A terminal machine monitoring management system for wire harness production

Abstract

This invention provides a terminal crimping machine monitoring and management system for wire harness production, relating to the field of wire harness terminal crimping monitoring technology. The system includes: real-time dynamic thresholding of the digitally sampled signal based on its amplitude and slope characteristics to extract effective waveform data segments; dividing the discrete transient data sequence within the effective waveform data segments into several continuous sub-intervals at equal intervals according to time sequence; within each sub-interval, connecting adjacent sampling points with straight line segments to form trapezoids; calculating the area of ​​each trapezoid and summing the results to obtain an integral calculation result; and obtaining an integral curve reflecting the energy accumulation characteristics of the entire crimping process based on the integral calculation result. This invention provides real-time monitoring and intelligent management of the terminal crimping machine's operating status, improving the quality and efficiency of wire harness production.

Description

Technical Field

[0001] This invention relates to the field of wire harness terminal crimping monitoring technology, and in particular to a terminal crimping monitoring and management system for wire harness production. Background Technology

[0002] In the field of industrial wire harness production, online monitoring of terminal crimping quality is of certain significance for ensuring the reliability of electrical connections. By setting a fixed amplitude threshold, when the sensor signal exceeds the threshold, data interception is triggered. Then, the entire crimping waveform obtained by interception is discretely integrated to calculate the total energy value. The total energy value is then compared with the preset normal range to determine whether the crimping is qualified.

[0003] In practical applications, taking the crimping production of a certain automotive wiring harness terminal as an example, it has been found that the wire diameter of different batches of wiring harnesses may have slight fluctuations, and the hardness of the terminal material may also have some differences. These factors will cause changes in the dynamic range of the crimping force signal amplitude. In this case, when using a fixed threshold interception method, if the threshold is set relatively high, for crimping processes with thinner wire diameters or softer materials, the signal amplitude may sometimes be difficult to stably trigger interception, which may lead to the loss of some data in the effective waveform segment. If the threshold is set relatively low, mechanical no-load fluctuations or electrical noise before crimping may be mistakenly included in the effective data segment, resulting in some invalid components in the subsequent integral calculation results. In addition, usually only the total integral area under the entire crimping waveform is calculated, while less attention is paid to the details of the cumulative change of energy over time during the crimping process. For example, in different stages of crimping, such as the initial deformation of the metal wire, the compression of the insulation layer in the middle stage, and the curling and forming of the terminal in the later stage, the change in the energy accumulation rate often reflects different defect types, such as insufficient insulation layer indentation or metal wire scattering. However, relying solely on a single total energy value may make it difficult to fully capture these temporal dynamic characteristics. Summary of the Invention

[0004] This invention provides a terminal block machine monitoring and management system for wire harness production, which monitors the terminal block machine's operating status in real time and manages it intelligently, thereby improving the quality and efficiency of wire harness production.

[0005] To solve the above-mentioned technical problems, the technical solution of the present invention is as follows:

[0006] In a first aspect, a terminal block monitoring and management system for wire harness production includes:

[0007] The conditioning module is used to perform anti-aliasing filtering on the differential voltage signal to obtain the processed analog signal; it performs adaptive gain adjustment according to the dynamic range of the analog signal amplitude to obtain the amplitude-matched conditioning signal, and converts the amplitude-matched conditioning signal into a digital sampling signal through analog-to-digital conversion;

[0008] The extraction module is used to perform real-time dynamic thresholding of the digitally sampled signal based on the amplitude and slope characteristics of the digitally sampled signal to extract the effective waveform data segment; the discrete transient data sequence in the effective waveform data segment is divided into several continuous sub-intervals at equal intervals according to the time sequence; within each sub-interval, adjacent sampling points are connected by straight line segments to form a trapezoid; the area of ​​each trapezoid is calculated and summed to obtain the integral operation result; based on the integral operation result, an integral curve reflecting the energy accumulation characteristics of the entire pressing process is obtained;

[0009] The analysis module is used to construct a reference curve under normal crimping conditions based on the integral curve, and compare the difference between the integral curve measured in the current crimping cycle and the reference curve to obtain the difference feature data sequence.

[0010] The control module is used to determine the current crimping quality status of the terminal in real time based on whether the difference feature data sequence exceeds the preset boundary envelope range. If the crimping quality status is determined to be unqualified, the control command is obtained through the preset PLC intelligent control component to perform closed-loop control on the crimping distance and crimping force of the terminal machine, so as to realize the process adjustment and quality control of the industrial wire harness production site.

[0011] In a second aspect, a computer-readable storage medium storing a program that, when executed by a processor, implements the method.

[0012] The above-described solution of the present invention has at least the following beneficial effects:

[0013] By performing anti-aliasing filtering, adaptive gain conditioning, and analog-to-digital conversion on the differential voltage signal during the terminal crimping process, on-site interference is suppressed and signal amplitude matching is ensured, improving the stability of the original sampled data. Real-time dynamic threshold truncation is achieved based on amplitude and slope characteristics to extract effective waveform segments reflecting the actual crimping process. An energy accumulation integral curve is constructed by accumulating and integrating the trapezoidal surface in sub-intervals. The difference between the measured integral curve and the reference curve is compared, and quality judgment is performed in combination with the preset boundary envelope range. This achieves online real-time identification of terminal crimping quality. At the same time, relying on the PLC intelligent control component, the crimping distance and crimping force are controlled in a closed loop. When unqualified conditions are found, the process parameters can be corrected in time, realizing automated control of terminal crimping quality in the wire harness production process. Attached Figure Description

[0014] Figure 1 This is a schematic diagram of a terminal block monitoring and management system for wire harness production provided by an embodiment of the present invention.

[0015] Figure 2 This is a schematic diagram of the integral curve difference comparison step provided in the embodiment of the present invention.

[0016] Figure 3 This is a graph showing the relationship between the width of the dynamic threshold truncation window and the number of triggered events, provided by an embodiment of the present invention.

[0017] Figure 4 This is a schematic diagram of trapezoidal integral segmented surface accumulation provided by an embodiment of the present invention.

[0018] Figure 5 This is a simulation diagram of the closed-loop adjustment selection convergence process provided in an embodiment of the present invention. Detailed Implementation

[0019] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

[0020] like Figure 1 As shown, an embodiment of the present invention proposes a terminal block machine monitoring and management system for wire harness production, comprising:

[0021] The conditioning module is used to perform anti-aliasing filtering on the differential voltage signal to obtain the processed analog signal; it performs adaptive gain adjustment according to the dynamic range of the analog signal amplitude to obtain the amplitude-matched conditioning signal, and converts the amplitude-matched conditioning signal into a digital sampling signal through analog-to-digital conversion;

[0022] The extraction module is used to perform real-time dynamic thresholding of the digitally sampled signal based on the amplitude and slope characteristics of the digitally sampled signal to extract the effective waveform data segment; the discrete transient data sequence in the effective waveform data segment is divided into several continuous sub-intervals at equal intervals according to the time sequence; within each sub-interval, adjacent sampling points are connected by straight line segments to form a trapezoid; the area of ​​each trapezoid is calculated and summed to obtain the integral operation result; based on the integral operation result, an integral curve reflecting the energy accumulation characteristics of the entire pressing process is obtained;

[0023] The analysis module is used to construct a reference curve under normal crimping conditions based on the integral curve, and compare the difference between the integral curve measured in the current crimping cycle and the reference curve to obtain the difference feature data sequence.

[0024] The control module is used to determine the current crimping quality status of the terminal in real time based on whether the difference feature data sequence exceeds the preset boundary envelope range. If the crimping quality status is determined to be unqualified, the control command is obtained through the preset PLC intelligent control component to perform closed-loop control on the crimping distance and crimping force of the terminal machine, so as to realize the process adjustment and quality control of the industrial wire harness production site.

[0025] In this embodiment of the invention, by performing anti-aliasing filtering, adaptive gain conditioning, and analog-to-digital conversion on the differential voltage signal during the terminal crimping process, on-site interference is suppressed and signal amplitude matching is ensured, thereby improving the stability of the original sampling data. Based on amplitude and slope characteristics, real-time dynamic threshold truncation is achieved to extract effective waveform segments reflecting the actual crimping process. An energy accumulation integral curve is constructed by accumulating and integrating the trapezoidal surface in sub-intervals. The difference between the measured integral curve and the reference curve is compared, and quality is judged in combination with the preset boundary envelope range. This achieves online real-time identification of terminal crimping quality. At the same time, relying on the PLC intelligent control component, the crimping distance and crimping force are controlled in a closed loop. When an unqualified state is detected, the process parameters can be corrected in time, realizing automated control of terminal crimping quality during wire harness production.

[0026] In a preferred embodiment of the present invention, during the physical crimping of wire harness terminals by the terminal crimping machine, the elastic deformation information of the mold base plate under dynamic crimping force is acquired in real time, and a differential voltage signal characterizing the actual deformation degree of the base plate is extracted based on the elastic deformation information, which may include:

[0027] In this embodiment of the invention, based on the micron-level elastic deformation information generated by the mold base plate under dynamic pressing force, multiple sets of resistive strain-sensitive elements attached to the stress-sensitive area of ​​the base plate are used to convert the elastic deformation information into the resistance change of each sensitive element in real time. Specifically, during the entire process of physical pressing the wire harness terminals by the terminal crimping machine, the crimping mold applies a continuously changing dynamic pressing force to the terminals and wire harness. This pressing force is transmitted to the mold base plate through the mold, causing the mold base plate to produce micron-level elastic deformation corresponding to the magnitude and direction of the pressing force. The degree of deformation increases with the increase of the pressing force and decreases with the decrease of the pressing force, and the deformation is always within the elastic range, without permanent deformation, ensuring that the correspondence between deformation and pressing force is consistent in each pressing process. To capture this micron-level elastic deformation information of the mold base plate, it is first necessary to determine the stress-sensitive area of ​​the mold base plate, which is the area with the most obvious deformation and the most concentrated stress under the pressing force. Through actual testing and verification, the center of the mold base plate is usually selected. The stress-bearing area and the surrounding support area are designated as stress-sensitive areas, as these areas exhibit the largest deformation amplitude and can most accurately reflect changes in the compression force. Multiple sets of resistive strain gauge elements are evenly attached to the defined stress-sensitive areas. The number of elements in each set is determined by the size of the stress-sensitive area, typically 4-8 sets, with each set containing 2-4 resistive strain gauge elements. The elements are attached in the same direction as the possible deformation of the base plate, i.e., along both the direction of the compression force and the perpendicular direction, ensuring comprehensive capture of the base plate's deformation in different directions. During the attachment process, the surface of the mold base plate must first be cleaned and polished to remove oil, rust, and oxide layers, ensuring a smooth and flat surface. Then, a special strain gauge adhesive is applied, and the resistive strain gauge elements are evenly attached to the designated positions. After attachment, the surface is allowed to cure for at least 24 hours to ensure a tight bond between the elements and the base plate, preventing inaccurate deformation transmission due to weak adhesion, which would affect the accuracy of impedance measurement.

[0028] The core working principle of a resistive strain gauge element is the strain resistance effect, meaning its resistance value changes linearly with its mechanical deformation. When the mold base plate undergoes elastic deformation, it causes the resistive strain gauge element attached to it to deform synchronously to the same degree, resulting in a corresponding change in the element's impedance. This achieves the initial conversion of mechanical deformation information into an electrical signal. The calculation of the impedance change of the resistive strain gauge element follows the strain resistance effect formula, as follows: ,in This refers to the change in resistance of a resistive strain gauge element. Its value is equal to the resistance value of the element after deformation minus the initial resistance value before deformation. When the element is stretched, the resistance value increases. It is a positive value; when the component is compressed, the resistance value decreases. It is a negative value; The initial resistance value of the resistive strain sensing element is the resistance value when the element has not undergone any deformation and is in its natural state. Usually, two specifications, 120Ω and 350Ω, are selected according to the actual crimping accuracy requirements. The 120Ω resistive strain sensing element is selected. The sensitivity coefficient of a resistive strain-sensitive element is determined by the material properties and manufacturing process of the element itself, and is usually in the range of 2.0-2.5. An element with a sensitivity coefficient of 2.2 is selected to ensure that it can capture the resistance change corresponding to small deformations. The strain value of the resistive strain sensor is equal to the ratio of the deformation displacement of the sensor to its original length. Because the sensor is tightly fitted to the mold base plate, the strain value of the sensor is completely consistent with the strain value at the corresponding position on the mold base plate. ,in This refers to the deformation displacement of the mold base plate, which is the difference between the deformed length and the original length of the base plate. Its range is usually 1-10μm, and it is positively correlated with the magnitude of the pressing force. This refers to the original length of the mold base plate, in millimeters, i.e., the length of the base plate before deformation. It is determined according to the specifications of the terminal block machine mold, typically 50-100mm. In actual calculations, the deformation displacement of the mold base plate is first measured in real time using a displacement sensor. Combined with the original length of the base plate Calculate the strain value Then, by substituting the values ​​into the strain resistance effect formula, the resistance change of each set of resistive strain-sensitive elements can be calculated in real time. This allows us to obtain the change in impedance of each sensitive element, thus completing the initial conversion of elastic deformation information into electrical signals.

[0029] Based on the change in resistance, a Wheatstone full-bridge measurement circuit is used to convert the difference in resistance between the sensing elements into an initial unbalanced differential voltage signal. The positive and negative components of this signal correspond to the deformation polarities of the tensile and compression zones of the base plate, respectively. Specifically, this includes the change in resistance of each resistive strain sensing element. This weak resistance change signal cannot be directly used for subsequent processing and needs to be converted into a voltage signal by a dedicated measurement circuit. The Wheatstone full-bridge measurement circuit is used to achieve this conversion. The Wheatstone full-bridge measurement circuit consists of four bridge arms, with the resistance values ​​of the four bridge arms initially being consistent. Multiple sets of attached resistive strain sensing elements are connected to the four bridge arms respectively. Specifically, the sensing elements attached to the tensile zone of the mold base plate are connected to two opposite bridge arms, and the sensing elements attached to the compression zone are connected to the other two opposite bridge arms. This ensures that the sensing elements in the tensile and compression zones correspond to different bridge arms. In this way, when the base plate deforms, the resistance of the sensing elements in the tensile zone increases, and the resistance of the sensing elements in the compression zone decreases, which can generate a significant imbalance in the full-bridge circuit, thereby outputting a stable voltage signal. A differential voltage signal is established. After the circuit connection is completed, a stable DC power supply voltage is provided to the Wheatstone full-bridge measurement circuit. The selection of the power supply voltage needs to be combined with the specifications of the sensitive element and the circuit sensitivity requirements. Usually, a DC voltage of 5 volts (V) or 12 volts (V) is selected. A 5-volt DC power supply is selected to ensure stable circuit operation and avoid damage to the sensitive element due to excessive voltage. When the mold base plate is not subjected to pressure and has no elastic deformation, the impedance of each resistive strain sensitive element remains at its initial value, and the resistance values ​​of the four bridge arms are equal. The Wheatstone full-bridge measurement circuit is in a stable state. In equilibrium, the circuit output voltage is zero, meaning there is no signal output. When the mold base plate is subjected to pressure and undergoes elastic deformation, the sensitive element in the stretching zone is stretched, its resistance increases, and the corresponding bridge arm resistance increases. The sensitive element in the compression zone is compressed, its resistance decreases, and the corresponding bridge arm resistance decreases. This causes the Wheatstone full-bridge measurement circuit to lose balance, thereby outputting an initial unbalanced differential voltage signal. The amplitude of this signal is proportional to the difference in resistance of each bridge arm resistive strain sensitive element. The greater the difference in resistance, the greater the amplitude of the output voltage signal.

[0030] The specific calculation of the initial unbalanced differential voltage signal follows the Wheatstone full-bridge output formula, as shown below. ,in This is the initial unbalanced differential voltage signal, measured in millivolts. Its amplitude is typically between 0.1 and 1 mV, and the specific value varies with the magnitude of the clamping force. The greater the clamping force, the greater the signal amplitude. The DC power supply voltage for the Wheatstone full-bridge measurement circuit is 5V in volts to ensure stable circuit operation and meet sensitivity requirements. , , , These represent the resistance changes of the resistive strain-sensitive elements in the four bridge arms, in ohms. , The change in resistance of the sensitive element in the stretching zone is positive; , The change in resistance of the sensitive element in the compression zone is negative; , , , These are the initial resistance values ​​of the resistive strain-sensitive elements in the four bridge arms, in ohms. The initial resistance value of each of the four bridge arms is 120Ω, ensuring that the circuit is in balance in the initial state. This initial unbalanced differential voltage signal has a clear polarity characteristic, which can reflect the deformation state of different areas of the mold base plate. The positive component corresponds to the deformation of the tensile zone of the mold base plate, that is, the voltage component generated when the impedance of the sensitive element in the tensile zone increases. The amplitude of the positive component increases with the increase of the tensile deformation. The negative component corresponds to the deformation of the compression zone of the mold base plate, that is, the voltage component generated when the impedance of the sensitive element in the compression zone decreases. The amplitude of the negative component increases with the increase of the compression deformation. Through this polarity characteristic, the deformation of the tensile zone and the compression zone of the base plate can be intuitively distinguished.

[0031] The initial unbalanced differential voltage signal undergoes common-mode noise suppression and primary differential amplification to obtain an intermediate-stage differential voltage signal with interference filtered out. Specifically, the initial unbalanced differential voltage signal, generated in an industrial wiring harness production environment, is subject to significant electrical and environmental interference, including inverter interference, motor operation interference, grounding loop noise, and electromagnetic radiation interference. This interference superimposes on the initial unbalanced differential voltage signal as common-mode noise, causing signal distortion, manifested as noise and excessive amplitude fluctuations. Without interference filtering, the accuracy of signal processing is severely affected, potentially preventing accurate extraction of valid waveform data segments. Therefore, common-mode noise suppression is first performed on the initial unbalanced differential voltage signal using a dedicated differential amplification circuit. This circuit uses an instrumentation amplifier, which can effectively suppress common-mode noise present at both input terminals simultaneously, amplifying only the differential signal between the two input terminals, i.e., the effective signal related to the deformation of the mold base plate. Common-mode rejection ratio (CMRR) is the core indicator for measuring the noise suppression capability of a differential amplifier circuit. The higher the CMRR value, the better the noise suppression effect. The selected instrumentation amplifier has a CMRR of over 80 dB to ensure that the amplitude of common-mode noise can be suppressed to less than 1% of the effective signal amplitude, minimizing the interference of noise on the effective signal. The specific suppression process is as follows: the initial unbalanced differential voltage signal is connected to the two input terminals of the instrumentation amplifier. The amplifier internally cancels the common-mode signals at the two input terminals through a differential amplification structure, while simultaneously amplifying the differential signal to achieve effective filtering of common-mode noise.

[0032] After common-mode noise suppression, the initial unbalanced differential voltage signal has a small amplitude, typically only 0.1-1mV. Subsequent analog-to-digital conversion (ADC) and waveform extraction require a signal with a certain amplitude. If the signal amplitude is too small, it will lead to excessive quantization errors during ADC, affecting the digitization accuracy. Therefore, primary differential amplification is needed to increase the signal amplitude to a suitable range after noise filtering. The amplification factor of the primary differential amplification is determined based on the requirements of subsequent processing. The amplification factor is calculated according to the principle: amplified signal amplitude = initial unbalanced differential voltage signal amplitude × primary differential amplification factor. The primary differential amplification factor is set to 50 times, increasing the initial 0.1-1mV amplitude... The signal is amplified to 5-50mV. This amplitude range meets the requirements of subsequent analog-to-digital conversion without causing signal distortion due to excessive amplification. During amplification, the amplitude change of the signal needs to be monitored in real time to ensure that the amplified signal does not exhibit saturation distortion, i.e., the signal amplitude does not exceed the output range of the instrumentation amplifier. If the signal amplitude is detected to be close to saturation, the primary differential amplification factor can be adjusted appropriately to ensure that the amplified signal retains complete waveform characteristics. After common-mode noise suppression and primary differential amplification, the intermediate stage differential voltage signal has filtered out most of the electrical interference in the industrial environment, and the signal amplitude is stable at 5-50mV, accurately preserving the effective signal characteristics corresponding to the deformation of the mold base plate.

[0033] The intermediate-stage differential voltage signal undergoes high-frequency mechanical vibration noise attenuation processing to obtain the final differential voltage signal that characterizes the actual dynamic deformation of the mold base plate in real time. Specifically, the intermediate-stage differential voltage signal, although having filtered out most electrical interference, still contains high-frequency mechanical vibration noise generated during the terminal crimping machine's operation. This noise mainly originates from the collision of the terminal crimping machine mold, the high-speed operation of the motor, and the friction of the transmission mechanism, with a frequency range typically between 1000-10000 Hz. However, the effective signal frequency range corresponding to the elastic deformation of the mold base plate is typically 10-100 Hz. The frequency of the high-frequency mechanical vibration noise is much higher than the effective signal frequency and will be superimposed on the noise as clutter. The purity of the effective signal is affected, which in turn affects the accuracy of effective waveform extraction and energy accumulation analysis. Therefore, it is necessary to attenuate the high-frequency mechanical vibration noise of the intermediate stage differential voltage signal. The attenuation of high-frequency mechanical vibration noise is achieved by using a low-pass filter. The core function of the low-pass filter is to allow signals below the cutoff frequency to pass through while blocking signals above the cutoff frequency, thereby achieving high-frequency noise attenuation. An RC low-pass filter is selected because it has a simple structure, stable operation, low cost, and can meet the requirements of high-frequency noise attenuation. It consists of a resistor and a capacitor connected in series. By reasonably selecting the parameters of the resistor and capacitor, a suitable cutoff frequency can be set.

[0034] Setting the cutoff frequency is the core of a low-pass filter. It needs to be determined by considering both the effective signal frequency and the high-frequency noise frequency. The cutoff frequency should be higher than the highest frequency of the effective signal and lower than the lowest frequency of the high-frequency noise. This allows the effective signal to pass completely while effectively blocking high-frequency noise. The highest frequency of the effective signal is 100Hz, and the lowest frequency of the high-frequency noise is 1000Hz. Therefore, the cutoff frequency of the low-pass filter is set to 300Hz. This cutoff frequency completely covers the frequency range of the effective signal while effectively blocking high-frequency mechanical vibration noise above 300Hz. The formula for calculating the cutoff frequency of a low-pass filter is as follows: ,in This is the cutoff frequency of the low-pass filter, measured in Hertz, with a value of 300 Hz. This is the resistance value of the low-pass filter, in kiloohms, with a value of 10kΩ. This is the capacitance value of the low-pass filter, in microfarads, calculated based on the cutoff frequency and resistance value. , Substituting into the formula, we can calculate... Therefore, we choose The capacitor; the specific attenuation process is as follows: the intermediate stage differential voltage signal is connected to an RC low-pass filter. When the signal passes through the filter, the effective signal below 300Hz can pass smoothly, while the high-frequency mechanical vibration noise above 300Hz will be blocked and attenuated by the filter. The amplitude of the attenuated high-frequency noise is reduced to less than 3% of the effective signal amplitude, ensuring that the noise will not affect the subsequent signal processing. During the attenuation process, the waveform of the signal needs to be monitored in real time to ensure that the waveform of the effective signal will not be distorted due to the attenuation process. If waveform distortion is found, the resistance and capacitance parameters of the filter need to be adjusted and the cutoff frequency setting optimized. After the high-frequency mechanical vibration noise attenuation processing, the obtained signal only contains the effective components related to the actual dynamic deformation of the mold base plate. This signal can accurately characterize the degree of dynamic deformation of the mold base plate during the pressing process in real time, and the signal amplitude is stable at 5-50mV, the waveform is smooth, and there is no obvious noise. This signal is the final differential voltage signal.

[0035] By capturing the micron-level elastic deformation of the mold base plate through multiple sets of resistive strain sensitive elements, and combining it with the Wheatstone full-bridge measurement circuit to convert the deformation information into an electrical signal, the stability of wire harness terminal crimping production is improved through multi-stage processing of common-mode noise suppression, primary amplification and high-frequency mechanical vibration noise attenuation.

[0036] In a preferred embodiment of the present invention, the differential voltage signal is subjected to anti-aliasing filtering to obtain a processed analog signal; adaptive gain adjustment is performed according to the amplitude dynamic range of the analog signal to obtain an amplitude-matched conditioning signal; and the amplitude-matched conditioning signal is converted into a digital sampling signal through analog-to-digital conversion, which may include:

[0037] In this embodiment of the invention, the final differential voltage signal, which characterizes the actual dynamic deformation of the mold base plate in real time, is input to the pre-stage anti-aliasing filter network. By setting the cutoff frequency of the low-pass filter, high-frequency noise components higher than the Nyquist frequency are suppressed to obtain a smooth and continuous analog signal. Specifically, although the final differential voltage signal has been attenuated by high-frequency mechanical vibration noise, it may still contain a small amount of residual noise with higher frequencies. Furthermore, if high-frequency components higher than the Nyquist frequency exist during subsequent analog-to-digital conversion, aliasing will occur, leading to distortion of the digitized signal and failing to accurately reflect the dynamic deformation of the mold base plate. This can affect the accuracy of effective waveform extraction and energy accumulation analysis. Therefore, it is necessary to perform pre-stage anti-aliasing filtering on the final differential voltage signal to suppress high-frequency noise components above the Nyquist frequency and obtain a smooth and continuous analog signal. The pre-stage anti-aliasing filter network adopts a low-pass filter structure and selects an active low-pass filter. At the same time, it avoids attenuation of the effective signal. The core of the filter network is to set an appropriate low-pass filter cutoff frequency. The setting of this cutoff frequency must strictly follow the Nyquist sampling theorem to ensure that the cutoff frequency does not exceed the Nyquist frequency, thereby completely suppressing high-frequency noise above the Nyquist frequency and avoiding the occurrence of aliasing.

[0038] The method for calculating the Nyquist frequency is as follows: The Nyquist frequency is half the analog-to-digital conversion sampling frequency. ,in Nyquist frequency, measured in Hertz, is the highest signal frequency that can avoid aliasing; signals above this frequency will produce aliasing distortion. The preset sampling frequency, measured in Hertz, is used for subsequent analog-to-digital conversion. Considering the signal characteristics of the crimping process (effective signal frequency 10-100Hz), the preset sampling frequency needs to be significantly higher than the highest frequency of the effective signal to ensure complete capture of dynamic signal changes. A preset sampling frequency of 1000Hz is chosen to meet signal capture requirements while avoiding increased energy consumption and data redundancy caused by excessively high sampling frequencies. Substituting 1000Hz into the formula, we can calculate the Nyquist frequency. =500Hz. To ensure complete suppression of high-frequency noise above the Nyquist frequency, the cutoff frequency of the low-pass filter needs to be set to 0.5-0.8 times the Nyquist frequency. Choosing 0.6 times, the calculated cutoff frequency is 500Hz × 0.6 = 300Hz. This cutoff frequency effectively blocks high-frequency noise above 500Hz while completely covering the effective signal frequency range (10-100Hz), ensuring the effective signal can pass smoothly without attenuation. Specifically, the final differential voltage signal, which characterizes the actual dynamic deformation of the mold base plate in real time, is connected to the input of the pre-stage anti-aliasing filter network through a shielded wire. Shielded wires can effectively reduce the impact of external electromagnetic interference on signals, ensuring that signals are not contaminated during transmission. After the signal enters the filtering network, it is processed by an active low-pass filter, and high-frequency residual noise above 300Hz is significantly attenuated by more than 80 dB, reducing the amplitude of high-frequency noise to less than 1% of the effective signal amplitude, thus avoiding aliasing during analog-to-digital conversion. After anti-aliasing filtering, the output signal is free of high-frequency residual noise, and the waveform becomes smooth and continuous without obvious noise and fluctuations. It can retain the effective characteristics related to the dynamic deformation of the mold base plate in the final differential voltage signal, which is the processed analog signal.

[0039] Based on the instantaneous amplitude distribution range of the analog signal, the matching coefficient between the current signal peak value and the full-scale input range of the analog-to-digital converter is dynamically calculated. Based on this matching coefficient, the gain factor of the programmable gain amplifier unit is automatically adjusted so that the amplitude of the output signal is mapped to the final quantization range of the analog-to-digital converter, resulting in an amplitude-matched conditioned signal. Specifically, the dynamic range of the analog signal amplitude fluctuates with changes in the crimping conditions. This is mainly due to slight tolerance fluctuations in the wire diameter of different batches of wire harnesses and batch-to-batch differences in the hardness of the terminal raw materials. These factors lead to different degrees of elastic deformation of the mold base plate, thus causing changes in the amplitude of the analog signal. The amplitude fluctuation range is typically 5-50mV, which is then used for subsequent digital conversion. The analog-to-digital converter (ADC) has a fixed full-scale input range. If the amplitude of the analog signal is too large, it will cause signal saturation distortion during ADC conversion; if the amplitude is too small, it will cause excessive quantization error in the digitized signal, affecting the digitization accuracy. Therefore, adaptive gain adjustment is needed based on the dynamic range of the analog signal amplitude to map the output signal amplitude to the final quantization range of the ADC, obtaining an amplitude-matched conditioned signal. The analog signal is monitored in real time, and signal data is collected over a certain period, typically one compression cycle (approximately 0.5-1 second). The instantaneous amplitude distribution range of the analog signal within this time period is extracted using a signal amplitude monitoring circuit to determine the current signal peak value, i.e., the maximum amplitude of the signal, denoted as . The matching coefficient between the current signal peak value and the full-scale input range of the analog-to-digital converter (ADC) is dynamically calculated. The matching coefficient measures the degree of fit between the current signal amplitude and the ADC input range. The formula for calculating the matching coefficient is as follows: ,in The matching coefficient is a dimensionless coefficient. Its value reflects the approximate range of the gain that needs to be adjusted. When the matching coefficient is greater than 1, the gain needs to be increased; when the matching coefficient is less than 1, the gain needs to be decreased. The full-scale input range of the analog-to-digital converter is measured in millivolts. A high-resolution analog-to-digital converter is selected, and its full-scale input range is set to 0-100mV. This range can cover the amplitude range of the analog signal after gain adjustment, ensuring that the signal will not saturate. This represents the peak value of the current analog signal, expressed in millivolts (mV). It is the maximum amplitude of the analog signal during the monitoring period, ranging from 5 to 50 mV. Multiplying by 1.2 in the formula is to reserve a certain amplitude margin, to avoid the amplitude exceeding the full-scale input range of the analog-to-digital converter due to instantaneous signal fluctuations, to ensure that the signal will not experience saturation distortion after adjustment, and to improve the stability of the signal.

[0040] After calculating the matching coefficient, this coefficient is input into the programmable gain amplifier (PGA). The PGA automatically adjusts its gain based on the matching coefficient. The gain adjustment follows the principle that the PGA's gain equals the matching coefficient, ensuring that the adjusted signal amplitude maps to the final quantization range of the analog-to-digital converter (ADC). Specifically, if the peak value of the current analog signal is 5mV, substituting into the formula, the matching coefficient is calculated as 100mV ÷ (5mV × 1.2) ≈ 16.7. The PGA automatically adjusts the gain to 17, resulting in a signal amplitude of 5mV × 17 = 85mV, which falls within the 0-100mV full-scale input range of the ADC. If the peak value of the current analog signal is 50mV, the PGA calculates... The matching coefficient can be calculated as 100mV ÷ (50mV × 1.2) ≈ 1.7. At this time, the programmable gain amplifier unit automatically adjusts the gain to 2 times. The adjusted signal amplitude is 50mV × 2 = 100mV, which just reaches the full-scale input range of the analog-to-digital converter and will not cause saturation distortion. During the gain adjustment process, the adjusted signal amplitude is monitored in real time to ensure that the signal amplitude is always within the final quantization range of the analog-to-digital converter (0-100mV). If the signal amplitude is detected to be outside this range, the gain is finely adjusted in time to ensure signal amplitude matching. After adaptive gain adjustment, the output signal amplitude is stable within the final quantization range of the analog-to-digital converter and can completely retain the effective characteristics of the analog signal. This signal is the amplitude-matched conditioning signal.

[0041] The amplitude-matched conditioned signal is input to a high-resolution analog-to-digital converter (ADC) device, which converts it into a discrete digital sequence, i.e., a digital sampled signal, at a preset sampling frequency and quantization bit depth. Specifically, the amplitude-matched conditioned signal is still an analog signal, but subsequent processing such as effective waveform extraction, integration, and quality judgment all require digital signals as a foundation. Therefore, the amplitude-matched conditioned signal needs to be converted into a discrete digital sequence, i.e., a digital sampled signal. This conversion process is achieved through a high-resolution ADC device. First, a suitable high-resolution ADC device is selected. Considering the accuracy requirements of the terminal machine monitoring and management system, a 16-bit quantization bit ADC device is chosen. The operating voltage of the device is set to 5V DC, consistent with the power supply voltage of the preamplifier circuit, to ensure stable operation. The amplitude-matched conditioning signal is connected to the input of the high-resolution analog-to-digital converter (ADC) via a dedicated signal line. Shielded wiring is used during connection to prevent external interference from affecting conversion accuracy. The ADC discretizes the conditioning signal according to a preset sampling frequency and quantization bit depth. The preset sampling frequency is consistent with the sampling frequency used to calculate the Nyquist frequency, i.e., 1000Hz, and the preset quantization bit depth is 16 bits. A sampling frequency of 1000Hz means that the ADC samples the conditioning signal 1000 times per second. Each sample acquires a digital value corresponding to an instantaneous amplitude. The sampling interval is 1 ÷ 1000 = 0.001 seconds. This sampling interval can capture the dynamic changes of the conditioned signal, ensuring that key signal features are not lost. The quantization bit is 16 bits, meaning that the analog-to-digital converter can divide the full-scale input range of 0-100mV into 2^16 quantization levels, or 65536 quantization levels. The voltage value corresponding to each quantization level is approximately 100mV ÷ 65536 ≈ 0.0015mV. The quantization accuracy is extremely high, effectively reducing quantization errors and ensuring that the digitized signal can accurately reflect the amplitude changes of the analog signal. The specific conversion process is as follows: the analog-to-digital converter... The internal sample-and-hold circuit instantaneously samples the conditioning signal according to a preset sampling frequency, temporarily storing the sampled analog amplitude signal. Then, the quantization circuit converts the analog amplitude signal into a corresponding binary digital quantity, and the encoding circuit converts the binary digital quantity into a decimal digital quantity, completing one analog-to-digital conversion. Following this process, sampling and conversion are completed every 0.001 seconds. By continuously acquiring the signal for one pressing cycle, a set of discrete digital quantity sequences arranged in time order can be obtained. This discrete digital quantity sequence can correspond to the instantaneous amplitude change of the conditioning signal and completely retain the effective characteristics of the dynamic deformation of the mold base plate, which is the digital sampling signal.

[0042] High-frequency noise is completely suppressed and aliasing is avoided by anti-aliasing filtering. The signal amplitude fluctuation under different working conditions is adapted by adaptive gain adjustment. The signal is digitized by high-resolution analog-to-digital conversion, ensuring that the output digital sampling signal is true and complete, and improving the wire harness production qualification rate.

[0043] In a preferred embodiment of the present invention, based on the amplitude and slope characteristics of the digitally sampled signal, real-time dynamic thresholding is performed on the digitally sampled signal to extract effective waveform data segments; the discrete transient data sequence in the effective waveform data segments is divided into several continuous sub-intervals at equal intervals according to time sequence; within each sub-interval, adjacent sampling points are connected by straight line segments to form trapezoids; the area of ​​each trapezoid is calculated and summed to obtain the integral operation result; based on the integral operation result, an integral curve reflecting the energy accumulation characteristics of the entire pressing process is obtained, which may include:

[0044] In this embodiment of the invention, the digital sampling signal is used as the input sequence. A sliding time window is used to traverse the time axis segment by segment. Within each window, the mean amplitude, local maxima, and local minima of the sampling points are calculated. First-order difference operations are performed on adjacent sampling points to extract the instantaneous slope values ​​of each point, resulting in a joint feature set that integrates amplitude statistical features and slope change features. Specifically, the digital sampling signal is a discrete digital sequence arranged in chronological order. This sequence includes not only the valid signals corresponding to the deformation of the mold base plate during the pressing process, but also invalid silent data before and after pressing. The amplitude and trend of the signal fluctuate with the crimping process. To distinguish between valid and invalid data, and to capture the dynamic characteristics of valid signals, it is necessary to first extract the amplitude statistical characteristics and slope change characteristics of the digitally sampled signal to form a joint feature set. In specific implementation, the digitally sampled signal is used as the input sequence, and a sliding time window is used to traverse the entire signal sequence segment by segment along the time axis. The setting of the sliding time window needs to be combined with the signal characteristics of the crimping process. The window size is determined according to the sampling frequency and crimping period. The sampling frequency is 1000Hz, and the crimping period is about 0.5-1 second. Therefore, the sliding time window... The aperture width is set to 0.01 seconds, corresponding to 10 sampling points. This ensures comprehensive capture of signal features within the window while avoiding feature lag due to an excessively large window or feature distortion due to an excessively small window. The sliding step size is set to one sampling point, meaning the window moves forward one sampling point for each iteration, ensuring point-by-point coverage of the entire digital sampling signal sequence without missing any signal details. Within each sliding time window, feature calculations are performed on the 10 sampling points within the window. First, the amplitude mean is calculated by summing the amplitudes of all sampling points within the window and then dividing by the number of sampling points (10) to obtain the mean amplitude. The average amplitude within the window reflects the overall amplitude level of the signal within the current window. Local maxima and minima are extracted by traversing all sampling points within the window, selecting the sampling point with the largest amplitude as the local maxima and the sampling point with the smallest amplitude as the local minima, reflecting the amplitude fluctuation range of the signal within the current window. First-order difference operations are performed on adjacent sampling points within the window to extract the instantaneous slope value of each sampling point. The instantaneous slope value reflects the rate of change of the signal; a positive slope indicates an upward trend, a negative slope indicates a downward trend, and the larger the absolute value of the slope, the faster the signal changes.

[0045] The instantaneous slope value is calculated using the first-order difference formula, as follows: ,in For the first The instantaneous slope value of each sampling point, in millivolts per millisecond, has no positive or negative sign and only indicates the rate of change; a positive or negative sign indicates the direction of change. For the first The amplitude of the nth sampling point, in millivolts, is the amplitude of the nth sampling point in the digitized sampled signal. The actual voltage amplitude corresponding to each digital quantity; For the first The amplitude at the sampling point is expressed in millivolts, which is the same as the amplitude at the sampling point. The amplitude of the previous sampling point adjacent to each sampling point; For the first The sampling time of the nth sampling point is in milliseconds, calculated based on the sampling frequency, i.e., the nth sampling point. The time of each sampling point = ( ) × sampling interval, where the sampling interval is 1ms; For the first The sampling time of the nth sampling point is in milliseconds, i.e., the nth sampling point. The time of each sampling point = ( ) × sampling interval; Since the sampling frequency is fixed at 1000Hz and the sampling interval is fixed at 1ms, therefore - =1ms, at this time the formula for calculating the instantaneous slope value can be simplified to: = - The calculation is more convenient and does not affect the trend and relative size of the slope. For each sliding time window, the above calculation process is repeated to obtain the mean amplitude, local maximum, local minimum and instantaneous slope value of each sampling point corresponding to each window. These feature parameters of all windows are summarized to form a joint feature set that integrates amplitude statistical features and slope change features.

[0046] The joint feature set is input into the dynamic threshold generator. A base threshold baseline is determined based on the average amplitude within the current window. The threshold offset is adjusted in real-time using the polarity and absolute value of the instantaneous slope value, dynamically generating an upper threshold curve and a lower threshold curve. Specifically, the joint feature set is input into the dynamic threshold generator. The generator first extracts the average amplitude within each sliding time window, using this average amplitude as the base threshold baseline. The base threshold baseline updates in real-time as the window moves, adapting to dynamic fluctuations in signal amplitude and ensuring that the threshold always matches the overall amplitude level of the current signal. For example, when the signal amplitude decreases due to a thinner wire diameter, the average amplitude within the window decreases accordingly, and the base threshold baseline decreases synchronously; when the signal amplitude increases due to a harder terminal material, the average amplitude decreases... As the mean value increases, the baseline threshold also increases synchronously. After determining the baseline threshold, the threshold offset is adjusted in real time using the instantaneous slope value from the joint feature set. The offset adjustment is based on the polarity and absolute value of the instantaneous slope value. When the instantaneous slope value is positive and the absolute value is large, it indicates that the signal is in a rapid rising phase. At this time, the offset of the upper threshold needs to be increased and the offset of the lower threshold needs to be decreased to avoid missing the initial trigger point due to the signal rising too fast. When the instantaneous slope value is negative and the absolute value is large, it indicates that the signal is in a rapid falling phase. At this time, the offset of the upper threshold needs to be decreased and the offset of the lower threshold needs to be increased to avoid misjudging the termination trigger point due to the signal falling too fast. When the instantaneous slope value is close to 0, it indicates that the signal is in a stable phase. At this time, the offset remains unchanged to ensure the stability of the threshold.

[0047] The specific adjustment rules for the offset are as follows: upper threshold offset = base threshold baseline × 0.1 × absolute value of instantaneous slope; lower threshold offset = base threshold baseline × 0.1 × absolute value of instantaneous slope. When the instantaneous slope is positive, the upper threshold = base threshold baseline + upper threshold offset, and the lower threshold = base threshold baseline - lower threshold offset. When the instantaneous slope is negative, the upper threshold = base threshold baseline - upper threshold offset, and the lower threshold = base threshold baseline + lower threshold offset. This adjustment method allows the upper and lower threshold curves to follow the signal's changing trend in real time, adapting to the signal's amplitude fluctuations and rate of change. According to the above rules, the offset of the base threshold baseline for each sliding time window is adjusted to obtain the upper and lower thresholds corresponding to each window. The upper thresholds of all windows are connected in chronological order to form a continuous upper threshold curve. The lower thresholds of all windows are connected in chronological order to form a continuous lower threshold curve. The two threshold curves change dynamically over time, always adapting to the characteristics of the digitally sampled signal.

[0048] The original digitized sampled signal is compared point-by-point with dynamically generated upper and lower threshold curves. When the signal amplitude crosses the upper threshold from below, the current sampling moment is marked as the start trigger point of the valid waveform. When the signal amplitude crosses the lower threshold from above and subsequent consecutive sampling points remain below the lower threshold, the current sampling moment is marked as the end trigger point of the valid waveform, thus obtaining the start and end trigger points. Specifically, this involves comparing the amplitude of the digitized sampled signal at each sampling moment with the corresponding upper and lower thresholds along the time axis, focusing on the instant the signal amplitude crosses the threshold curve. The direction of the crossing is used to determine whether it is a trigger point. For the start trigger point, when the amplitude of the digitized sampled signal crosses the upper threshold curve from below, it indicates that the signal has entered the valid stage, i.e., the compression begins. At this time, the current sampling moment is marked as the start trigger point of the valid waveform, avoiding misjudging invalid silent data before compression as a valid signal. For the end trigger point, the judgment condition is more stringent to avoid misjudgment due to instantaneous signal fluctuations. When the amplitude of the digitally sampled signal crosses the lower threshold curve from above, and the amplitude of the subsequent three consecutive sampling points remains below the lower threshold curve, it indicates that the signal has entered the invalid stage, i.e., the crimping ends. At this point, the current sampling time is marked as the termination trigger point of the valid waveform. Setting the condition of three consecutive sampling points is to eliminate the brief drop in signal caused by instantaneous noise, ensure the accuracy of the termination trigger point determination, and avoid premature termination of the valid waveform, which would lead to the loss of valid data. During the entire point-by-point comparison process, the sampling time, signal amplitude, and threshold value are recorded in real time. If the signal amplitude crosses the threshold curve multiple times, only the moment when it first crosses the upper threshold from below is taken as the starting trigger point, and the moment when the termination condition is met is taken as the termination trigger point. This ensures that only one set of starting and ending trigger points is marked for each crimping cycle, avoiding confusion of the valid waveform caused by multiple sets of trigger points. In this way, the starting and ending trigger points of the valid waveform are marked, and the time range of the valid signal is clearly defined.

[0049] Using the start and end trigger points as the truncation boundaries, all sampling points between the two boundaries are extracted from the original digital sampling signal sequence. All invalid silent data segments before the start point and after the end point are automatically removed, thus completing the extraction of valid waveform data segments. Specifically, this includes: first, determining the sampling point numbers corresponding to the start and end trigger points. Assuming the start trigger point corresponds to the m-th sampling point and the end trigger point corresponds to the k-th sampling point (m < k), all sampling points between the m-th and k-th sampling points are extracted from the original digital sampling signal sequence. The signals corresponding to these sampling points are the valid signals of the mold base plate deformation during the pressing process, constituting the valid waveform data segments. Simultaneously, all sampling points before the start trigger point (from the 1st to the (m-1st)th sampling point) and after the end trigger point (from the (k+1st)th to the last sampling point) are automatically removed. All invalid silent data segments, mainly signals generated by the terminal block under no-load operation before crimping and the equipment reset after crimping, do not contain any valid information related to crimping quality. Removing these segments reduces the workload of subsequent data processing and avoids interference from invalid data in integration calculations and quality judgment. After extraction, the valid waveform data segments are initially verified to check their integrity. If the number of sampling points in the valid waveform data segment is between 500 and 1000, corresponding to a crimping period of 0.5 to 1 second, and the signal amplitude shows a typical crimping characteristic of rising-stable-falling, then the extraction is valid. If the number of sampling points is too few or too many, or the signal amplitude does not have obvious crimping characteristics, then the starting and ending trigger point markings are rechecked, the threshold curve offset is adjusted, and the segment is re-trunculated until a complete and valid waveform data segment is extracted.

[0050] Based on the effective waveform data segment, discrete sampling points are arranged in ascending order of sampling time. Using a fixed time interval as the sub-interval width, the entire effective waveform data segment is sequentially divided into multiple consecutive, non-overlapping sub-intervals, starting from the starting point. Each sub-interval contains two adjacent sampling points, resulting in an ordered set of sub-intervals. Specifically, this involves: first, arranging all discrete sampling points in the effective waveform data segment in ascending order of sampling time to ensure the time order of the sampling points is consistent with the compression process, avoiding errors in subsequent integration calculations due to disordered time order; using a fixed time interval as the sub-interval width, consistent with the sampling interval of the digital sampling signal (0.001 seconds), ensuring that each sub-interval contains only two adjacent sampling points, which can capture instantaneous changes in the signal and simplify the calculation of the trapezoidal area. The division process starts from the starting point of the effective waveform data segment (the m-th sampling point). Starting from the first sub-interval, the entire valid waveform data segment is sequentially divided into multiple continuous sub-intervals that are connected end-to-end and do not overlap. The first sub-interval contains the m-th and (m+1)-th sampling points, the second sub-interval contains the (m+1)-th and (m+2)-th sampling points, and so on, with the last sub-interval containing the (k-1)-th and (k)-th sampling points (k being the sampling point number corresponding to the termination trigger point). The start time of each sub-interval is the time of the previous sampling point, and the end time is the time of the next sampling point. The width of each sub-interval is fixed at 1 ms. All sub-intervals are connected end-to-end and do not overlap, completely covering the entire valid waveform data segment and forming an ordered set of sub-intervals. For example, if the valid waveform data segment contains 600 sampling points and the corresponding compression period is 0.6 seconds, it can be divided into 599 continuous sub-intervals, each with a width of 1 ms and containing two adjacent sampling points. After being arranged in an orderly manner, they form a set of sub-intervals.

[0051] Based on an ordered set of sub-intervals, each sub-interval is extracted sequentially. For the current sub-interval, the area of ​​the trapezoid formed by the lines connecting adjacent sampling points within the sub-interval and the time axis is calculated, with the left sampling point amplitude as the left height of the trapezoid, the right sampling point amplitude as the right height of the trapezoid, and a fixed time interval as the width of the trapezoid. This yields the local area value representing the instantaneous energy contribution of the sub-interval, and a sequence of local area values ​​is obtained according to the time order of the sub-intervals. Specifically, this involves: first, starting the sub-interval traversal program, extracting each sub-interval sequentially according to the time order of the ordered set of sub-intervals for individual calculation, ensuring that each sub-interval is processed without omissions. To prevent omissions and duplicate calculations, each sub-interval is uniquely identified during the traversal process, marked as the i-th sub-interval, starting from 1 and incrementing sequentially until the last sub-interval. This facilitates the recording, verification, and association of calculation results. For the currently extracted i-th sub-interval, the three key parameters constituting the trapezoid must be clearly defined. The determination of each parameter must be combined with the characteristics of the digital sampling signal and the sub-interval division rules described above to ensure the accuracy and rationality of the parameters. The specific parameter determination process is as follows: The first key parameter is the left height of the trapezoid, which is determined by the amplitude of the left sampling point within the sub-interval, denoted as . The left sampling point is the earliest sampling point in this sub-interval, corresponding to the (m+i-1)th sampling point in the effective waveform data segment, where m is the sampling point number corresponding to the starting trigger point of the effective waveform data segment. The amplitude of this sampling point is the digital amplitude obtained after anti-aliasing filtering, adaptive gain adjustment and analog-to-digital conversion mentioned above. Its value directly corresponds to the degree of deformation of the mold base plate at that moment, and thus corresponds to the magnitude of the pressing force. The larger the amplitude, the greater the pressing force at that moment and the stronger the instantaneous energy contribution; the smaller the amplitude, the smaller the pressing force at that moment and the weaker the instantaneous energy contribution. Before determining the amplitude of the left sampling point, the validity of the amplitude of the sampling point needs to be verified to check for any abnormal values, such as an amplitude of 0 or an amplitude exceeding the normal range of 5-100mV. If there are abnormal values, the average amplitude of two adjacent sampling points is used to replace them to ensure the accuracy of the left height parameter and avoid abnormal data affecting the trapezoidal area calculation results.

[0052] The second key parameter is the right height of the trapezoid, which is explicitly defined as the amplitude of the right sampling point within the sub-interval, denoted as... The right sampling point is the later sampling point in this sub-interval, corresponding to the (m+i)th sampling point in the effective waveform data segment. It is adjacent to the left sampling point, and its amplitude determination method and validity verification standard are completely consistent with those of the left sampling point, ensuring the consistency and reliability of the amplitude data of the two sampling points. The amplitude of the right sampling point also corresponds to the signal strength at that moment, reflecting the magnitude of the intensification force at that moment. The amplitude difference between the left and right sampling points can reflect the trend of intensification force change in this sub-interval. If the amplitude of the right sampling point is greater than that of the left sampling point, it indicates that the intensification force is increasing in this time period, and the instantaneous energy contribution is gradually increasing; if the amplitude of the right sampling point is less than that of the left sampling point, it indicates that the intensification force is decreasing in this time period, and the instantaneous energy contribution is gradually weakening; if the amplitudes of the two are basically the same, it indicates that the intensification force tends to be stable in this time period, and the instantaneous energy contribution remains stable.

[0053] The third key parameter is the width of the trapezoid, which is explicitly defined as the width of the trapezoid at a fixed time interval of 1ms, denoted as . This fixed time interval is completely consistent with the sampling interval of the digital sampling signal mentioned earlier. The preset sampling frequency in the previous text is 1000Hz, and the sampling interval = 1 ÷ sampling frequency = 1 ÷ 1000 = 0.001 seconds = 1ms. Setting the sub-interval width to 1ms ensures that each sub-interval contains only two adjacent sampling points, simplifying the calculation process of the trapezoidal area, and also ensures the uniformity of the time dimension, making the time length of each sub-interval consistent. This ensures that the trapezoidal areas of different sub-intervals are comparable and avoids the quantization deviation of instantaneous energy contribution caused by inconsistent time widths. At the same time, this width is fixed and does not change with the fluctuation of the signal amplitude, further improving the stability and accuracy of the calculation results.

[0054] Once the three key parameters are determined, the area of ​​the trapezoid in the i-th subinterval is precisely calculated using the general formula for calculating the area of ​​a trapezoid. The formula is as follows: ,in Let be the trapezoidal area of ​​the i-th sub-interval, expressed in millivolt-milliseconds. Its core function is to characterize the instantaneous energy contribution of the crimping process within that sub-interval. Its value directly reflects the amount of crimping energy within that 1ms time interval, serving as the core foundational data for subsequent cumulative summation and generation of the integral curve. Its value range is determined based on the amplitude range of the left and right sampling points. Since... and The value range is 5-100mV. =1ms, therefore The value range is from (5+5)×1÷2=5mV·ms to (100+100)×1÷2=100mV·ms, with different sub-intervals. The numerical differences directly reflect the instantaneous energy changes at different times during the crimping process; The amplitude of the left sampling point of the i-th sub-interval is in millivolts. It is the voltage amplitude corresponding to the earlier sampling point in the sub-interval. It is derived from the obtained digital sampling signal and determined after validity verification. The value range is 5-100mV. Its value is positively correlated with the deformation degree of the mold base plate, and thus positively correlated with the pressing force. Let be the amplitude of the right sampling point in the i-th sub-interval, in millivolts, that is, the voltage amplitude corresponding to the later sampling point in this sub-interval, and . Originating from the same digital sampling signal and having undergone validity verification, the value ranges from 5 to 100mV, and its numerical changes reflect the changes in the pressure relay force within this sub-interval. The width of the trapezoid, i.e., the fixed time interval of the sub-interval, is measured in milliseconds and is fixed at 1ms. It is consistent with the sampling interval of the digital sampling signal to ensure the uniformity of the time dimension and the comparability of the calculation results. Its value does not change with the pressing conditions or the signal amplitude.

[0055] To illustrate the calculation process more intuitively, let's take a concrete example. Suppose we are currently processing the 5th sub-interval, and the left sampling point is the (m+4)th sampling point of the valid waveform data segment, with its amplitude... =20mV, the right sampling point is the (m+5)th sampling point, its amplitude =25mV, trapezoidal width =1ms, substituting these parameters into the formula, the area of ​​the trapezoid in this subinterval can be calculated. = (20+25)×1÷2=22.5mV·ms. This value indicates that the instantaneous energy contribution is 22.5mV·ms within the 1ms pressing time period. Since the amplitude of the right sampling point is greater than that of the left sampling point, it indicates that the pressing force is increasing during this time period, and the instantaneous energy contribution is gradually increasing. According to the above calculation method and example, the trapezoidal area is calculated for each sub-interval in the ordered sub-interval set in sequence. After the calculation of each sub-interval is completed, the obtained trapezoidal area value, i.e., the local area value, is recorded in the data register in real time to ensure that the calculation results are not lost or confused. The size of the local area value is positively correlated with the instantaneous energy contribution during this time period. The larger the local area value, the greater the pressing force during this time period. Under the condition of a fixed duration, the instantaneous energy contribution is greater. The smaller the product value, the smaller the crimping force and the smaller the instantaneous energy contribution during that time period. After the trapezoidal area of ​​all sub-intervals is calculated, all the recorded local area values ​​are organized and arranged in strict accordance with the time order of the sub-intervals, from the first sub-interval to the last sub-interval, to form a complete local area value sequence. The length of this sequence is exactly the same as the number of sub-intervals in the ordered sub-interval set. Each local area value corresponds one-to-one with the corresponding sub-interval, and then one-to-one with each 1ms time period of the crimping process. Through this local area value sequence, the instantaneous energy changes at different time periods during the crimping process can be reflected, clearly showing the differences in energy contribution in the initial stage (metal wire begins to deform), the middle stage (insulation layer compression, metal wire shaping), and the later stage (terminal curling).

[0056] Based on the sequence of local area values, starting from the first local area value, the current local area value is sequentially summed with the cumulative sum of all previous local area values. Each calculation yields an intermediate integral value that increases with the pressing process. This intermediate integral value is then associated with the end time of the current sub-interval, transforming the local area value sequence into a sequence of integral operation results that increases with time. Specifically, starting from the first local area value in the sequence, a cumulative summation operation is performed sequentially. Each cumulative summation yields an intermediate integral value that increases with the pressing process, and this intermediate integral value is associated with the end time of the current sub-interval, ensuring a one-to-one correspondence between the intermediate integral value and the pressing time. The specific rules for cumulative summation are as follows: the first intermediate integral value equals the first local area value, corresponding to the end time of the first sub-interval; the second intermediate integral value equals the first local area value plus the second local area value, corresponding to the end time of the second sub-interval; the third intermediate integral value equals the cumulative sum of the first two local area values ​​plus the third local area value, corresponding to the end time of the third sub-interval. The nth integral intermediate value is equal to the cumulative sum of the first n-1 local area values ​​plus the nth local area value, corresponding to the end time of the nth sub-interval. For example, if the local area value sequence is S1, S2, S3, ..., Sn (n is the number of sub-intervals), then the first integral intermediate value = S1, corresponding to the end time of the first sub-interval; the second integral intermediate value = S1 + S2, corresponding to the end time of the second sub-interval; the third integral intermediate value = S1 + S2 + S3, corresponding to the end time of the third sub-interval; the nth integral intermediate value = S1 + S2 + ... + Sn, corresponding to the end time of the nth sub-interval, which is the end time of the effective waveform data segment. Through the above cumulative summation operation, the discrete local area value sequence is transformed into an integral operation result sequence that increases with time. Each integral intermediate value in this sequence corresponds to the cumulative energy at a certain moment in the pressing process. The final value of the integral operation result sequence is the total energy of the entire pressing process, which retains both the total energy information and the details of the cumulative change of energy over time.

[0057] Based on the sequence of integral calculation results, with the time of the pressing process as the x-axis and the integral calculation result as the y-axis, the points corresponding to each integral calculation result are connected by straight line segments in chronological order, forming a continuous curve that starts from the origin and rises monotonically. This curve is the integral curve reflecting the energy accumulation characteristics of the entire pressing process. Specifically, this includes: establishing a two-dimensional coordinate system with the pressing process time as the x-axis and the integral calculation result as the y-axis. The x-axis ranges from the start time to the end time of the effective waveform data segment, in milliseconds, with the scale evenly distributed according to the end time of the sub-interval, ensuring that each intermediate value of the integral can be found at a corresponding time point on the x-axis; the y-axis ranges from 0 to the final value of the integral calculation result sequence, in millivolt-milliseconds, with the scale set at even intervals to ensure that the trend of the curve is clearly visible; and connecting the points corresponding to each integral calculation result in chronological order... The sequence of integral results is marked with corresponding coordinate points in a two-dimensional coordinate system for each intermediate value of the integral operation. The x-coordinate of each coordinate point represents the end time of the sub-interval corresponding to the intermediate value, and the y-coordinate represents the magnitude of the intermediate value. After marking, adjacent coordinate points are connected sequentially with straight line segments to form a continuous curve. Since the integral operation result is obtained by accumulating and summing local area values, and each local area value is positive, the integral operation result always increases with time. The generated curve starts from the origin, the start time of the effective waveform, and the integral value is 0, showing a monotonically increasing trend. The slope of the curve corresponds to the magnitude of the local area value. The larger the slope, the faster the energy accumulation rate in that time period; the smaller the slope, the slower the energy accumulation rate in that time period. This curve is the integral curve that reflects the energy accumulation characteristics of the entire pressing process.

[0058] By extracting valid waveforms through dynamic thresholding, invalid data is avoided from being mixed in and valid data is lost, and signal fluctuations caused by different batches of wire harnesses and terminal materials are adapted. By trapezoidal integration and cumulative summation, the energy accumulation time sequence changes throughout the crimping process are captured, which helps to improve the pass rate and stability of wire harness and terminal production.

[0059] like Figure 2 As shown, in a preferred embodiment of the present invention, a reference curve for normal crimping is constructed based on the integral curve, and the difference between the measured integral curve and the reference curve within the current crimping cycle is compared to obtain a difference feature data sequence, which may include:

[0060] In this embodiment of the invention, a benchmark sample set is established based on the integral curves corresponding to multiple crimping cycles that have passed quality inspection during historical production. Specifically, this includes selecting multiple crimping cycles that have passed rigorous quality inspection during historical production. Each qualified crimping cycle undergoes three inspection processes: manual visual inspection, electrical continuity testing, and mechanical tensile testing. This ensures that the corresponding crimped terminals are free from defects such as insufficient insulation layer pressing, wire fraying, and abnormal crimping depth. The electrical continuity is good, the mechanical connection is firm, and the quality standards for wire harness production are met. The selection of the sample size must consider both production batch size and statistical reliability requirements. Too few samples will result in a lack of representativeness in the benchmark curve, while too many samples will increase the workload of data processing. Practical verification has shown that selecting… Taking 50-100 qualified crimping cycles as samples ensures both the reliability of statistical results and the efficiency of data processing. Simultaneously, the selected samples must cover qualified crimping cycles from different production batches, wire diameters, and terminal material hardnesses to ensure sample diversity. This allows the constructed benchmark curve to adapt to normal crimping processes under different operating conditions, avoiding the inability of the benchmark curve to adapt to batch fluctuations due to sample limitations. The integral curves corresponding to each qualified crimping cycle are collected. These integral curves are all generated through the above process and have uniform time resolution, amplitude units, and curve characteristics. All these qualified integral curves are organized and archived to establish a benchmark sample set. Each sample is labeled with its corresponding production batch, wire diameter, and terminal material information.

[0061] For each integral curve in the benchmark sample set, the amplitude at each sampling moment is extracted along the time axis of the lamination process with a uniform time resolution. The amplitudes of all benchmark integral curves at the same moment are statistically averaged to obtain the mean integral curve. Simultaneously, the sample standard deviation estimate of the amplitudes of each benchmark integral curve at the same moment is calculated to obtain the amplitude dispersion information of the mean integral curve at each time point. Specifically, this includes: firstly, for each integral curve in the benchmark sample set, the amplitude at each sampling moment is extracted along the time axis of the lamination process with a uniform time resolution, consistent with the previous step of 1ms, ensuring that the amplitude of each integral curve is extracted at the same time point, achieving time alignment between different integral curves and avoiding statistical bias caused by inconsistent time resolution. During the extraction process, for each integral curve, starting from the initial moment (0ms), an amplitude data point is extracted every 1ms until the end of the lamination process, ensuring that each time point has a corresponding amplitude. If an integral curve lacks amplitude at certain time points due to slight differences in the lamination period, linear interpolation is used to supplement it, i.e., based on the amplitudes of two adjacent points before and after that time point. For each time point, the amplitude is calculated, and a supplementary amplitude is obtained to ensure that the amplitude data of all integral curves are perfectly aligned in the time dimension. After extracting the amplitudes of all integral curves, for each unified sampling time, the amplitudes of all benchmark integral curves at that time are statistically averaged to obtain the mean amplitude at that time. The mean amplitudes of all times are arranged in chronological order to form a continuous mean integral curve. The mean amplitude is calculated as follows: mean amplitude at a certain time = sum of amplitudes of all benchmark integral curves at that time ÷ number of samples in the benchmark sample set. This statistical averaging process can offset the random fluctuations of individual qualified samples and highlight the typical pattern of energy accumulation under normal pressing conditions. Simultaneously, for each unified sampling time, the sample standard deviation estimate of the amplitudes of each benchmark integral curve at the same time is calculated. This value characterizes the dispersion of the mean integral curve amplitude at that time point. The greater the dispersion, the greater the fluctuation in energy accumulation of qualified samples at that time; the smaller the dispersion, the more stable the energy accumulation of qualified samples at that time. The calculation of the sample standard deviation estimate follows the sample standard deviation formula, as follows: ,in For the first The sample standard deviation estimate at each sampling time, in millivolt-milliseconds, characterizes the dispersion of the amplitudes of all benchmark integral curves at that time. The number of samples in the baseline sample set, i.e., the number of qualified crimping cycles, is between 50 and 100. For the first The reference integral curve is in the 1st The amplitude at each sampling time point, in millivolt-milliseconds. The value range is 1 to , The value range is from 1 to the total number of sampling times in the crimping cycle; For the first The mean amplitude at each sampling time, in millivolt-milliseconds, is the statistical average of the amplitudes of all reference integral curves at that time, obtained through the averaging process described above. Dividing by [value] in the formula... This is to perform an unbiased estimate of the sample standard deviation, ensuring that the calculation results can more accurately reflect the dispersion of the whole (all qualified pressing cycles) and avoid bias caused by the limited sample size. The sample standard deviation estimate is calculated sequentially for each sampling time to obtain the amplitude dispersion information at all time points.

[0062] The mean-integral curve is used as the central reference trajectory reflecting the typical trend of energy accumulation under normal crimping conditions. The upper and lower offset distances are obtained by expanding the sample standard deviation estimate by a predetermined confidence interval factor. This constructs a baseline reference curve for normal crimping conditions. Specifically, the core function of the mean-integral curve is determined, using it as the central reference trajectory reflecting the typical trend of energy accumulation under normal crimping conditions. The trend of this curve corresponds one-to-one with the stages of the normal crimping process: the curve slope is smaller in the early stages of crimping (slower energy accumulation, corresponding to the beginning of wire deformation), and the curve slope is larger in the middle stages of crimping (faster energy accumulation, corresponding to insulation layer compression and wire shaping). The slope of the curve gradually decreases in the later stages of crimping (energy accumulation slows down, corresponding to terminal curling), which can characterize the energy accumulation law of normal crimping. The upper and lower offset distances of the reference curve are determined. This distance is obtained by multiplying the sample standard deviation estimate by a preset confidence interval coefficient. The selection of the confidence interval coefficient needs to be combined with the quality control requirements of industrial production, taking into account both the false positive rate and the false negative rate. A value between 2 and 3 is usually selected; 2.5 is chosen because this value can cover the fluctuation range of more than 99% of qualified samples, avoiding both misjudging normal fluctuations as abnormal due to a coefficient that is too small and omitting abnormal fluctuations due to a coefficient that is too large. The calculation method for the upper and lower offset distances is as follows: upper offset... The distance is calculated as: (Estimated standard deviation of the sample at the current moment) × confidence interval coefficient; the lower offset distance is also calculated as: (Estimated standard deviation of the sample at the current moment) × confidence interval coefficient. Since the dispersion fluctuates bidirectionally, the values ​​of the upper and lower offset distances are equal. Based on the mean integral curve and the upper and lower offset distances, two continuous reference curves are constructed, forming a reference curve for normal crimping. The upper reference curve = mean integral curve + upper offset distance, reflecting the upper limit of energy accumulation under normal crimping conditions. If the measured integral curve exceeds the upper reference curve, it indicates that the current crimping energy accumulation is too strong, potentially indicating over-crimping, terminal deformation, or other defects. The lower reference curve = mean integral curve - lower offset distance. The offset distance reflects the lower limit of energy accumulation under normal crimping conditions. If the measured integral curve exceeds the lower reference curve, it indicates that the current crimping energy accumulation is weak, and there may be defects such as insufficient crimping or incomplete shaping of the metal wire. After construction, the reference curve is verified by selecting 10 new qualified crimping cycle integral curves and comparing them with the reference curve. If all curves are between the upper and lower reference curves, it indicates that the reference curve is reasonably constructed. If individual curves exceed the range, the confidence interval coefficient is finely adjusted and the upper and lower offset distances are recalculated until the reference curve can cover the fluctuation range of all qualified samples, ensuring the reliability and applicability of the reference curve.

[0063] The measured integral curve within the current crimping cycle is aligned point-by-point with the reference curve along the time axis. After alignment, for each aligned time point, a difference calculation is performed to obtain a set of discrete difference data reflecting the degree of energy accumulation deviation of the current crimping process relative to the normal state. Specifically, this includes: first, performing point-by-point alignment. The core of alignment is to ensure that the time axis of the current measured integral curve is completely consistent with that of the reference curve, avoiding errors in difference calculation caused by slight differences in the crimping cycle or sampling time deviations. The alignment process is as follows: using the time axis of the reference curve as the standard, the starting time of the current measured integral curve is aligned with the starting time of the reference curve. If there is a slight difference between the current measured crimping cycle and the crimping cycle of the reference sample, such as... If the quasi-period is 0.6 seconds and the current measured period is 0.58 seconds, then linear interpolation is used to supplement the amplitude at the end of the current measured integral curve if there are missing time points. If the current measured period is slightly longer, then a time period consistent with the reference period is selected to ensure that the time length and sampling time of the two are completely corresponding and that each time point is aligned. After alignment, a difference calculation is performed for each aligned time point. The specific method of the difference calculation is as follows: the discrete difference data of the current time point = the amplitude of the current measured integral curve at that time - the amplitude of the mean integral curve in the reference curve at that time. It is not directly calculated with the upper and lower reference curves, but only through the difference with the mean curve to accurately reflect the direction and degree of deviation of the current pressing energy from the normal typical state.

[0064] There are three possible results for the difference calculation. If the discrete difference data is positive, it indicates that the energy accumulation at the current moment is higher than the normal typical state. The larger the positive value, the greater the deviation, which may indicate problems such as over-pressing or excessive pressing force. If the discrete difference data is negative, it indicates that the energy accumulation at the current moment is lower than the normal typical state. The larger the absolute value of the negative value, the greater the deviation, which may indicate problems such as insufficient pressing or insufficient pressing force. If the discrete difference data is close to 0, it indicates that the energy accumulation at the current moment is basically consistent with the normal typical state, and the pressing process is normal. After completing the difference calculation point by point, a set of discrete difference data is obtained. Each data point corresponds to a corresponding time point, reflecting the degree of deviation of the energy accumulation from the normal state in each time period during the current pressing process. This provides a basis for the subsequent formation of the difference feature data sequence. At the same time, the discrete difference data is initially verified to check for abnormal differences. For example, if the absolute value of the difference is too large, exceeding 1.5 times the upper and lower offset distance, if abnormal differences are found, the alignment process is checked for accuracy and the measured integral curve is complete. Alignment and difference calculation are then performed again to ensure the accuracy of the discrete difference data.

[0065] The discrete difference data are arranged sequentially according to the time order of the pressing process to form a complete difference feature data sequence. Specifically, this involves arranging all the discrete difference data calculated point-by-point from the start to the end of the pressing process, ensuring the temporal correlation of the difference data and avoiding disorder. For example, the discrete difference data corresponding to the start time (0ms) is placed first in the sequence, the discrete difference data corresponding to 1ms is placed second, and so on, until the discrete difference data corresponding to the end of the pressing is placed last. The resulting difference feature data sequence has a length exactly matching the number of sampling times of the current measured integral curve. Each data point corresponds to a specific pressing time period, and the trend of the sequence reflects the current pressing time. The energy accumulation deviation during the pre-crimping process is analyzed. If the sequence fluctuates around 0 and the fluctuation range is within the upper and lower offset distances, it indicates that the current crimping process is generally normal. If the sequence shows continuous positive fluctuations, it indicates that the energy accumulation during the current crimping process is too strong. If it shows continuous negative fluctuations, it indicates that the energy accumulation during the current crimping process is too weak. If the sequence shows obvious positive and negative abrupt changes in a certain period of time, it indicates that the crimping energy fluctuates abnormally during that period of time, and there may be local defects. The integrity of the difference feature data sequence is checked to see if the sequence length is consistent with the number of sampling times and whether there are any missing or duplicate data. If there are any problems, the difference calculation and arrangement are re-performed. At the same time, the upper and lower offset distances of the sequence and the benchmark reference curve are initially compared, and the difference data exceeding the offset distance are marked to provide preliminary anomaly markers for quality judgment.

[0066] By constructing a benchmark reference curve using multiple batches of qualified samples, misjudgments or omissions caused by a single standard are avoided. Through point-by-point alignment and difference calculation, the degree of energy accumulation deviation of the current crimping process relative to the normal state is quantified. The resulting difference feature data sequence can clearly show the deviation differences of each crimping stage, helping to improve the pass rate and stability of wire harness terminal production.

[0067] In a preferred embodiment of the present invention, the current crimping quality status of the terminal is determined in real time based on whether the difference feature data sequence exceeds a preset boundary envelope range; if the crimping quality status is determined to be unqualified, a control command is obtained through a preset PLC intelligent control component to perform closed-loop control on the crimping distance and crimping force of the terminal machine, thereby realizing process adjustment and quality control in the industrial wire harness production site, which may include:

[0068] In this embodiment of the invention, each discrete difference data in the difference feature data sequence is compared point-by-point with a preset boundary envelope range. Based on the comparison results, all discrete difference data exceeding the boundary envelope range are counted as abnormal deviation points, and the deviation magnitude of each abnormal deviation point is recorded. Based on the number of abnormal deviation points and the cumulative deviation magnitude of each abnormal deviation point, a comprehensive deviation index characterizing the overall abnormality of the current crimping process is calculated. Specifically, this includes: comparing each discrete difference data in the difference feature data sequence with a preset boundary envelope range point-by-point. The preset boundary envelope range is not arbitrarily set, but is determined based on a constructed benchmark reference curve, combined with the quality control precision requirements of industrial wire harness production and the normal fluctuation range of qualified samples. It has been preset after extensive experimental verification to ensure that this range can accommodate the normal energy fluctuations of a qualified crimping process while also identifying abnormal deviations exceeding the normal range, avoiding misjudgment or missed judgment. The preset logic is as follows: the upper limit of the boundary envelope is equal to the upper offset distance (i.e., the sample standard deviation estimate × confidence interval coefficient) at each sampling time, and the lower limit of the boundary envelope is equal to the lower offset distance (i.e., the sample standard deviation estimate × confidence interval coefficient) at each sampling time. The boundary envelope changes dynamically with the sampling time and corresponds completely to the upper and lower reference curves of the benchmark reference curve, ensuring that the boundary envelope at each time point can adapt to the normal fluctuation pattern at that time. For example, in the middle stage of pressing (the stage of rapid energy accumulation), the energy fluctuation of qualified samples is relatively large, and the corresponding boundary envelope also expands accordingly; in the early and late stages of pressing (the stage of slow energy accumulation), the energy fluctuation of qualified samples is relatively small, and the corresponding boundary envelope also shrinks accordingly. The preset boundary envelope value range is usually from -5mV·ms to 5mV·ms, and the specific value is dynamically adjusted according to the sample standard deviation estimate at each sampling time.

[0069] The point-by-point amplitude comparison process is strictly performed sequentially along the timeline of the crimping process. Starting from the crimping start time (0ms), every 1ms (consistent with the sampling interval mentioned earlier), the discrete difference data at the current moment is compared with the preset boundary envelope range (upper and lower limits) for that moment. No difference data at any time point is overlooked, ensuring comprehensive and complete screening of abnormal deviation points. During the comparison, a clear judgment standard is established: if the discrete difference data is greater than the upper limit of the boundary envelope at that moment, or less than the lower limit, the difference data is determined to be outside the boundary envelope range; if the discrete difference data is equal to the upper or lower limit, it is judged as normal fluctuation and not included in the abnormal deviation point category. This judgment standard is also preset to avoid misjudging minor normal fluctuations as abnormalities and improve the accuracy of screening. After the comparison is completed, all discrete difference data exceeding the boundary envelope range are screened out and formally defined as abnormal deviation points. Simultaneously, the deviation magnitude of each abnormal deviation point is recorded. The calculation method for the deviation magnitude is a preset fixed logic to ensure that the calculation results are consistent and comparable. If the abnormal deviation point is higher than the upper limit of the boundary envelope, the deviation magnitude = the difference data of the abnormal deviation point - the upper limit of the boundary envelope; if the abnormal deviation point is lower than the lower limit of the boundary envelope, the deviation magnitude = the lower limit of the boundary envelope - the difference data of the abnormal deviation point. The deviation magnitude values ​​are all positive numbers. The larger the value, the more serious the deviation of the energy accumulation from the normal state at that time point. For example, if the upper limit of the boundary envelope is 4mV·ms at a certain moment, and the discrete difference data at that moment is 6mV·ms, then the deviation magnitude = 6mV·ms - 4mV·ms = 2mV·ms, indicating that the energy accumulation at that moment exceeds the normal upper limit by 2mV·ms, which is a moderate degree of deviation; if the discrete difference data is 9mV·ms, the deviation magnitude = 5mV·ms, indicating that the energy accumulation at that moment is seriously deviating from the normal state.

[0070] After screening abnormal deviation points and recording their deviation magnitudes, all abnormal deviation points are statistically analyzed. This includes the total number of abnormal deviation points, the deviation magnitude of each abnormal deviation point, and the cumulative deviation magnitude of all abnormal deviation points (i.e., the sum of all deviation magnitudes). Based on the number of abnormal deviation points and the cumulative deviation magnitude of each abnormal deviation point, a comprehensive deviation index characterizing the overall abnormality of the current crimping process is calculated. This index comprehensively reflects the severity of the abnormality, avoiding misjudgments caused by judging solely by the number of abnormal points (ignoring the degree of deviation) or a single deviation magnitude (ignoring the range of abnormality). Its calculation logic and related parameters are preset to ensure accurate and comparable results. The calculation of the comprehensive deviation index follows the formula below. ,in This is a comprehensive deviation index, without units, with a preset value range of 0 to 1. This range is preset based on the comprehensive deviation index distribution of qualified and unqualified samples. The closer the value is to 1, the more serious the overall abnormality of the current crimping process; the closer the value is to 0, the closer the current crimping process is to the normal state. If the index is 0, it means that there are no abnormal deviation points and the crimping process is completely normal. This is a weighting coefficient for the number of outliers, without units, with a preset range of 0.4 to 0.6. This range is preset considering the combined impact of the number of outliers and the cumulative deviation on the overall degree of anomaly. If the value is too large, it will overemphasize the number of outliers and ignore serious single-point deviations; if If the value is too small, it will overemphasize the deviation range and ignore the minor anomalies in a large area. The preset value is 0.5, which is used to balance the impact of the number of anomalies and the cumulative deviation range, to ensure the rationality of the comprehensive index and meet the control requirements of the industrial wire harness production that the range and degree of anomalies are equally important. The number of abnormal deviation points, i.e. the number of discrete difference data points that exceed the boundary envelope range, is expressed in units of 0 and ranges from 0 to the total length of the difference feature data sequence. Its statistical rule is a preset fixed logic, that is, only difference data that exceed the upper or lower limit of the boundary envelope are counted, while difference data that are equal to the boundary envelope line are not counted. This represents the total length of the difference feature data sequence, i.e., the total number of sampling moments in the current crimping cycle, expressed in units of [number of sampling moments]. This is consistent with the sampling frequency and crimping cycle mentioned earlier, and the preset value is typically between 500 and 1000. The specific value is determined based on the crimping cycle. For example, if the crimping cycle is 0.5 seconds and the sampling frequency is 1000Hz, then... =500; if the crimping cycle is 1 second. =1000, the purpose of this preset value is to ensure that the denominator is consistent and that the proportion of abnormal points in different crimping cycles is comparable; This is the cumulative deviation magnitude weighting coefficient, which has no unit and a preset value range of 0.4 to 0.6, and is related to the outlier count weighting coefficient. The value range is consistent, and the preset and The sum is 1, the default value is 0.5, and The values ​​are equal to ensure a balanced weight distribution, neither ignoring the number of outliers nor the severity of their deviations, so that the comprehensive deviation index can fully reflect the overall anomaly situation. For the first The deviation magnitude value of each abnormal deviation point is in millivolt-milliseconds and is a positive number, reflecting the severity of the deviation of the abnormal point. The calculation method is a preset fixed logic, which has been explained in detail above to ensure that the calculation standard of the deviation magnitude value of each abnormal deviation point is consistent. For the first The absolute value of each discrete difference data point, in millivolts per millisecond. This is the sum of the absolute values ​​of all discrete difference data in the difference feature data sequence. The preset function of this parameter is to normalize the cumulative deviation amplitude, avoiding distortion caused by differences in the pressing cycle (i.e., ...). The cumulative deviation caused by differences cannot be directly compared, so it is necessary to ensure that the comprehensive deviation index of different crimping cycles is comparable. For example, if the cumulative deviation of two crimping cycles is 100mV·ms, if one cycle... One set of data contains 500 data points, while the other contains 1000. The normalized data can more accurately reflect the actual degree of deviation.

[0071] The specific calculation steps for the comprehensive deviation index are also based on a pre-set fixed process to ensure that the calculation process is standardized and unbiased, and to calculate the percentage of outliers, i.e., the number of outliers. Divide by the total length of the difference feature data sequence The percentage of outliers is obtained; the percentage of outliers is calculated and compared with... The product of the values; calculate the cumulative deviation, which is the sum of the deviation values ​​of all abnormal deviation points. ; Calculate the sum of the absolute values ​​of all discrete difference data. ; Calculate the ratio of the cumulative deviation to the sum of the absolute values; calculate this ratio and The product of the ... The product and the ratio are calculated together. The products of these factors are added together to obtain the comprehensive deviation index. .

[0072] The overall deviation index is compared with a preset quality non-compliance threshold. Based on the comparison result, it is determined whether the overall deviation index exceeds the quality non-compliance threshold. If it does, the current terminal crimping quality status is deemed unqualified. Specifically, after calculating the overall deviation index, it is compared with a preset quality non-compliance threshold to determine the current terminal crimping quality status. This threshold is set in accordance with the quality control requirements of industrial wire harness production and has been verified through a large number of qualified and unqualified samples to ensure that the threshold can effectively distinguish between qualified and unqualified crimping, balancing false positives and false negatives. The preset quality non-compliance threshold is 0.3. This value means that when the overall deviation index exceeds 0.3, it indicates that the overall abnormality of the current crimping process has exceeded the acceptable range, and the crimping quality does not meet production standards; when the overall deviation index is less than or equal to 0.3, it indicates that the current terminal crimping quality is unqualified. If the degree of abnormality in the pre-crimping process is within an acceptable range, the crimping quality is judged as qualified. The calculated comprehensive deviation index is compared with 0.3 using a direct numerical comparison method. If the comprehensive deviation index > 0.3, the current terminal crimping quality is judged as unqualified, and the basis for the unqualified judgment is recorded, including the number of abnormal deviation points, the cumulative deviation range, and the specific value of the comprehensive deviation index, providing a reference for subsequent process adjustments. If the comprehensive deviation index ≤ 0.3, the current terminal crimping quality is judged as qualified, and no further closed-loop control is required; the process proceeds directly to the next crimping cycle. If the process is judged as unqualified, the subsequent closed-loop control process is immediately triggered to ensure timely adjustment of process parameters and avoid the production of more unqualified products. If the process is judged as qualified, the difference characteristic data sequence of the next crimping cycle is continuously monitored to achieve real-time control of crimping quality.

[0073] Based on the non-conformance judgment result, the deviation analysis unit in the preset PLC intelligent control component is invoked to extract the maximum positive deviation and the maximum negative deviation from the difference feature data sequence. The maximum positive deviation is converted into a pressing distance compensation amount according to a preset mapping relationship, and the maximum negative deviation is converted into a pressing force compensation amount according to a preset mapping relationship. Based on the pressing distance compensation amount and the pressing force compensation amount, control instructions are obtained. Specifically, this includes: based on the non-conformance judgment result, invoking the preset PLC intelligent control component, which is pre-configured according to the industrial wire harness crimping process requirements and terminal machine equipment parameters, and has built-in deviation... The analysis unit, with its preset fixed difference analysis rules, can extract key deviation parameters from the difference feature data sequence and convert them into compensation amounts for pressing distance and pressing force, generating control commands. To improve the accuracy of the compensation amount, accelerate the convergence speed of closed-loop regulation, and avoid overshoot (excessive compensation leading to secondary anomalies) or lag (insufficient compensation failing to quickly correct anomalies) during regulation, a reaction path optimization algorithm is incorporated into this stage. The core parameters and optimization logic of this algorithm are preset. By optimizing the generation path of the compensation amount, it ensures that the compensated pressing distance and pressing force can accurately adapt to the current abnormal deviation, achieving rapid and stable process adjustment.

[0074] First, the preset deviation analysis unit in the PLC intelligent control component is invoked. Following preset analysis rules, the difference characteristic data sequence is comprehensively analyzed to extract the maximum positive deviation and the maximum negative deviation. The maximum positive deviation is the maximum value of all positive difference data in the difference characteristic data sequence, reflecting the maximum degree to which energy accumulation exceeds the normal state during the current crimping process, corresponding to problems such as over-crimping, excessive pressure, or insufficient pressing distance. The maximum negative deviation is the minimum value (the largest absolute negative value) of all negative difference data in the difference characteristic data sequence. This reflects the maximum extent to which the energy accumulation during the current crimping process is lower than normal, corresponding to problems such as insufficient crimping, insufficient crimping force, or excessive crimping distance. After extraction, the crimping distance compensation and crimping force compensation are initially obtained according to a preset basic mapping relationship. This basic mapping relationship is preset and determined based on crimping process experience and verification from a large number of abnormal samples. The core logic is that the crimping distance compensation is proportional to the maximum positive deviation, and the preset ratio is determined according to the terminal specifications and crimping material. The larger the maximum positive deviation, the larger the crimping distance compensation, aiming to alleviate the problem of over-crimping by increasing the crimping distance. The crimping force compensation is proportional to the absolute value of the maximum negative deviation, and the preset ratio is also adapted to the actual production conditions. The larger the absolute value of the maximum negative deviation, the larger the crimping force compensation, aiming to alleviate the problem of insufficient crimping by increasing the crimping force. A reaction path optimization algorithm is incorporated to optimize the initially obtained compensation. The core purpose of this algorithm is to optimize the compensation path to ensure that the adjustment process is smooth and converges quickly. Its optimization parameters and iteration rules are preset. The specific implementation process is as follows: First, the preset optimization parameters are initialized, including the compensation adjustment. The parameters are: step size, convergence threshold, and maximum number of iterations. The step size for adjusting the pressing distance is preset to 0.01 mm, and the step size for adjusting the pressing force is preset to 0.1 N. This step size is preset to balance the adjustment accuracy and speed, ensuring accurate adjustment without causing adjustment delay due to an excessively small step size. The convergence threshold is preset to 0.05, meaning that when the predicted deviation corresponding to the optimized compensation amount is less than 0.05, it indicates that the compensation effect has met expectations, and optimization can be stopped. The maximum number of iterations is preset to 10. This value is preset to avoid excessive iterations that could cause adjustment delays and affect production continuity.

[0075] During the optimization process, each iteration calculates the predicted comprehensive deviation index based on the current compensation amount and a preset mapping relationship. The core objective is to minimize the predicted comprehensive deviation index. If the predicted comprehensive deviation index is less than the preset convergence threshold, the iteration stops immediately, and the current compensation amount is taken as the optimal compensation amount. If the convergence threshold is not reached, the compensation amount is adjusted according to the trend of the predicted comprehensive deviation index—if the predicted index is too large, the compensation amount is increased appropriately; if the predicted index is close to the threshold, the adjustment step size is decreased, until the convergence threshold or the preset maximum number of iterations is reached. After optimization, the final pressing distance compensation amount and pressing force compensation amount are obtained. Corresponding control commands are generated based on these two compensation amounts. The format and content of the control commands are preset, explicitly including the adjustment value of the pressing distance and the adjustment value of the pressing force. The pressing distance adjustment value is equal to the current pressing distance plus the pressing distance compensation amount, and the pressing force adjustment value is equal to the current pressing force plus the pressing force compensation amount, ensuring that the servo drive execution unit of the terminal machine can recognize and execute the adjustment operation.

[0076] The control commands are input to the servo drive execution unit of the terminal crimping machine to adjust the crimping distance and pressure, forming a closed-loop feedback adjustment until the difference feature data sequence obtained in the crimping cycle converges again within the preset boundary envelope range, realizing process adjustment and quality control in the industrial wire harness production site. Specifically, after the control commands are generated, they are input to the servo drive execution unit of the terminal crimping machine. This unit is the actuator for the crimping distance and pressure of the terminal crimping machine. It can accurately adjust relevant process parameters according to the control commands to form a closed-loop feedback adjustment, ensuring that the crimping quality quickly converges to the normal range, realizing real-time process adjustment and quality control in the industrial wire harness production site. After receiving the control commands, the servo drive execution unit first parses the crimping distance adjustment value and pressure adjustment value in the command, and then starts the adjustment process. For the adjustment of the crimping distance, the crimping die stroke of the terminal crimping machine is driven by the servo motor to adjust the crimping distance to the value required by the command. During the adjustment process, the actual value of the crimping distance is monitored in real time, and data is fed back through the displacement sensor to ensure that the adjustment error is controlled within 0.001mm to avoid excessive adjustment deviation. For pressure adjustment, the pressure is adjusted to the required value by adjusting the cylinder pressure or the output torque of the servo motor. During the adjustment process, the actual pressure value is monitored in real time, and data is fed back through the pressure sensor to ensure that the adjustment error is controlled within 0.1N, ensuring accurate pressure adjustment. After the adjustment is completed, the terminal crimping machine immediately enters the next crimping cycle. Following the above process, the signal is re-acquired, the integral curve is generated, and the difference characteristic data sequence is calculated. The new crimping quality status is repeatedly judged. If the new difference characteristic data sequence converges within the preset boundary envelope range and the comprehensive deviation index is ≤0.3, it indicates that the closed-loop adjustment is effective and the crimping quality has returned to normal. Continued monitoring is then possible. If the new difference characteristic data sequence still exceeds the boundary envelope range and the comprehensive deviation index is >0.3, it indicates that the adjustment has not achieved the expected effect. The PLC intelligent control component is called back, the compensation amount is adjusted, and the adjustment is executed again until the difference characteristic data sequence converges within the boundary envelope range, realizing fully automatic real-time adjustment. The adjustment cycle is consistent with the crimping cycle to ensure timely correction of crimping process deviations and avoid the continuous generation of unqualified products.

[0077] By comprehensively considering deviation indicators, the crimping quality is determined, avoiding misjudgments and omissions caused by relying on a single indicator. The reaction path optimization algorithm is incorporated to optimize the generation of compensation amounts, improving the accuracy and convergence speed of process adjustments. Through closed-loop feedback adjustment of PLC intelligent control components and servo drive execution units, fully automatic real-time adjustment of crimping distance and crimping force is achieved, enhancing the automation level of the terminal machine monitoring and management system and improving production efficiency.

[0078] Embodiments of the present invention also provide a computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method described above. All implementations in the above method embodiments are applicable to this embodiment and can achieve the same technical effects.

[0079] Experimental example:

[0080] I. Experiment Overview and Parameter Configuration:

[0081] This experimental example demonstrates the specific parameterization of the terminal crimping machine signal acquisition and crimping quality monitoring system. The experiment was conducted at room temperature (25℃), collecting dynamic signal data for 500 complete crimping cycles. The system sampling frequency was 500kHz, the quantization bit depth was 16 bits, and the total number of effective sampling points was 50,000.

[0082] Table 1 System Core Experimental Parameter Configuration

[0083]

[0084] II. Specific Experimental Examples

[0085] (I) Example of Physical Crimping and Differential Voltage Signal Acquisition

[0086] During the physical crimping of wire harness terminals by the terminal crimping machine, multiple sets of resistive strain sensing elements attached to the stress-sensitive area of ​​the mold base plate are used to obtain micron-level elastic deformation information of the base plate under dynamic crimping force in real time. When the crimping force is applied to the mold base plate, the sensing element generates a change in resistance. The change is converted into an initial unbalanced differential voltage signal using a Wheatstone full-bridge measurement circuit. Its positive and negative components correspond to the deformation polarities of the tensile and compression zones of the base plate, respectively.

[0087] Experimental results show that under normal crimping, the differential voltage signal amplitude ranges from -2.5V to +2.5V, and the signal rise slope is approximately 250V / ms. When the crimping force is too large, the signal amplitude exceeds +2.5V in the positive direction; when the crimping force is insufficient, the signal amplitude exceeds -2.5V in the negative direction. After common-mode noise suppression and high-frequency vibration attenuation processing, the differential voltage signal is finally used as a clean differential voltage signal for subsequent processing, with a measured signal-to-noise ratio of 42.3dB.

[0088] (II) Signal Conditioning and Digital Sampling Experiment Examples

[0089] The conditioning module sequentially passes the differential voltage signal through a pre-stage anti-aliasing filter network (low-pass cutoff frequency set to 200kHz), a programmable gain amplifier unit (dynamic gain range 0.5× to 8×), and a high-resolution analog-to-digital converter (500kHz sampling, 16-bit quantization) to finally obtain a digital sampled signal sequence. In this experimental example, a total of 50,000 sampling points were collected (corresponding to 500 crimping cycles, each cycle 100ms), with a total sampling time of 50 seconds.

[0090] (III) Experimental Example of Dynamic Threshold Extraction and Effective Waveform Extraction

[0091] The extraction module uses a sliding time window (0.01ms wide) to traverse the digitally sampled signal segment by segment along the time axis. Within each window, it calculates the mean amplitude and instantaneous slope value, and constructs a joint feature set that integrates amplitude statistical features and slope change features. Based on the joint feature set, it dynamically generates an upper threshold curve and a lower threshold curve. The upper threshold curve is used to detect the starting position of the rising edge of the crimped waveform, and the lower threshold curve is used to detect the ending position of the falling edge of the waveform. When the signal crosses the upper threshold from below, the starting trigger point is marked. When the signal crosses the lower threshold from above and subsequent consecutive sampling points remain below the lower threshold, the ending trigger point is marked.

[0092] This experiment set seven different sliding window widths (0.005ms, 0.01ms, 0.02ms, 0.05ms, 0.1ms, 0.2ms, and 0.5ms), and counted the number of valid waveform trigger events for each. The experimental results show that when the window width is 0.005ms, the number of trigger events is 387, indicating a relatively high number of missed events. When the window width is increased to 0.01ms, the number of trigger events increases rapidly to 452, and when it is further increased to 0.05ms, it reaches 478 and tends to stabilize.

[0093] like Figure 3 The figure shows the trend of the number of effective waveform trigger events under different sliding window widths, as well as the corresponding upper and lower threshold boundaries. The number of effective trigger events (blue solid line) begins to converge rapidly at a window width of 0.01ms and eventually stabilizes at 479. The light blue filled area between the upper threshold boundary (red dashed line) and the lower threshold boundary (orange dashed line) is the effective trigger interval. The experimental results show that 0.01ms is the final window width of the dynamic threshold under the experimental conditions, which can effectively suppress noise triggering while ensuring the event capture rate.

[0094] (iv) Experimental example of trapezoidal integral and integral curve construction

[0095] The extraction module divides the effective waveform data segment into several continuous sub-intervals with a width of 0.002ms (2μs) in time sequence. The sub-intervals are connected end to end and do not overlap. In each sub-interval, adjacent sampling points are connected by straight line segments to form a trapezoid. The area of ​​each trapezoid is calculated and summed. Here, x is the sampling time and y is the corresponding amplitude.

[0096] like Figure 4 The diagram illustrates the trapezoidal integral segmented accumulation process of the effective waveform signal (blue curve). The orange filled area represents the trapezoidal area of ​​each sub-interval. The integral result is obtained by summing the areas of each sub-interval.

[0097] (V) Experimental Example of Constructing Integral Curves

[0098] After completing the trapezoidal integral calculation, the extraction module accumulates and sums the local area values ​​according to the time sequence of the pressing process, transforming the local area value sequence into an integral calculation result sequence that monotonically increases with time. This integral calculation result sequence is the integral curve that reflects the energy accumulation characteristics of the entire pressing process. In this experimental example, the peak value range of the integral curve for 500 pressing cycles is 4.2 to 9.8. The upward slope of the integral curve gradually slows down with the pressing process, which is consistent with the actual physical process of the pressing force decreasing from large to small.

[0099] (vi) Statistical Experiment Example of Multi-stage Processing Funnel for Integral Curve

[0100] The analysis module constructs a reference curve under normal crimping conditions based on the obtained integral curve, and compares the measured integral curve of the current crimping cycle with the reference curve point by point to obtain the difference feature data sequence. When the difference feature data sequence exceeds the preset boundary envelope range (±0.3), the current terminal crimping quality status is determined to be unqualified or in a warning state.

[0101] (vii) PLC Closed-Loop Control and Regulation Convergence Experiment Example

[0102] Based on the non-conformance judgment result, the control module calls the deviation analysis unit in the preset PLC intelligent control component to extract the maximum positive deviation (corresponding to over-pressing) and the maximum negative deviation (corresponding to under-pressing) from the difference feature data sequence. Based on the preset mapping relationship, the deviation is converted into pressing distance compensation and pressing force compensation. The reaction path optimization algorithm is incorporated to iteratively optimize the compensation. During the optimization process, each iteration calculates and predicts the comprehensive deviation index based on the current compensation. If the index is less than the convergence threshold of 0.05, the iteration stops; otherwise, the compensation is adjusted according to the changing trend.

[0103] In this experimental example, four closed-loop regulation events were triggered. The optimization iteration convergence process of each event is recorded as follows: the initial comprehensive deviation index of regulation event A was 0.78, which converged to 0.05 after 7 iterations; regulation event B converged from 0.65 to 0.06 after 6 iterations; regulation event C converged from 0.55 to 0.09 after 7 iterations; and regulation event D converged from 0.82 to 0.05 after 6 iterations. All four regulation events reached or approached the convergence threshold within 10 iterations of the maximum number of iterations.

[0104] like Figure 5 The figure shows the convergence curves of the comprehensive deviation index of the four closed-loop adjustment events as a function of the number of optimization iterations. Each event reached the convergence threshold of 0.05 (red dashed line) within 7 iterations, which is below the unqualified threshold of 0.3 (gray dashed line).

[0105] III. Experimental Conclusions

[0106] This experimental example verifies the signal acquisition and processing chain through 500 crimping cycles, leading to the following conclusions: First, the conditioning module, based on anti-aliasing filtering, adaptive gain adjustment, and high-resolution analog-to-digital conversion signal chain design, can effectively suppress noise interference and accurately acquire crimping dynamic deformation signals. Second, the extraction module, based on sliding window dynamic threshold truncation and trapezoidal integral algorithm, can extract effective waveform data segments and quantify crimping energy accumulation characteristics. Third, the analysis module, by constructing a reference curve and setting a ±0.3 boundary envelope, can identify crimping process deviations in the early stages, reserving sufficient adjustment time and space for the closed-loop adjustment of the control module. Fourth, the control module incorporates a PLC closed-loop control strategy with a reaction path optimization algorithm, enabling rapid convergence within a limited number of iterations. All four adjustment events reached the convergence threshold within seven iterations, effectively achieving fully automatic real-time adjustment of crimping distance and crimping force. Fifth, the closed-loop feedback adjustment mechanism ensures that unqualified crimping cycles reconverge within the boundary envelope range after adjustment, improving the automated monitoring and management level of the terminal block machine and the quality of wire harness production.

[0107] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A terminal block machine monitoring and management system for wire harness production, characterized in that, include: The conditioning module is used to perform anti-aliasing filtering on the differential voltage signal to obtain the processed analog signal; it performs adaptive gain adjustment according to the dynamic range of the analog signal amplitude to obtain the amplitude-matched conditioning signal, and converts the amplitude-matched conditioning signal into a digital sampling signal through analog-to-digital conversion; The extraction module is used to perform real-time dynamic thresholding of the digitized sampled signal based on its amplitude and slope characteristics, extracting effective waveform data segments. This includes using the digitized sampled signal as an input sequence, traversing the time axis segment by segment using a sliding time window, calculating the mean amplitude, local maxima, and local minima within each window, and performing a first-order difference operation on adjacent sampled points to extract the instantaneous slope values, resulting in a joint feature set that integrates amplitude statistical features and slope change features. This joint feature set is then input to a dynamic threshold generator, which determines the baseline threshold based on the mean amplitude within the current window and adjusts the threshold offset in real-time using the polarity and absolute value of the instantaneous slope values, dynamically generating an upper threshold curve and a lower threshold curve. The original digitized sampled signal is then compared point-by-point with the dynamically generated upper and lower threshold curves. When the signal amplitude crosses below the threshold curve, the generator performs a point-by-point amplitude comparison. When the signal amplitude exceeds the threshold, the current sampling time is marked as the starting trigger point of the valid waveform. When the signal amplitude crosses the lower threshold from above and subsequent consecutive sampling points remain below the lower threshold, the current sampling time is marked as the ending trigger point of the valid waveform, thus obtaining the starting and ending trigger points. Using the starting and ending trigger points as the truncation boundaries, all sampling points between the two boundaries are extracted from the original digital sampling signal sequence, and all invalid silent data segments before the starting point and after the ending point are automatically removed, completing the extraction of the valid waveform data segment. The discrete transient data sequence in the valid waveform data segment is divided into several continuous sub-intervals at equal intervals according to the time sequence. In each sub-interval, adjacent sampling points are connected by straight line segments to form a trapezoid. The area of ​​each trapezoid is calculated and summed to obtain the integral operation result. Based on the integral operation result, an integral curve reflecting the energy accumulation characteristics of the entire pressing process is obtained. The analysis module is used to construct a reference curve under normal crimping conditions based on the integral curve, and compare the difference between the integral curve measured in the current crimping cycle and the reference curve to obtain the difference feature data sequence. The control module is used to determine the current crimping quality status of the terminal in real time based on whether the difference feature data sequence exceeds the preset boundary envelope range. If the crimping quality status is determined to be unqualified, the control command is obtained through the preset PLC intelligent control component to perform closed-loop control on the crimping distance and crimping force of the terminal machine, so as to realize the process adjustment and quality control of the industrial wire harness production site.

2. The terminal block machine monitoring and management system for wire harness production according to claim 1, characterized in that, During the physical crimping of the wire harness terminals by the terminal crimping machine before the conditioning module, the elastic deformation information of the mold base plate under dynamic crimping force is acquired in real time, and a differential voltage signal characterizing the actual deformation degree of the base plate is extracted based on the elastic deformation information, including: Based on the micron-level elastic deformation information generated by the mold base plate under dynamic pressing force, multiple sets of resistive strain sensitive elements attached to the stress-sensitive area of ​​the base plate are used to convert the elastic deformation information into the resistance change of each sensitive element in real time. Based on the change in impedance, the impedance difference between each sensitive element is converted into an initial unbalanced differential voltage signal using a Wheatstone full-bridge measurement circuit. The positive and negative components of this signal correspond to the deformation polarities of the tensile and compression zones of the base plate, respectively. Common-mode noise suppression and primary differential amplification are performed on the initial unbalanced differential voltage signal to obtain an intermediate-stage differential voltage signal after interference filtering. The intermediate-stage differential voltage signal is attenuated by high-frequency mechanical vibration noise to obtain the final differential voltage signal that characterizes the actual dynamic deformation of the mold base plate in real time.

3. The terminal block machine monitoring and management system for wire harness production according to claim 2, characterized in that, The differential voltage signal is subjected to anti-aliasing filtering to obtain a processed analog signal; adaptive gain adjustment is performed based on the dynamic range of the analog signal's amplitude to obtain an amplitude-matched conditioned signal; and the amplitude-matched conditioned signal is converted into a digital sampled signal through analog-to-digital conversion, including: The final differential voltage signal, which characterizes the actual dynamic deformation of the mold base plate in real time, is input into the pre-stage anti-aliasing filter network. By setting the cutoff frequency of the low-pass filter, high-frequency noise components higher than the Nyquist frequency are suppressed to obtain a smooth and continuous analog signal. Based on the instantaneous amplitude distribution range of the analog signal, the matching coefficient between the current signal peak and the full-scale input range of the analog-to-digital converter is dynamically calculated. Based on the matching coefficient, the gain factor of the programmable gain amplifier unit is automatically adjusted so that the amplitude of the output signal is mapped to the final quantization range of the analog-to-digital converter, thus obtaining an amplitude-matched conditioned signal. The amplitude-matched conditioning signal is input into a high-resolution analog-to-digital converter, which converts it into a discrete digital sequence, i.e., a digital sampling signal, at a preset sampling frequency and quantization bit depth.

4. The terminal block machine monitoring and management system for wire harness production according to claim 3, characterized in that, The upper threshold curve is used to detect the starting position of the rising edge of the crimping waveform, and the lower threshold curve is used to detect the ending position of the falling edge of the waveform.

5. The terminal block machine monitoring and management system for wire harness production according to claim 4, characterized in that, The discrete transient data sequence in the effective waveform data segment is divided into several continuous sub-intervals at equal intervals according to time sequence. Within each sub-interval, adjacent sampling points are connected by straight line segments to form trapezoids. The area of ​​each trapezoid is calculated and summed to obtain the integral operation result. Based on the integral operation result, an integral curve reflecting the energy accumulation characteristics of the entire pressing process is obtained, including: Based on the effective waveform data segment, the discrete sampling points are arranged in ascending order of sampling time, and the width of the sub-interval is used as a fixed time interval. Starting from the starting point, the entire effective waveform data segment is divided into multiple continuous sub-intervals that are connected end to end and do not overlap. Each sub-interval includes two adjacent sampling points, resulting in an ordered set of sub-intervals. Based on an ordered set of sub-intervals, each sub-interval is extracted sequentially. For the current sub-interval, the area of ​​the trapezoid formed by the line connecting adjacent sampling points and the time axis is calculated, with the amplitude of the left sampling point as the left height of the trapezoid, the amplitude of the right sampling point as the right height of the trapezoid, and the fixed time interval as the width of the trapezoid. This yields the local area value that characterizes the instantaneous energy contribution of the sub-interval, and the local area value sequence is obtained according to the time order of the sub-intervals. Based on the local area value sequence, starting from the first local area value, the current local area value is added to the cumulative sum of all previous local area values ​​in turn. Each calculation yields an intermediate integral value that increases with the pressing process. The intermediate integral value is then associated with the end time of the current sub-interval, thus transforming the local area value sequence into a sequence of integral calculation results that increases with time. Based on the sequence of integral calculation results, with the time of the pressing process as the horizontal axis and the integral calculation result as the vertical axis, the points corresponding to each integral calculation result are connected by straight line segments in chronological order to form a continuous curve that starts from the origin and rises monotonically. This curve is the integral curve that reflects the energy accumulation characteristics of the entire pressing process.

6. The terminal block machine monitoring and management system for wire harness production according to claim 5, characterized in that, A reference curve for normal crimping is constructed based on the integral curve, and the difference between the measured integral curve and the reference curve within the current crimping cycle is compared to obtain a sequence of difference characteristic data, including: A benchmark sample set is established based on the integral curves corresponding to multiple crimping cycles that have been judged to be qualified through quality inspection during the historical production process. For each integral curve in the benchmark sample set, the amplitude at each sampling time is extracted along the time axis of the pressing process with a uniform time resolution. The amplitudes of all benchmark integral curves at the same time are statistically averaged to obtain the mean integral curve. At the same time, the sample standard deviation estimate of the amplitude of each benchmark integral curve at the same time is calculated to obtain the amplitude dispersion information of the mean integral curve at each time point. The mean integral curve is used as the central reference trajectory to reflect the typical trend of energy accumulation under normal crimping conditions. The sample standard deviation estimate is expanded by a preset confidence interval coefficient to obtain the value as the upper and lower offset distance, thus constructing the benchmark reference curve for normal crimping conditions. The integral curve measured in the current crimping cycle is aligned point by point with the reference curve along the time axis. After alignment, for each aligned time point, the difference calculation is performed to obtain a set of discrete difference data reflecting the degree of energy accumulation deviation of the current crimping process relative to the normal state. The discrete difference data are arranged sequentially according to the time order of the pressing process to form a complete difference feature data sequence.

7. The terminal block machine monitoring and management system for wire harness production according to claim 6, characterized in that, The reference curve corresponds to a central amplitude and an allowable positive and negative fluctuation range at each time point in the crimping process.

8. The terminal block machine monitoring and management system for wire harness production according to claim 7, characterized in that, Based on whether the difference feature data sequence exceeds the preset boundary envelope range, the current crimping quality status of the terminal is determined in real time. If the crimping quality status is determined to be unqualified, control instructions are obtained through the preset PLC intelligent control component to perform closed-loop control on the crimping distance and crimping force of the terminal machine, realizing process adjustment and quality control in the industrial wire harness production site, including: Each discrete difference data in the difference feature data sequence is compared point by point with the preset boundary envelope range. Based on the comparison results, all discrete difference data that exceed the boundary envelope range are counted as abnormal deviation points, and the deviation magnitude of each abnormal deviation point is recorded. Based on the number of abnormal deviation points and the cumulative deviation magnitude of each abnormal deviation point, a comprehensive deviation index that characterizes the overall abnormality of the current crimping process is calculated. The comprehensive deviation index is compared with the preset quality failure threshold. Based on the comparison result, it is determined whether the comprehensive deviation index exceeds the quality failure threshold. If it does, the current terminal crimping quality status is determined to be unqualified. Based on the non-compliance judgment result, the deviation analysis unit in the preset PLC intelligent control component is called to extract the maximum positive deviation and the maximum negative deviation from the difference feature data sequence. The maximum positive deviation is converted into the pressing distance compensation amount according to the preset mapping relationship, and the maximum negative deviation is converted into the pressing force compensation amount according to the preset mapping relationship. Based on the pressing distance compensation amount and the pressing force compensation amount, the control command is obtained. The control commands are input to the servo drive execution unit of the terminal crimping machine to adjust the crimping distance and crimping force, forming a closed-loop feedback adjustment until the difference feature data sequence obtained in the crimping cycle converges again within the preset boundary envelope range, thereby realizing process adjustment and quality control in the industrial wire harness production site.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a program that, when executed by a processor, implements the system as described in any one of claims 1 to 8.

Images