Method and device for diagnosing fault of electric compressor based on multi-source data analysis

Abstract

The present application relates to the technical field of electric data processing, in particular to a motor compressor fault diagnosis method and device based on multi-source data analysis. The method collects the direct current intensity sequence and vibration intensity sequence of the motor compressor, removes the baseline under the respective dimension data in the step response section, iteratively accumulates and analyzes each time in the step response section, and respectively obtains the current drive cumulative intensity signal and the mechanical vibration cumulative intensity signal; the maximum nonlinear hysteresis is determined through the matching process, the maximum hysteresis time corresponding to the maximum nonlinear hysteresis is taken as the reference, the drive power is queried back as the verification power, the verification power and the maximum nonlinear hysteresis are taken as the fault judgment reference, and the fault judgment is performed. The present application can effectively analyze the motor compressor fault from the two dimensions, take the verification power as the judgment constraint, avoid the influence of inductance working condition on the final fault diagnosis result, and improve the detection accuracy of the motor compressor fault.

Description

Technical Field

[0001] This invention relates to the field of electrical data processing technology, and more specifically to a method and apparatus for diagnosing electric compressor faults based on multi-source data analysis. Background Technology

[0002] As a core component of the thermal management system in new energy vehicles, the electric compressor typically consists of a motor driving a scroll assembly via a coupling or keyway structure. During long-term operation or assembly, mechanical failures such as excessive clearance or loosening of the drivetrain may occur.

[0003] Currently, the mainstream fault detection method in the industry involves controlling the compressor to run at a constant speed during the final inspection stage of the production line and collecting the casing vibration signal for spectrum analysis. However, this detection method based on steady-state operating conditions has obvious limitations. When the motor rotates at a constant speed, the continuous unidirectional centrifugal force and gas back pressure will tightly press loose mechanical parts in one direction, causing the gap to close, thereby masking the impact vibration characteristics and resulting in missed detection.

[0004] To further extract impact vibration characteristics, existing technologies focus on analyzing transient signals during motor startup or acceleration. These transient signals mainly consist of transient current and transient vibration signals. The time difference between these two types of signals is analyzed to determine if there is hysteresis in the mechanical response. However, under high load or high inductance conditions, motors exhibit a physical phenomenon where the magnetic field builds up ahead of the rotor motion. This makes it impossible for existing technologies to distinguish between abnormal hysteresis caused by mechanical backlash idling and normal hysteresis caused by this inductance condition, further leading to misjudgments and affecting fault diagnosis efficiency. Summary of the Invention

[0005] To address the technical problem of false fault detection caused by directly analyzing transient signals under high inductance conditions in electric compressors, the present invention aims to provide a method and apparatus for fault diagnosis of electric compressors based on multi-source data analysis. The specific technical solution adopted is as follows: This invention proposes a fault diagnosis method for electric compressors based on multi-source data analysis, the method comprising: Obtain the DC current intensity sequence and vibration intensity sequence of the electric compressor; The analysis period is constructed with the motor start-up time as the center. The analysis period includes a background noise segment and a step response segment. In the step response segment, the DC current intensity sequence and the vibration intensity sequence are processed respectively. For each moment, the baseline in the background noise segment is removed to obtain the current driving intensity and mechanical vibration intensity at each moment. In the step response segment, the current driving intensity and mechanical vibration intensity at each moment are iteratively accumulated and analyzed to obtain the current driving intensity signal and the mechanical vibration intensity signal respectively; the current driving intensity signal and the mechanical vibration intensity signal are matched point by point to obtain the best matching pair set. The difference between the timestamps of the cumulative mechanical vibration intensity and the cumulative current drive intensity in the best-matched pair is used as the nonlinear hysteresis. The maximum nonlinear hysteresis in the set of best-matched pairs is obtained, and the time of the cumulative current drive intensity corresponding to the maximum nonlinear hysteresis is taken as the maximum hysteresis time. The driving power of the electric compressor at the maximum lag time is used as the verification power, and the fault diagnosis is performed in combination with the maximum nonlinear lag.

[0006] Further, obtaining the DC current intensity sequence and vibration intensity sequence of the electric compressor includes: Acquire the three-phase current signal and vibration acceleration signal of the electric compressor; perform DC processing on the three-phase current signal to obtain a DC current intensity sequence; perform envelope processing on the vibration acceleration signal, and use the obtained vibration envelope sequence as the vibration intensity sequence.

[0007] Furthermore, the method for determining the motor start-up time includes: If a start command is issued at a certain moment, and the gradient in the DC current intensity sequence is greater than a preset gradient threshold, then the motor is considered to have started.

[0008] Furthermore, the method for constructing the time period to be analyzed includes: Taking the motor start-up time as the center, a time period of a preset first length is divided forward, and a time period of a preset second length is divided backward to obtain the time period to be analyzed; the time period of the preset first length before the motor start-up time is the background noise segment, and the time period of the preset second length after the motor start-up time is the step response segment.

[0009] Furthermore, obtaining the current driving intensity and mechanical vibration intensity at each moment includes: The average value of the DC current intensity sequence in the background noise segment is used as the current baseline value, and the average value of the vibration intensity sequence in the background noise segment is used as the vibration baseline value. For each moment in the step response segment, the corresponding vibration intensity is subtracted from the vibration baseline value to obtain the mechanical vibration intensity; if the mechanical vibration intensity is less than 0, then the mechanical vibration intensity is assigned a value of 0. For each moment in the step response segment, the current driving intensity is obtained by subtracting the current baseline value from the corresponding DC current intensity; if the current driving intensity is less than 0, the current driving intensity is assigned a value of 0.

[0010] Furthermore, after obtaining the DC current intensity sequence and the vibration intensity sequence, the process also includes: The DC current intensity sequence and the vibration intensity sequence are resampled and aligned.

[0011] Furthermore, the method for obtaining the set of best matching pairs includes: The normalized results of the current-driven cumulative intensity signal and the mechanical vibration cumulative intensity signal are used as the row dimension and column dimension, respectively, to construct a matching matrix. Each element in the matching matrix is ​​the difference between the corresponding normalized current-driven cumulative intensity and the normalized mechanical vibration cumulative intensity. Using the dynamic time warping algorithm and the Sakoe-Chiba band as the constraint window, the minimum cumulative distance path of the matching matrix is ​​obtained according to the constraint conditions. The constraints include: the path starts at the bottom left corner of the matching matrix and ends at the top right corner of the matching matrix; the path can only move right, up, or to the upper right. The normalized current-driven cumulative intensity and the normalized mechanical vibration cumulative intensity corresponding to each path point in the minimum cumulative distance path constitute the best matching pair, and all path points constitute the set of the best matching pairs.

[0012] Furthermore, the method for obtaining the nonlinear hysteresis includes: For each optimal match, the nonlinear hysteresis is obtained by subtracting the timestamp of the normalized cumulative mechanical vibration intensity from the timestamp of the normalized cumulative current drive intensity.

[0013] Furthermore, the fault diagnosis method includes: If the maximum nonlinear hysteresis is less than or equal to the preset hysteresis threshold, it is deemed qualified. If the maximum nonlinear hysteresis is greater than the preset hysteresis threshold and the verification power is greater than or equal to the preset power threshold, then it is determined to be a high-sensitivity qualified operating condition. If the maximum nonlinear hysteresis is greater than a preset hysteresis threshold and the verification power is less than a preset power threshold, then it is determined to be a fault condition.

[0014] This invention also proposes a fault diagnosis device for electric compressors based on multi-source data analysis, the device comprising: The electric compressor operation data acquisition module is used to acquire the DC current intensity sequence and vibration intensity sequence of the electric compressor. The data feature extraction module is used to construct the analysis period centered on the motor start-up time. The analysis period includes a background noise segment and a step response segment. In the step response segment, the DC current intensity sequence and vibration intensity sequence are processed respectively. For each moment, the baseline in the background noise segment is removed to obtain the current driving intensity and mechanical vibration intensity at each moment. The drive response coupling analysis module is used to perform iterative cumulative analysis at each time step in the step response segment to obtain the current drive cumulative intensity signal and the mechanical vibration cumulative intensity signal; the current drive cumulative intensity signal and the mechanical vibration cumulative intensity signal are matched point by point to obtain the best matching pair set; The maximum lag time determination module is used to take the difference between the timestamps of the cumulative mechanical vibration intensity and the cumulative current drive intensity in the best matching pair as the nonlinear lag amount, obtain the maximum nonlinear lag amount in the set of best matching pairs, and take the time of the cumulative current drive intensity corresponding to the maximum nonlinear lag amount as the maximum lag time. The fault diagnosis module is used to backtrack and query the drive power of the electric compressor at the maximum lag time as the verification power, and perform fault diagnosis in combination with the maximum nonlinear lag.

[0015] The present invention has the following beneficial effects: This invention acquires the DC current intensity sequence and vibration intensity sequence of an electric compressor, avoiding triggering confusion and subsequent power calculation problems caused by the zero-crossing point of the original AC data. Further feature processing is performed on the acquired instantaneous signals to identify the motor start-up moment and construct the analysis period. Within this analysis period, a step response segment and a background noise segment are divided. Baselines in each dimension of the data can be removed within the step response segment, resulting in a clean and characteristically significant signal intensity data segment. Considering the random error inherent in simple intensity threshold judgment, this invention further performs iterative cumulative analysis at each moment within the step response segment to obtain the current-driven cumulative intensity signal and the mechanical vibration cumulative intensity signal, respectively. Then, the two signals are matched point-by-point. The maximum nonlinear hysteresis is determined through the matching process, enabling precise quantification of minute idling slippage during the start-up process. This invention further uses the maximum lag time corresponding to the maximum nonlinear lag as a benchmark, backtracks to query the driving power as the verification power, and uses the verification power and the maximum nonlinear lag together as the fault judgment benchmark. It can effectively analyze electric compressor faults from two dimensions. Using the verification power as a judgment constraint avoids the influence of inductive conditions on the final fault diagnosis results and improves the monitoring accuracy of electric compressor faults. Attached Figure Description

[0016] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of a fault diagnosis method for an electric compressor based on multi-source data analysis, provided as an embodiment of the present invention. Detailed Implementation

[0018] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of an electric compressor fault diagnosis method and apparatus based on multi-source data analysis proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0019] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0020] The following description, in conjunction with the accompanying drawings, details the specific scheme of the electric compressor fault diagnosis method and device based on multi-source data analysis provided by the present invention.

[0021] Please see Figure 1 The diagram illustrates a flowchart of a fault diagnosis method for an electric compressor based on multi-source data analysis, according to an embodiment of the present invention. The method includes: Step S1: Obtain the DC current intensity sequence and vibration intensity sequence of the electric compressor.

[0022] This invention addresses the mechanical clearance fault detection scenario during the quality inspection process of electric compressor production. Since electric compressors are typically driven by permanent magnet synchronous motors, the original phase current fed back by their controller is a sinusoidal fluctuating AC signal with periodic zero-crossing points and negative half-cycles. If the original signal is directly used for power calculation or gradient analysis, the zero-crossing value will return to zero, leading the system to misjudge that the motor is not working or has extremely low power. Therefore, this invention acquires and analyzes the corresponding DC current intensity sequence of the electric compressor, and combines it with the vibration intensity sequence collected during the same period to perform a drive-response coupling analysis in subsequent processes.

[0023] Preferably, in this embodiment of the invention, obtaining the DC current intensity sequence and vibration intensity sequence of the electric compressor includes: The three-phase current signal and vibration acceleration signal of the electric compressor are acquired; the three-phase current signal is processed by DC to obtain a DC current intensity sequence. The DC processing includes existing techniques such as RMS calculation and Park transform. In this embodiment, Park transform is chosen to convert the original three-phase current into two components in a rotating coordinate system. The vector magnitudes of the two components at each moment are calculated as the DC current intensity to obtain the final DC current intensity sequence. This process is a well-known technique and will not be elaborated or limited here.

[0024] Vibration signals from electric compressors are typically acquired by directly collecting vibration acceleration signals. However, since vibrations generated by mechanical impacts usually exhibit high-frequency modulated waveforms, direct analysis of the original waveform is not intuitive. Therefore, this invention performs envelope processing on the vibration acceleration signals, and the resulting vibration envelope sequence is used as the vibration intensity sequence. The envelope processing technique can be any existing technology such as Hilbert transform or detection, rectification, and filtering. This invention selects the Hilbert transform method, and the resulting vibration intensity sequence reflects the instantaneous change trend of mechanical vibration energy over time. The specific method is well-known to those skilled in the art and will not be elaborated or limited here.

[0025] It should be noted that in the actual acquisition of current and vibration signals from an electric compressor, the sampling frequency of the vibration signal is usually higher than that of the current signal. For example, the sampling frequency of the vibration signal is usually 50kHz, while that of the current signal is usually 1kHz. This necessitates point-by-point matching in subsequent processes, resulting in different numbers of data points in the two sequences. Therefore, in some preferred embodiments of this invention, after obtaining the DC current intensity sequence and the vibration intensity sequence, a resampling and alignment operation is performed on the two sequences to ensure that the number of data points is the same, facilitating subsequent point-by-point matching. The resampling and alignment operation is a well-known technique. In this embodiment, considering the higher sampling frequency of the vibration signal, the vibration intensity sequence can be directly downsampled using the sampling frequency of the lower-frequency current intensity sequence as the reference frequency. First, the ratio of the vibration signal sampling frequency to the reference frequency can be used as the downsampling ratio R. In the vibration intensity sequence, one data point is extracted every R points, ultimately generating the downsampled vibration intensity sequence. The specific downsampling operation is a well-known technique and will not be elaborated or limited here. In specific implementation scenarios, an appropriate sampling operation can be selected based on the actual signal acquisition situation.

[0026] Step S2: Construct the analysis period centered on the motor start-up time. The analysis period includes a background noise segment and a step response segment. In the step response segment, process the DC current intensity sequence and the vibration intensity sequence respectively. For each moment, remove the baseline in the background noise segment to obtain the current driving intensity and mechanical vibration intensity at each moment.

[0027] The time period encompassed by the motor startup time is the target time period for this embodiment of the invention; therefore, the time period to be analyzed can be constructed centered on the motor startup time. Due to the zero-point drift of the sensor and the continuous environmental vibration background noise in industrial environments, the acquired raw signals are usually superimposed with a static DC bias or random noise baseline. Since the time period to be analyzed is constructed centered on the motor startup time, it includes both the time period before and after the motor startup time. The information acquired before the motor startup time can be considered background or noise information, while the information acquired after the motor startup time is the truly meaningful signal strength information. Therefore, this embodiment of the invention divides the time period to be analyzed into a background noise segment and a step response segment.

[0028] Based on the background noise segment, the baselines of the current and vibration dimensions can be obtained respectively. Then, in the step response segment, the DC current intensity sequence and the vibration intensity sequence are processed respectively. By removing the baseline of each dimension at each time point, the current driving intensity and mechanical vibration intensity, which characterize the net signal intensity, can be obtained respectively.

[0029] Preferably, in some implementations of the present invention, considering that conventional electric compressors generate obvious command signals during operation, in order to more accurately identify the motor start-up time, the method for determining the motor start-up time in the embodiments of the present invention includes: if a start command is generated at a certain moment, and the gradient in the DC current intensity sequence is greater than a preset gradient threshold, then it is considered the motor start-up time. That is, the motor start-up time is considered to be when the current undergoes an effective operational change at the same time as the command is generated, avoiding the problem of inaccurate motor start-up time identification caused by factors such as command lag. The gradient at each moment in the DC current intensity sequence can be obtained by using the first derivative, which represents the rate of change of data at a single moment. The gradient threshold can be set between 5% and 10% of the rated current of the motor, and in the specific implementation of the present invention, it is set to 5%.

[0030] Preferably, in some implementations of the embodiments of the present invention, the method for constructing the time period to be analyzed includes: Centered on the motor start-up time, a time period of preset first length is divided forward, and a time period of preset second length is divided backward to obtain the time period to be analyzed; the time period of preset first length before the motor start-up time is the background noise segment, and the time period of preset second length after the motor start-up time is the step response segment. In a specific implementation of this invention, the first length is set to 0.2 seconds, and the second length is set to 0.5 seconds.

[0031] Preferably, in this embodiment of the invention, obtaining the current driving intensity and mechanical vibration intensity at each moment includes: The average value of the DC current intensity sequence in the background noise segment is used as the current baseline value, and the average value of the vibration intensity sequence in the background noise segment is used as the vibration baseline value. For each moment in the step response segment, the corresponding vibration intensity is subtracted from the vibration baseline value to obtain the mechanical vibration intensity. Considering that the subtraction result may be less than 0, the purpose of this embodiment of the invention is to quantify a net intensity. Therefore, if the mechanical vibration intensity is less than 0, the mechanical vibration intensity is assigned a value of 0.

[0032] Similarly, for each moment in the step response segment, the corresponding DC current intensity is subtracted from the current baseline value to obtain the current driving intensity; if the current driving intensity is less than 0, then the current driving intensity is assigned a value of 0. Finally, the information at each moment in the step response segment is statistically analyzed to obtain two current driving intensity sequences and a mechanical vibration intensity sequence of equal length.

[0033] Step S3: In the step response segment, perform iterative cumulative analysis on the current driving intensity and mechanical vibration intensity at each moment to obtain the current driving cumulative intensity signal and the mechanical vibration cumulative intensity signal respectively; perform point-by-point matching on the current driving cumulative intensity signal and the mechanical vibration cumulative intensity signal to obtain the best matching pair set.

[0034] In step S3, the aim is to determine each drive-response event by point-by-point matching of the two types of data. In step S2, the current drive intensity and mechanical vibration intensity obtained at each moment in the step response segment are instantaneous information, and the trend information within a single data set is not obvious. Therefore, to further reflect the changing trend in the data, iterative cumulative analysis is performed on the current drive intensity and mechanical vibration intensity at each moment in the step response segment, obtaining the current drive cumulative intensity signal and the mechanical vibration cumulative intensity signal respectively. Through iterative accumulation, the data at each moment is integrated with the preceding data. If there is a significant changing trend at a certain moment, such as a significant increasing trend, the cumulative intensity will be significantly greater than the cumulative intensity at the previous moment, thus making the data contain trend information. Point-by-point matching of the current drive cumulative intensity signal and the mechanical vibration cumulative intensity signal yields the optimal matching pair set. In the optimal matching pair combination, each optimal matching pair represents a current drive-mechanical vibration response event.

[0035] Taking the current-driven cumulative intensity signal as an example, for each moment, the sum of the current-driven intensity at this moment and the current-driven cumulative intensity at the previous moment is taken as the current-driven cumulative intensity at this moment; where the current-driven cumulative intensity at the initial moment is the current-driven intensity at the initial moment itself. Similarly, the mechanical vibration cumulative intensity signal can be obtained.

[0036] It should be noted that, to facilitate matching analysis and eliminate the influence of dimensions between the two dimensions, after obtaining the current-driven cumulative intensity signal and the mechanical vibration cumulative intensity signal, both signals need to be normalized. In this embodiment of the invention, the last element of the signal is selected as the maximum value element. By using the maximization normalization method, to prevent the denominator from being 0, the sum of the maximum value element and the preset hyperparameter can be used as the denominator, and each data point in the signal can be used as the numerator for normalization. The preset hyperparameter can be set to 0.1, and there is no specific limitation.

[0037] In one specific implementation of this invention, validity determination logic is further added. If the maximum value of the cumulative mechanical vibration intensity signal is lower than a preset validity threshold, the data acquisition is deemed invalid, indicating a possible fault in the sensor data acquisition process. Subsequent calculations are then terminated, and the operator is reminded to re-perform the test. The validity threshold can be set to 1% of the sensor's range.

[0038] Preferably, in some implementations of the present invention, considering that the current data and vibration data have already been sampled and aligned during the data processing stage, the lengths of the current-driven cumulative intensity signal and the mechanical vibration cumulative intensity signal are the same. Therefore, in the embodiments of the present invention, the normalized results of the current-driven cumulative intensity signal and the mechanical vibration cumulative intensity signal are used as the row dimension and column dimension, respectively, to construct a matching matrix. That is, the matching matrix is ​​a square matrix, with the current-driven cumulative intensity signal as the row and the mechanical vibration cumulative intensity signal as the column. The elements in the matrix... This represents the element value corresponding to the i-th current-driven cumulative intensity and the j-th mechanical vibration cumulative intensity. Each element in the matching matrix represents the difference between the corresponding normalized current-driven cumulative intensity and the normalized mechanical vibration cumulative intensity. In this embodiment of the invention, since both signals have been normalized, the absolute value of the difference between the corresponding data can be directly used as the corresponding element value in the matching matrix; the smaller the element value, the closer the cumulative intensities at the two times.

[0039] Furthermore, a dynamic time warping algorithm is adopted, using the Sakoe-Chiba band as a constraint window, to obtain the minimum cumulative distance path of the matching matrix according to the constraint conditions; The constraints are as follows: the path starts at the bottom left corner of the matching matrix and ends at the top right corner; the path can only move right, up, or to the upper right. These constraints allow us to find the event matching result that best matches the real physical scenario.

[0040] The normalized current-driven cumulative intensity and the normalized mechanical vibration cumulative intensity corresponding to each path point in the final minimum cumulative distance path constitute the best matching pair, and all path points constitute the set of the best matching pairs.

[0041] It should be noted that the dynamic time warping algorithm is a well-known technique in the field of science, and the specific algorithm content will not be elaborated or limited.

[0042] Step S4: Use the time stamp difference between the cumulative mechanical vibration intensity and the cumulative current drive intensity in the best-matched pair as the nonlinear hysteresis quantity, obtain the maximum nonlinear hysteresis quantity in the set of best-matched pairs, and take the time of the cumulative current drive intensity corresponding to the maximum nonlinear hysteresis quantity as the maximum hysteresis time.

[0043] In step S3, the point-by-point matching method, compared to directly linearly translating the two sequences, yields a nonlinear analysis result. This is because the idling acceleration of the mechanical system is nonlinear. Therefore, the matching result of the optimal matching pair obtained in step S3 represents the result after nonlinear alignment. Thus, the difference in timestamps between the cumulative mechanical vibration intensity and the cumulative current drive intensity in the optimal matching pair can be used as the nonlinear hysteresis, and the maximum nonlinear hysteresis in the set of optimal matching pairs can be statistically determined. The maximum nonlinear hysteresis is used as the detection basis for the current electric compressor, and the time of the current drive cumulative intensity corresponding to the maximum nonlinear hysteresis is taken as the maximum hysteresis time.

[0044] Preferably, in some implementations of the present invention, considering that both elements in the best matching pair are normalized, for each best matching pair, the timestamp of the normalized mechanical vibration cumulative intensity is subtracted from the timestamp of the normalized current drive cumulative intensity to obtain the nonlinear hysteresis. If the final maximum nonlinear hysteresis is greater than 0, it indicates that the mechanical response lags behind the electromagnetic drive, and the larger the value, the more obvious the hysteresis; if the maximum nonlinear hysteresis is less than 0, it indicates that the mechanical response is ahead, which is usually unlikely. If this situation exists, it may be due to the sensor installation polarity being reversed, requiring feedback to the staff to check the sensor settings and stop the subsequent processing steps.

[0045] Preferably, in some implementations of the present invention, considering that the waveform of the current-driven cumulative intensity signal can be regarded as an S-shaped waveform, and the lag exhibited in the beginning and end segments of the waveform is likely non-physical lag, the search is performed over a time interval of 10% to 90% when searching for the maximum nonlinear lag, reducing the computational load while improving data reliability. If multiple maximum nonlinear lags exist, the moment with the highest current-driven cumulative intensity is selected as the maximum lag moment.

[0046] Step S5: Backtrack and query the drive power of the electric compressor at the maximum lag time as the verification power, and combine it with the maximum nonlinear lag amount to perform fault diagnosis.

[0047] The maximum lag moment is the point at which mechanical lag is most severe. However, the maximum nonlinear lag at this moment alone cannot directly determine the fault, because when the motor starts under high load or high inductance conditions, the establishment of the magnetic field also affects rotor movement, resulting in significant lag. Therefore, to effectively distinguish this process, this embodiment of the invention further backtracks and queries the drive power of the electric compressor at the maximum lag moment as a verification power. The verification power directly reflects how much energy is injected into the motor stator windings at the maximum lag moment. If the verification power is high, it indicates that the motor is struggling to overcome a large back electromotive force or load resistance, and the lag at this time is a normal physical response. If the verification power is low, it indicates that the motor is almost powering, and the rotor is under light load. However, a large lag under light load can only mean that the rotor is idling, i.e., there is a gap in the drive train. Therefore, the maximum nonlinear lag and the verification power can be combined for effective fault diagnosis.

[0048] It should be noted that the power verification query can be directly exported from the monitoring platform, or it can be calculated using the prior power calculation formula under the condition of current information and electric compressor resistance at the known maximum lag time. The specifics will not be elaborated or limited.

[0049] Preferably, in this embodiment of the invention, the fault diagnosis method includes: If the maximum nonlinear hysteresis is less than or equal to the preset hysteresis threshold, it indicates that the mechanical response closely follows the electromagnetic drive, the hysteresis is within the normal range, there is no obvious fit gap inside the transmission chain, and it is judged as qualified. If the maximum nonlinear hysteresis is greater than the preset hysteresis threshold and the verification power is greater than or equal to the preset power threshold, it indicates that although there is a large response hysteresis, it is accompanied by high power input. This is consistent with the physical law that the magnetic field of a motor leads the mechanical action due to its inductive characteristics under high current excitation, which is a normal electrical response characteristic. Therefore, it is judged as a high-inductance qualified operating condition. If the maximum nonlinear hysteresis is greater than the preset hysteresis threshold and the verification power is less than the preset power threshold, it indicates that the electric compressor is experiencing a large hysteresis at low power. The motor is idling, but the mechanical end is slow to respond. This is a typical characteristic of a loose transmission chain, indicating that the rotor is undergoing an ineffective idling stroke, and is therefore determined to be a fault condition.

[0050] As a specific example, in this embodiment of the invention, the method for setting the preset hysteresis threshold and the preset power threshold includes: Statistically analyze the average maximum nonlinear hysteresis, standard deviation of maximum nonlinear hysteresis, average calibration power, and standard deviation of calibration power for several healthy electric compressors. By combining the average maximum nonlinear hysteresis and the standard deviation of the maximum nonlinear hysteresis, a preset hysteresis threshold is obtained using the three sigma principle; similarly, by combining the average verification power and the standard deviation of the verification power, a preset power threshold is obtained using the three sigma principle. The three sigma principle is a concept well-known to those skilled in the art, using the sum of the average value and three times the standard deviation as the final threshold result, without specifying any particular details or limitations.

[0051] It should be noted that the power threshold can also be set to 10% to 15% of the motor's rated power, or half of the minimum power value of a healthy electric compressor when starting under load. The specifics are not elaborated or limited, and can be set according to the implementation scenario.

[0052] In summary, this invention collects the DC current intensity sequence and vibration intensity sequence of an electric compressor. Within the analysis period, a step response segment and a background noise segment are defined. Baselines in each dimension of the data are removed from the step response segment. Iterative cumulative analysis is performed at each moment within the step response segment to obtain the current-driven cumulative intensity signal and the mechanical vibration cumulative intensity signal, respectively. The two signals are then matched point-by-point to determine the maximum nonlinear hysteresis. Using the maximum hysteresis time corresponding to the maximum nonlinear hysteresis as a benchmark, the driving power is back-checked as the verification power. The verification power and the maximum nonlinear hysteresis are used together as the fault judgment benchmark for fault diagnosis. This invention can effectively analyze electric compressor faults from two dimensions, using the verification power as a judgment constraint to avoid the influence of inductive operating conditions on the final fault diagnosis result, thus improving the detection accuracy of electric compressor faults.

[0053] Based on the same inventive concept, this invention also proposes an electric compressor fault diagnosis device based on multi-source data analysis, the device comprising: The electric compressor operation data acquisition module is used to acquire the DC current intensity sequence and vibration intensity sequence of the electric compressor. The data feature extraction module is used to construct the analysis period centered on the motor start-up time. The analysis period includes a background noise segment and a step response segment. In the step response segment, the DC current intensity sequence and vibration intensity sequence are processed respectively. For each moment, the baseline in the background noise segment is removed to obtain the current driving intensity and mechanical vibration intensity at each moment. The drive response coupling analysis module is used to perform iterative cumulative analysis at each time step in the step response segment to obtain the current drive cumulative intensity signal and the mechanical vibration cumulative intensity signal; the current drive cumulative intensity signal and the mechanical vibration cumulative intensity signal are matched point by point to obtain the best matching pair set; The maximum lag time determination module is used to take the difference between the timestamps of the cumulative mechanical vibration intensity and the cumulative current drive intensity in the best matching pair as the nonlinear lag amount, obtain the maximum nonlinear lag amount in the set of best matching pairs, and take the time of the cumulative current drive intensity corresponding to the maximum nonlinear lag amount as the maximum lag time. The fault diagnosis module is used to backtrack and query the drive power of the electric compressor at the maximum lag time as the verification power, and perform fault diagnosis in combination with the maximum nonlinear lag.

[0054] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0055] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A fault diagnosis method for electric compressors based on multi-source data analysis, characterized in that, The method includes: Obtain the DC current intensity sequence and vibration intensity sequence of the electric compressor; The analysis period is constructed with the motor start-up time as the center. The analysis period includes a background noise segment and a step response segment. In the step response segment, the DC current intensity sequence and the vibration intensity sequence are processed respectively. For each moment, the baseline in the background noise segment is removed to obtain the current driving intensity and mechanical vibration intensity at each moment. In the step response segment, the current driving intensity and mechanical vibration intensity at each moment are iteratively accumulated and analyzed to obtain the current driving intensity signal and the mechanical vibration intensity signal respectively; the current driving intensity signal and the mechanical vibration intensity signal are matched point by point to obtain the best matching pair set. The difference between the timestamps of the cumulative mechanical vibration intensity and the cumulative current drive intensity in the best-matched pair is used as the nonlinear hysteresis. The maximum nonlinear hysteresis in the set of best-matched pairs is obtained, and the time of the cumulative current drive intensity corresponding to the maximum nonlinear hysteresis is taken as the maximum hysteresis time. The driving power of the electric compressor at the maximum lag time is used as the verification power, and fault diagnosis is performed in combination with the maximum nonlinear lag.

2. The method for fault diagnosis of an electric compressor based on multi-source data analysis according to claim 1, characterized in that, The acquisition of the DC current intensity sequence and vibration intensity sequence of the electric compressor includes: Acquire the three-phase current signal and vibration acceleration signal of the electric compressor; perform DC processing on the three-phase current signal to obtain a DC current intensity sequence; perform envelope processing on the vibration acceleration signal, and use the obtained vibration envelope sequence as the vibration intensity sequence.

3. The method for fault diagnosis of an electric compressor based on multi-source data analysis according to claim 1, characterized in that, The method for determining the motor start-up time includes: If a start command is issued at a certain moment, and the gradient in the DC current intensity sequence is greater than a preset gradient threshold, then the motor is considered to have started.

4. The method for fault diagnosis of an electric compressor based on multi-source data analysis according to claim 1, characterized in that, The method for constructing the time period to be analyzed includes: Taking the motor start-up time as the center, a time period of a preset first length is divided forward, and a time period of a preset second length is divided backward to obtain the time period to be analyzed; the time period of the preset first length before the motor start-up time is the background noise segment, and the time period of the preset second length after the motor start-up time is the step response segment.

5. The method for fault diagnosis of an electric compressor based on multi-source data analysis according to claim 1, characterized in that, The process of obtaining the current driving intensity and mechanical vibration intensity at each moment includes: The average value of the DC current intensity sequence in the background noise segment is used as the current baseline value, and the average value of the vibration intensity sequence in the background noise segment is used as the vibration baseline value. For each moment in the step response segment, the corresponding vibration intensity is subtracted from the vibration baseline value to obtain the mechanical vibration intensity; if the mechanical vibration intensity is less than 0, then the mechanical vibration intensity is assigned a value of 0. For each moment in the step response segment, the current driving intensity is obtained by subtracting the current baseline value from the corresponding DC current intensity; if the current driving intensity is less than 0, the current driving intensity is assigned a value of 0.

6. The method for fault diagnosis of an electric compressor based on multi-source data analysis according to claim 1, characterized in that, After obtaining the DC current intensity sequence and the vibration intensity sequence, the following steps are also included: The DC current intensity sequence and the vibration intensity sequence are resampled and aligned.

7. The method for fault diagnosis of an electric compressor based on multi-source data analysis according to claim 6, characterized in that, The methods for obtaining the set of best matching pairs include: The normalized results of the current-driven cumulative intensity signal and the mechanical vibration cumulative intensity signal are used as the row dimension and column dimension, respectively, to construct a matching matrix. Each element in the matching matrix is ​​the difference between the corresponding normalized current-driven cumulative intensity and the normalized mechanical vibration cumulative intensity. Using the dynamic time warping algorithm and the Sakoe-Chiba band as the constraint window, the minimum cumulative distance path of the matching matrix is ​​obtained according to the constraint conditions. The constraints include: the path starts at the bottom left corner of the matching matrix and ends at the top right corner of the matching matrix; the path can only move right, up, or to the upper right. The normalized current-driven cumulative intensity and the normalized mechanical vibration cumulative intensity corresponding to each path point in the minimum cumulative distance path constitute the best matching pair, and all path points constitute the set of the best matching pairs.

8. The method for fault diagnosis of an electric compressor based on multi-source data analysis according to claim 7, characterized in that, The method for obtaining the nonlinear hysteresis includes: For each optimal match, the nonlinear hysteresis is obtained by subtracting the timestamp of the normalized cumulative mechanical vibration intensity from the timestamp of the normalized cumulative current drive intensity.

9. The method for fault diagnosis of an electric compressor based on multi-source data analysis according to claim 1, characterized in that, The fault diagnosis method includes: If the maximum nonlinear hysteresis is less than or equal to the preset hysteresis threshold, it is deemed qualified. If the maximum nonlinear hysteresis is greater than the preset hysteresis threshold and the verification power is greater than or equal to the preset power threshold, then it is determined to be a high-sensitivity qualified operating condition. If the maximum nonlinear hysteresis is greater than a preset hysteresis threshold and the verification power is less than a preset power threshold, then it is determined to be a fault condition.

10. A fault diagnosis device for an electric compressor based on multi-source data analysis, characterized in that, The device includes: The electric compressor operation data acquisition module is used to acquire the DC current intensity sequence and vibration intensity sequence of the electric compressor. The data feature extraction module is used to construct the analysis period centered on the motor start-up time. The analysis period includes a background noise segment and a step response segment. In the step response segment, the DC current intensity sequence and vibration intensity sequence are processed respectively. For each moment, the baseline in the background noise segment is removed to obtain the current driving intensity and mechanical vibration intensity at each moment. The drive response coupling analysis module is used to perform iterative cumulative analysis at each time step in the step response segment to obtain the current drive cumulative intensity signal and the mechanical vibration cumulative intensity signal; the current drive cumulative intensity signal and the mechanical vibration cumulative intensity signal are matched point by point to obtain the best matching pair set; The maximum lag time determination module is used to take the difference between the timestamps of the cumulative mechanical vibration intensity and the cumulative current drive intensity in the best matching pair as the nonlinear lag amount, obtain the maximum nonlinear lag amount in the set of best matching pairs, and take the time of the cumulative current drive intensity corresponding to the maximum nonlinear lag amount as the maximum lag time. The fault diagnosis module is used to backtrack and query the drive power of the electric compressor at the maximum lag time as the verification power, and perform fault diagnosis in combination with the maximum nonlinear lag.

Images