytical input data and the resistance dynamic model to obtain the characteristic dimension data; Extract characteristic features according to characteristic dimension data to obtain characteristic feature data; Perform frequency domain analysis according to the characteristic feature data to obtain frequency domain characteristic data; Performing clustering calculation on the frequency domain characteristic data to obtain frequency domain characteristic clustering data; Multi-characteristic fusion analysis is performed based on frequency domain characteristic clustering data to obtain resistance characteristic analysis data.

9. The method according to claim 1, characterized in that: Step S4 is specifically as follows: Perform data compliance matching based on resistance characteristic analysis data to obtain matching matrix data; Perform threshold determination according to matching matrix data to obtain threshold determination data; Weight evaluation is performed based on the threshold judgment data to obtain compliance matrix data for auxiliary work in electrical performance testing of home wiring harnesses.

10. A system for detecting electrical performance of household wiring harnesses, characterized in that: Used to perform the household wiring harness electrical performance detection method as claimed in claim 1, the household wiring harness electrical performance detection system comprises: The original electrical parameter matrix acquisition module is used to collect the initial current and voltage data of the wiring harness using sensors to obtain the original electrical parameter matrix data; The original electrical parameter matrix building module is used to perform resistor modeling based on the original electrical parameter matrix data to obtain a resistor dynamic model; A resistance characteristic analysis module, used to perform resistance characteristic analysis according to a resistance dynamic model to obtain resistance characteristic analysis data; The resistance compliance determination module is used to perform resistance compliance determination based on resistance characteristic analysis data to obtain compliance matrix data for auxiliary work in electrical performance detection of household wiring harnesses.

Images