Multi-mode sensing and analyzing method for sensing abnormal operation of household air conditioner

Abstract

The invention relates to the technical field of state monitoring and fault diagnosis of an air conditioner compressor, and particularly discloses a multi-mode sensing analysis method for sensing abnormal operation of a household air conditioner. The compressor is controlled to execute V-shaped frequency change in the low-load or self-checking period, and an active excitation signal is generated; synchronously extracting energy change contours of the current and vibration signals, calculating a cross-correlation value, and comparing the cross-correlation value with a health reference to judge whether electromechanical coupling abnormity exists or not; and if the abnormity exists, abnormal characteristics are further analyzed to distinguish internal mechanical state degradation, external installation abnormity or abnormity to be checked, and corresponding alarms are triggered. Through active excitation and bimodal synchronous analysis, the ability to excite and capture early faults is enhanced, the fault identification precision is improved, the professional detection cost is reduced, and state monitoring is seamlessly integrated into daily operation of the air conditioner.

Description

Technical Field

[0001] This invention relates to the field of air conditioner compressor condition monitoring and fault diagnosis technology, and more specifically, to a multimodal sensing method for detecting and analyzing abnormal operation of a household air conditioner. Background Technology

[0002] As the core power component of an air conditioning system, the long-term stable operation of the air conditioning compressor is crucial to the system's reliability. Compressor failures typically originate from internal mechanical wear, loose components, or deterioration of electrical performance. These potential problems often do not cause obvious performance degradation or severe abnormal noises under stable operating conditions in their early stages, making them difficult to detect using traditional steady-state parameter-based monitoring methods. Currently, industry monitoring of compressor health mainly relies on threshold judgments of single physical signals, such as continuously monitoring whether the vibration amplitude exceeds a certain fixed limit, or analyzing the harmonic components of the steady-state operating current.

[0003] Research has found that prior art patent CN105137251B discloses a method, device, and system for detecting air conditioners and their faults, including: acquiring a first parameter and a second parameter, wherein the first parameter is the power grid parameter to which the air conditioner is connected, and the second parameter is the operating parameter of the air conditioner; detecting whether the air conditioner's compressor has activated its protection action, wherein the compressor's protection action is the action of cutting off the compressor's power supply; and, if the compressor's protection action is detected, sending the first parameter and the second parameter to a terminal device so that the terminal device can analyze the first parameter and the second parameter respectively. This invention solves the technical problem of low accuracy in detecting the cause of compressor malfunctions in prior art.

[0004] The shortcomings of existing technologies are as follows: First, they employ a passive, reactive approach, requiring data collection and analysis only after the compressor triggers a protective shutdown due to serious issues such as overvoltage. This makes it impossible to proactively predict and warn of early mechanical wear or performance degradation that is still in its nascent stages and has not yet caused a shutdown. Second, the diagnostic methods rely on simple static threshold judgments of power grid and air conditioner operating parameters. This method is not sensitive enough to dynamic anomalies caused by weakened electromechanical coupling, and its diagnostic conclusions can only roughly distinguish between power grid faults and air conditioner faults, failing to delve into the physical root cause of the fault and differentiate between internal mechanical wear and external installation problems. Finally, there is a lack of an online, non-invasive detection method that can adaptively adapt to the current state of the compressor, proactively stimulate potential fault characteristics, and automatically perform multi-information fusion diagnosis. Summary of the Invention

[0005] In view of this, in order to solve the problems mentioned in the background technology, a multimodal sensing method for detecting and analyzing abnormal operation of household air conditioners is proposed.

[0006] The objective of this invention can be achieved through the following technical solution: This invention provides a multimodal sensing method for detecting and analyzing abnormal operation of a household air conditioner, comprising: acquiring the original three-phase current signal and the original high-frequency vibration signal of the casing when the air conditioner compressor is running, filtering the original three-phase current signal and the original high-frequency vibration signal of the casing respectively, and generating preprocessed current signal and vibration signal.

[0007] While continuously generating preprocessed current and vibration signals, when preset triggering conditions are met, the air conditioning compressor is controlled to perform a preset frequency change operation, causing the operating frequency of the air conditioning compressor to undergo a non-steady-state change, generating an active excitation signal for anomaly analysis.

[0008] During the duration of the active excitation signal, the energy change profiles of the preprocessed current signal and the preprocessed vibration signal are extracted simultaneously to generate current envelope signals and vibration envelope signals, respectively.

[0009] The similarity between the current envelope signal and the vibration envelope signal on the time axis is calculated to generate quantified cross-correlation values.

[0010] The quantified cross-relation values ​​are compared with the pre-stored health benchmark correlation coefficients. Based on the comparison results, it is determined whether there is an electromechanical coupling abnormality in the air conditioner, and an abnormality determination result is generated.

[0011] If the anomaly determination result indicates the presence of electromechanical coupling anomaly, further analysis of the anomaly characteristics is conducted to distinguish the fault type, and corresponding alarm prompts are triggered, generating and outputting anomaly alarm information. The fault type is internal mechanical condition deterioration, external installation anomaly, or an anomaly to be investigated.

[0012] Compared with the prior art, the embodiments of the present invention have at least the following advantages or beneficial effects: (1) The present invention artificially creates a non-steady-state operation process by actively controlling the compressor to perform V-shaped frequency change operation during the low load or self-test period of the air conditioner. This active excitation process can amplify the response differences of internal mechanical components under dynamic load, so that the potential fault characteristics that are masked during steady-state operation are highlighted in the current and vibration signals. Compared with monitoring only under steady state, this method enhances the ability to excite and capture early electromechanical coupling anomalies.

[0013] (2) This invention constructs a three-in-one closed-loop diagnostic model of "excitation-response-transfer function" by simultaneously extracting and analyzing the dynamic response of current and vibration signals under active excitation. This invention no longer treats current and vibration as isolated signals, but precisely maps the former to the "instantaneous synthetic electromagnetic torque" of the drive system and the latter to the "instantaneous impact energy response" of the mechanical system. The "electromagnetic-mechanical dynamic coupling stiffness" inside the compressor is quantitatively evaluated by calculating the morphological correlation between the two. This diagnostic logic based on the physical essence of things improves the accuracy and robustness of identifying electromechanical coupling anomalies.

[0014] (3) After determining that an anomaly exists, this invention further analyzes the deviation of the cross-correlation coefficient and the maximum amplitude of the vibration envelope in parallel, and makes logical judgments based on multiple preset threshold combinations. This design enables the system to distinguish different manifestations of anomalies: if the correlation coefficient drops significantly while the vibration amplitude is normal, it indicates that the transmission path from electromagnetic force to mechanical vibration has deteriorated, pointing to an internal mechanical problem; if the correlation coefficient is normal while the vibration amplitude is abnormally high, it indicates that the vibration originates from external excitation, pointing to an installation or environmental problem. This achieves a preliminary distinction of anomaly types, providing directional information for subsequent maintenance.

[0015] (4) This invention defines the steps for learning and storing the correlation coefficient of the health benchmark and the vibration amplitude benchmark through the same process under the health condition of the equipment, and comparing the current detection results with these personalized benchmarks. At the same time, the entire detection process can be executed automatically without human intervention. This constructs a complete and adaptive online diagnostic closed loop, reduces the manpower and time costs of implementing periodic professional inspections, and enables condition monitoring to be seamlessly integrated into the daily operation of the air conditioner. Attached Figure Description

[0016] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments 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 schematic diagram of the method steps of the present invention.

[0018] Figure 2 This is a complete system workflow diagram of the present invention. Detailed Implementation

[0019] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0020] Example 1

[0021] like Figure 1 and Figure 2 Embodiment 1 of the present invention provides a multimodal sensing method for detecting and analyzing abnormal operation of a household air conditioner. It constructs a three-in-one closed-loop diagnostic system of "excitation-response-transfer function". The technical features are deeply interlocked in function, forming a strong coupling mechanism in which a change in one part affects the whole. Specifically, it includes: S1, acquiring the original three-phase current signal and the original high-frequency vibration signal of the casing of the air conditioner compressor during operation, filtering the original three-phase current signal and the original high-frequency vibration signal of the casing respectively, and generating pre-processed current signal and vibration signal.

[0022] In a specific embodiment of the present invention, the generation of preprocessed current signal and vibration signal includes: synchronously acquiring the current waveform of the three-phase drive circuit of the compressor as the original three-phase current signal through a current sensor, and performing anti-aliasing filtering processing using a low-pass filter to obtain the preprocessed current signal.

[0023] The vertical acceleration waveform is collected by a vibration sensor attached to the compressor housing as the original high-frequency vibration signal of the housing. The signal is then filtered by a bandpass filter to remove low-frequency vibration interference from the environment, resulting in a pre-processed vibration signal.

[0024] It should be noted that step S1 first involves deploying the current sensor and vibration sensor. The current sensor is non-contactly clamped to each phase of the three-phase drive circuit of the air conditioner compressor. When the compressor starts and runs, the current sensor synchronously acquires the current intensity waveform of each phase circuit as it changes over time. This waveform is the raw three-phase current signal, a data set containing three independent time series. Simultaneously, the vibration sensor is firmly attached to the central area of ​​the compressor housing's outer surface, with its sensitive axis perpendicular to the compressor's main shaft axis, primarily capturing radial vibration. The acceleration waveform acquired in this direction is the raw high-frequency vibration signal of the housing.

[0025] Next, signal preprocessing is performed. For each single-phase time series in the acquired three-phase current raw signal, anti-aliasing filtering is applied using a low-pass filter with a cutoff frequency of 500 Hz to remove high-frequency noise components that could distort subsequent analysis. The filtering of the raw high-frequency vibration signal from the casing is not a simple broadband filter; instead, multiple narrow-bandpass filter banks are pre-set based on the physical model of the resonant frequencies of typical internal compressor components (such as valve plates, bearings, and pistons). For example, the passband frequency range is set to 300 Hz to 5 kHz. This design aims to retain the high-frequency vibration characteristic signals generated by the internal mechanical motion of the compressor, while acting like a tuning knob to focus subsequent analysis on specific potential fault source frequency bands. After these two independent filtering operations, two processed signals are obtained for subsequent analysis: the preprocessed current signal and the preprocessed vibration signal.

[0026] The three-phase current raw signal refers to a sequence of three analog or digital signals output by a current sensor, reflecting the continuous change of the instantaneous current value in the compressor's three-phase windings over time. Its characteristic attribute is multi-channel time-series data, with strict time synchronization of the three phases. The setting is based on the monitoring requirements of industrial electrical equipment standards, used to comprehensively reflect the motor drive status. The casing high-frequency vibration raw signal refers to a sequence of analog or digital signals output by a vibration sensor, reflecting the continuous change of the instantaneous acceleration value of the compressor casing surface in the vertical direction over time. Its characteristic attribute is single-channel time-series data, with the direction set to vertical based on engineering practice that is most sensitive to the axial vibration of the compressor's main shaft. Anti-aliasing filtering is a signal preprocessing operation. Its function is to limit the signal bandwidth before or after digital sampling of the continuous-time signal using a low-pass filter with a cutoff frequency of 500 Hz, preventing high-frequency components from aliasing into the low-frequency band after sampling and causing distortion. This cutoff frequency is set at least 10 times the highest operating frequency of the compressor drive current's fundamental frequency to ensure that all relevant harmonic components are preserved. Bandpass filtering is a signal preprocessing operation that selectively retains signal components within a passband frequency range of 300 Hz to 5 kHz. The lower cutoff frequency of 300 Hz is set to effectively filter out the main low-frequency energy of common environmental and building structural vibrations, while the upper cutoff frequency of 5 kHz is set based on the typical range of high-frequency vibration components that may be generated by mechanical shock and friction in compressors.

[0027] For example, taking a compressor with a rated frequency of 50 Hz as an example, step S1 is executed at a certain operating moment. The current sensor synchronously collects three-phase current at a sampling rate of 1000 points per second for 1 second, obtaining three sets of raw current signals, each containing 1000 data points, with current values ​​ranging between ±30 amperes. The vibration sensor collects vertical acceleration at a sampling rate of 10000 points per second for 1 second, obtaining a set of raw vibration signals containing 10000 data points, with acceleration values ​​ranging between ±50 times the force of gravity. After applying an anti-aliasing low-pass filter with a cutoff frequency of 500 Hz to the raw current signal, minor spikes above 500 Hz in the signal are smoothed, and the waveform is clearer. After applying a bandpass filter with a passband of 300 Hz to 5 kHz to the raw vibration signal, a strong environmental background vibration at a frequency of 25 Hz is completely suppressed, while a component suspected to be compressor valve plate impact vibration at a frequency of 1.2 kHz is completely preserved. These two sets of filtered signals constitute the preprocessed current signal and vibration signal, which can be directly used for subsequent analysis.

[0028] S2. While continuously generating preprocessed current and vibration signals, when the preset triggering conditions are met, control the air conditioning compressor to perform a preset frequency change operation, causing the operating frequency of the air conditioning compressor to undergo a non-steady-state change, generating an active excitation signal for anomaly analysis.

[0029] In a specific embodiment of the present invention, the preset trigger condition for controlling the air conditioner compressor to perform a preset frequency change operation is: continuously monitoring the operating status of the air conditioner, and when the air conditioner is detected to be in a low-load operating state or entering a user-defined self-test period, reading the current operating frequency of the compressor and triggering the frequency change operation, wherein the low-load operating state is when the current operating frequency of the compressor is continuously lower than half of its rated frequency.

[0030] In a specific embodiment of the present invention, generating an active excitation signal for anomaly analysis includes: sending a control command to the compressor's motor driver to control the compressor frequency to linearly decrease from the current operating frequency to a first frequency within a preset time window, and then linearly increase from the first frequency to a second frequency, forming a V-shaped frequency change process. The non-steady-state operating state of the compressor generated by the V-shaped frequency change process constitutes the active excitation signal.

[0031] It should be noted that step S2 first determines the timing of active excitation. The system continuously monitors the air conditioner's operating status. When it detects that the air conditioner is operating under low load, such as when the compressor operates at 30% below its rated power for more than 5 minutes, or when the system enters a self-test period preset by the user via remote control or panel, step S2 is triggered. Next, the system reads the current output frequency of the inverter driving the compressor through the internal bus; this value is the compressor's current operating frequency. Then, the system sends a set of preset digital control commands to the compressor's motor driver. These commands cause the driver to control the compressor's operating frequency to change specifically within a preset time window: the frequency first starts from the current operating frequency, then linearly decreases to a lower set value, i.e., the first frequency, within the first time period.

[0032] Subsequently, in the following second time period, the frequency linearly rises from the first frequency to a higher set value, namely the second frequency. The second frequency can be equal to or slightly higher than the initial current operating frequency. The entire frequency change trajectory over time exhibits a continuous V-shaped profile, first decreasing and then increasing. This artificially created, rapid frequency change process forces the compressor to leave steady-state operation and enter a non-steady-state operating state where speed and torque are constantly adjusted. During this dynamic change, the inertial and electromagnetic forces acting on the internal mechanical components of the compressor undergo transient changes. These changes act as active excitation signals that trigger potential mechanical fault characteristics, simultaneously affecting the compressor's electrical drive system and mechanical structure, and ultimately reflected in the subsequently continuously acquired pre-processed current and vibration signals.

[0033] Low-load operation refers to a condition where the compressor output power required by the air conditioning system to meet cooling or heating needs is significantly lower than its maximum capacity. The criterion is that the compressor's current operating frequency is consistently below 50% of its rated frequency. This threshold is set based on engineering experience that air conditioners operate more smoothly under partial load, making them suitable for insertion testing. The user-defined self-test period refers to a silent testing period initiated or scheduled by the user through the device interface. Its function is to allow the system to run the testing program independently when cooling or heating is not required. The current operating frequency refers to the actual output frequency value of the compressor motor driver at the moment step S2 is triggered, measured in Hertz (Hz). It is set based on the required frequency calculated by the air conditioning control system according to real-time operating conditions. The first frequency is a preset target frequency value lower than the current operating frequency. Its setting is based on causing a sufficiently significant speed reduction in the compressor to stimulate a dynamic response; it is typically set 10 to 20 Hz lower than the current operating frequency. The second frequency is a preset target frequency value higher than the first frequency. Its setting is based on causing the compressor to undergo a complete deceleration and acceleration cycle; it is typically set equal to or slightly higher than the initial current operating frequency. The V-shaped frequency change process refers to a complete control trajectory in which the compressor's operating frequency first decreases monotonically over time and then increases monotonically, with the change being continuous and without abrupt changes. Its function is to construct a symmetrical or asymmetrical dynamic process as a standard test excitation. Unsteady-state operation refers to a physical state in which parameters such as compressor speed, current, and torque are in a continuous process of change, as opposed to a stable constant-speed operation state.

[0034] For example, following the example from step S1, after the system determines that it has entered a low-load operating state, it reads that the current operating frequency of the compressor is 45 Hz. The system then sends a control command to the motor driver, requiring it to complete a V-shaped frequency change within a total duration of 6 seconds: the command causes the compressor frequency to linearly decrease from 45 Hz to a first frequency of 30 Hz within 2 seconds, then linearly increase from 30 Hz to a second frequency of 45 Hz within 2 seconds, and finally stabilize at 45 Hz in the last 2 seconds. This entire 6-second dynamic change process constitutes the active excitation signal used to stimulate fault characteristics. During this process, the pre-processed current signal and vibration signal are continuously acquired, and their waveforms will reflect the dynamic characteristics brought about by the frequency change.

[0035] S3. During the duration of the active excitation signal, the energy change profiles of the preprocessed current signal and the preprocessed vibration signal are extracted simultaneously to generate current envelope signals and vibration envelope signals, respectively.

[0036] In a specific embodiment of the present invention, generating the current envelope signal and the vibration envelope signal respectively includes: transforming the preprocessed three-phase current signal into current components in a two-phase orthogonal stationary coordinate system in real time through Clark transform, calculating the instantaneous magnitude of the current space vector, and generating a low-phase-delay current envelope signal that directly reflects the magnitude of the instantaneous synthesized electromagnetic torque.

[0037] The preprocessed vibration signal is subjected to Hilbert transform to construct an analytic signal, and the instantaneous amplitude of the analytic signal is calculated to generate a vibration envelope signal with low phase delay that can be accurately aligned with real mechanical impact events.

[0038] It should be noted that this step is not simply about acquiring the energy profiles of two signals in parallel, but rather requires strict synchronization at the microsecond level to ensure that the analyzed current and vibration data originate from the same physical excitation event on the time axis. This process specifically includes:

[0039] For current signals, the instantaneous electromagnetic torque vector magnitude is generated as the current envelope signal: this method abandons the ambiguity and delay inherent in traditional low-pass filtering after taking the absolute value. First, the synchronously acquired instantaneous three-phase current values ​​are transformed using Clarke transform. , , Real-time transformation to two orthogonal stationary coordinate systems and Then, calculate the instantaneous magnitude of the current space vector. .this This refers to the current envelope signal, whose physical meaning is no longer abstract energy, but directly proportional to the magnitude of the instantaneous synthesized electromagnetic torque driving the compressor rotor. This is a scalar time series with clear physical meaning, low phase delay, and real-time reflection of the driving force, which greatly improves the fidelity and time resolution of the input signal.

[0040] For vibration signals, an instantaneous impact energy envelope based on Hilbert transform is generated as the vibration envelope signal: this method abandons the full-wave rectification + low-pass filtering method that introduces significant group delay. First, the high-frequency vibration signal after band-selective filtering in step S1 is... Perform a Hilbert transform to obtain its orthogonal components. Construct analytical signals Then, calculate the instantaneous amplitude of the analytic signal. .this This refers to the vibration envelope signal, which represents the instantaneous magnitude of vibrational energy excited by internal impact or friction within a specific mechanical frequency band. This method completely eliminates the group delay introduced by filters, ensuring that the peaks and valleys of the envelope waveform are strictly aligned with the actual mechanical impact event on the time axis, providing a high-fidelity data foundation for capturing the microsecond-level morphological mismatch between driving force and mechanical response.

[0041] S4. Calculate the morphological similarity between the current envelope signal and the vibration envelope signal on the time axis, and generate quantified cross-correlation values.

[0042] In a specific embodiment of the present invention, the calculation of the similarity of the current envelope signal and the vibration envelope signal on the time axis to generate quantified cross-relationship values ​​includes: on the time axis, taking the triggering time of the active excitation signal as a reference, aligning the current envelope signal reflecting the instantaneous synthesized electromagnetic torque and the vibration envelope signal reflecting the instantaneous impact energy.

[0043] Extract the current envelope data sequence and vibration envelope data sequence within the same time period from the start of the frequency change operation to the frequency recovery and stabilization.

[0044] The Pearson correlation coefficient was used to calculate the degree of linear correlation between the current envelope data sequence and the vibration envelope data sequence. The calculation results were used as numerical values ​​to quantitatively evaluate the integrity of the electromagnetic-mechanical dynamic transmission coordination of the compressor.

[0045] It should be noted that the core of this step is to verify the physical causal chain, that is, to calculate the morphological similarity between the instantaneous synthesized electromagnetic torque (current envelope signal) generated in step S3 and the instantaneous mechanical impact energy (vibration envelope signal). The system reads the time series of the current envelope signal and the vibration envelope signal generated in step S3. It identifies the absolute timestamps at the start of data acquisition for these two signal sequences, and uses the start time point of the current envelope signal as a reference. Through a translation method in digital signal processing, the vibration envelope signal is moved on the time axis so that its starting point is precisely aligned with the starting point of the current envelope signal, ensuring that the two signals are completely synchronized in the time dimension.

[0046] Next, data extraction is performed. Based on the start and end times of the V-shaped frequency change of the active excitation signal recorded in step S2, a specific time window is determined on the time axis. From the two aligned envelope signals, data segments within the same time period, from the start of the frequency decrease to the frequency stabilization, are extracted, resulting in current envelope data sequences and vibration envelope data sequences, each containing the same number of data points. Finally, the linear correlation between the extracted current envelope data sequences and vibration envelope data sequences is calculated using the Pearson correlation coefficient formula. The formula is: .in, This represents the final generated cross-relationship value, and its value range is... . Represents the current envelope sequence. This represents the vibrational envelope sequence. Represents the current envelope sequence With vibrational envelope sequence The covariance. Represents the current envelope sequence The standard deviation. Represents the vibration envelope sequence The standard deviation is calculated as follows: First, calculate the mean of each of the two sequences; then, calculate the deviation of each data point in each sequence from its mean, and calculate the average of the products of the two deviation sequences to obtain the covariance; simultaneously, calculate the square root of the average of the squares of the deviations of the two sequences to obtain their respective standard deviations.

[0047] Finally, the covariance is divided by the product of the two standard deviations to obtain the Pearson correlation coefficient. .this The value represents the quantified cross-correlation value, which quantitatively assesses the integrity of the electromagnetic-mechanical dynamic transmission coordination. A Pearson correlation coefficient close to 1 indicates that the input electromagnetic torque is efficiently and linearly converted into the expected mechanical vibration response; conversely, a decrease in the value indicates abnormalities such as "softening" or "nonlinearity" in the transmission chain.

[0048] It should also be noted that the electromagnetic-mechanical dynamic coordination degree transfer characteristic is used to describe the dynamic relationship between the input (electromagnetic torque) and output (casing vibration energy) of the compressor drive system during non-steady-state operation. To quantify this characteristic, this invention defines a current envelope sequence. With vibrational envelope sequence Pearson correlation coefficient during the active incentive period This serves as an integrity evaluation metric for the transmission characteristics. Under ideal, tightly coupled, healthy conditions, changes in electromagnetic torque will instantly and proportionally induce a vibration response. The value approaches 1; when the transmission path has softening, gaps, or nonlinearity, The value decreased. Therefore, the correlation coefficient... It essentially serves as the output value of the equivalent transfer function for fault diagnosis as defined in this invention.

[0049] Alignment on the time axis refers to adjusting the start times of two signal sequences to the same time reference point. Its function is to eliminate errors caused by small initial time differences that may exist in the sensors or acquisition channels, which can affect subsequent correlation analysis. The same time period refers to a completely consistent time interval defined by the point where the frequency of the active excitation signal begins to change and the point where the frequency recovers and stabilizes. This is set to ensure that the comparison is made based on the responses under the same physical excitation process. The current envelope data sequence refers to a set of discrete data points extracted from the aligned current envelope signal, corresponding to a segment within the active excitation time period. The vibration envelope data sequence refers to a set of discrete data points extracted from the aligned vibration envelope signal, corresponding to a segment within the same active excitation time period.

[0050] S5. Compare the quantified cross-relation values ​​with the pre-stored health benchmark correlation coefficients, determine whether there is an electromechanical coupling abnormality in the air conditioner based on the comparison results, and generate an abnormality determination result.

[0051] In a specific embodiment of the present invention, the pre-stored health baseline correlation coefficient is stored in a non-volatile memory, and the health baseline correlation coefficient is obtained by averaging the cross-correlation values ​​calculated multiple times under the same active excitation conditions during the air conditioner's factory commissioning or initial installation learning phase.

[0052] It should be noted that the system accesses the non-volatile memory inside the air conditioner controller and reads the pre-stored health baseline correlation coefficient. This health baseline correlation coefficient is obtained by repeatedly executing steps S1 to S4 during the air conditioner's factory commissioning phase or the learning phase after initial installation, under the condition that the compressor and the whole unit are confirmed to be fault-free. That is, applying the same active excitation signal and calculating multiple cross-correlation values, and then taking the arithmetic mean of the multiple cross-correlation values.

[0053] It should be further explained that the specific process for establishing a health baseline during the learning phase after the initial installation is as follows: After the air conditioner is installed and confirmed to be operating normally, the user or installer initiates the baseline learning mode via remote control. The system will automatically capture at least five instances of low-load operation within the next 72 hours. During each instance, steps S1 to S4 are executed once, recording the calculated cross-correlation coefficients and the maximum amplitude of the vibration envelope. After the learning phase, the system processes the five sets of data: First, the arithmetic mean of the cross-correlation coefficients and the maximum amplitude of the vibration envelope is calculated; then, the deviation of each cross-correlation coefficient from the arithmetic mean is calculated. If the maximum deviation exceeds 20% of the average, the set of data with the largest deviation is discarded, and the average of the remaining data is recalculated. Finally, the cross-correlation values ​​and the maximum amplitude of the vibration envelope obtained after screening and averaging are used as the health baseline correlation coefficient and health baseline amplitude of the air conditioner, respectively, and permanently stored in non-volatile memory for use as subsequent preset values. This process ensures the representativeness and stability of the baseline values.

[0054] In a specific embodiment of the present invention, the step of determining whether the air conditioner has an electromechanical coupling abnormality based on the comparison result and generating an abnormality determination result includes: calculating the percentage deviation of the current mutual relationship value from the correlation coefficient of the health benchmark.

[0055] The deviation percentage is compared with a first preset threshold. If the deviation percentage is greater than or equal to the first preset threshold, it is determined that the air conditioner has an electromechanical coupling abnormality, and an abnormal judgment result with abnormal content is generated. If the deviation percentage is less than the first preset threshold, it is determined that the air conditioner does not have an electromechanical coupling abnormality, and an abnormal judgment result with normal content is generated.

[0056] It should be noted that, next, the system compares the current cross-correlation value calculated in step S4 with the health baseline correlation coefficient read from memory. The percentage deviation of the current cross-correlation value from the health baseline correlation coefficient is then calculated. The calculation formula is: .in, This represents the current cross-relationship value generated in step S4. This represents the correlation coefficient with the health baseline read from memory, with the symbol "||" indicating the absolute value. The numerator of this formula is the absolute difference between the current value and the baseline value, and the denominator is the baseline value. The result is a dimensionless percentage value used to quantify the deviation between the current state and the healthy state. Finally, a logical judgment is made based on a threshold. The calculated deviation percentage is compared with a first preset threshold. If the calculated deviation percentage is greater than or equal to the first preset threshold, the logic judgment unit outputs a judgment result of "abnormal," indicating that an electromechanical coupling abnormality has been detected. If the calculated deviation percentage is less than the first preset threshold, the logic judgment unit outputs a judgment result of "normal."

[0057] It should be further explained that the first preset threshold is set based on the following: by performing the aforementioned benchmark learning on a large number of healthy compressors of the same model and batch, the natural fluctuation range of the correlation coefficient of the health benchmark of all devices in long-term (e.g., within one year) repeated testing is statistically analyzed. Statistics show that the fluctuation range of the cross-correlation coefficient repeated testing value of more than 95% of healthy devices does not exceed 12% of its own health benchmark value. Therefore, in order to effectively capture anomalies while controlling the false alarm rate, the first preset threshold is set to 15%. This threshold can be fine-tuned according to the statistical characteristics of specific models, but its setting principle is that it is greater than the upper limit of natural fluctuation in a healthy state.

[0058] It should also be noted that this step is essentially a non-invasive online quantitative assessment of the electromagnetic-mechanical dynamic coupling stiffness inside the compressor. When this stiffness decreases (i.e., the currently calculated cross-correlation coefficient deviates from the healthy baseline), it indicates a potential fault in the transmission chain from electromagnetic force to effective mechanical motion. This assessment capability is a novel function resulting from the strong coupling of active excitation and dual-modal synchronous coherent analysis, and is not a simple superposition of the functions of each part.

[0059] The health baseline correlation coefficient refers to a specific correlation coefficient value stored in non-volatile memory, representing the degree of electromechanical coupling of the air conditioner in a healthy state. Its characteristic attribute is a dimensionless scalar constant, and its function is to serve as a comparison benchmark for all subsequent online detection results. Non-volatile memory refers to a storage chip that can retain data even after power failure, and its function is to permanently store health baseline data. The first preset threshold is a pre-set percentage threshold value used to determine abnormalities. Its characteristic attribute is a fixed percentage constant, and its function is to serve as a decision boundary distinguishing between normal and abnormal states.

[0060] S6. If the anomaly determination result is that there is an electromechanical coupling anomaly, then further analyze the anomaly characteristics to distinguish the fault type, trigger the corresponding alarm prompt, generate and output the anomaly alarm information, wherein the fault type is internal mechanical condition deterioration, external installation anomaly, or an anomaly to be investigated.

[0061] In a specific embodiment of the present invention, the further analysis of abnormal features to distinguish fault types includes: when the abnormality determination result is that there is an electromechanical coupling abnormality, obtaining the maximum amplitude of the vibration envelope signal during the active excitation signal, and comparing the maximum amplitude with a preset health benchmark amplitude.

[0062] If the percentage deviation of the cross-correlation values ​​is greater than or equal to the second preset threshold, and the percentage deviation of the maximum value is less than the third preset threshold, then the abnormality is determined to be the deterioration of the internal mechanical state, the physical basis of which is the softening of the electromagnetic-mechanical dynamic coupling stiffness of the compressor.

[0063] If the percentage deviation of the cross-correlation values ​​is less than the second preset threshold, but the percentage deviation of the maximum amplitude is greater than or equal to the fourth preset threshold, then the abnormality is determined to be an external installation abnormality. The physical basis for this is that the electromagnetic-mechanical dynamic coupling stiffness remains intact, but there is an external excitation source unrelated to the driving force.

[0064] If the percentage deviation of the cross-correlation values ​​is greater than or equal to the first preset threshold and less than the second preset threshold, and the percentage deviation of the maximum value is less than the fourth preset threshold, or if the percentage deviation of the cross-correlation values ​​is greater than or equal to the second preset threshold and the percentage deviation of the maximum value is greater than or equal to the third preset threshold, then the abnormality is determined to be an abnormality to be investigated.

[0065] It should be noted that, firstly, while retrieving the health baseline correlation coefficient, the system also reads the amplitude baseline of the vibration envelope signal in a healthy state from non-volatile memory, i.e., the health baseline amplitude. This health baseline amplitude is obtained by averaging the maximum amplitude values ​​of the vibration envelope signal recorded during multiple active excitation processes in the same factory debugging or learning phase as when the health baseline correlation coefficient was obtained. Next, the system analyzes the vibration envelope signal recorded during the same active excitation process that triggered this anomaly determination, and finds its maximum amplitude over the entire V-shaped frequency change time period. Then, the system executes parallel logic judgments.

[0066] 1) Internal Mechanical Condition Deterioration: When the deviation percentage is greater than or equal to the second preset threshold, and the deviation percentage of the maximum amplitude is less than the third preset threshold, the fault type is determined to be internal mechanical condition deterioration. The physical basis is that the decrease in the cross-correlation coefficient indicates that the instantaneous synthesized electromagnetic torque has not been effectively transmitted to the mechanical structure, that is, the electromagnetic-mechanical dynamic coupling stiffness has softened. This strongly points to the performance degradation of internal moving parts due to wear, fatigue, or increased fit clearance. 2) External Installation Abnormality: When the deviation percentage is less than the second preset threshold (but still greater than or equal to the first preset threshold for determining abnormality), and the deviation percentage of the maximum amplitude is greater than or equal to the fourth preset threshold, the fault type is determined to be external installation abnormality. The physical basis is that the electromagnetic-mechanical coupling stiffness remains intact (cross-correlation coefficient is normal), but the vibration energy is abnormally increased, indicating the existence of an external vibration source unrelated to electromagnetic drive. This strongly points to abnormal vibration caused by loose air conditioner outdoor unit mounting bracket, unstable base, or external strong winds or impacts. 3) Anomalies to be Investigated: To ensure the integrity of the logic and cover all abnormal ranges, this invention also defines categories of anomalies to be investigated. When the abnormal characteristics do not meet the two clearly defined fault modes mentioned above, they are classified into this category. Specifically, the fault type is determined to be an anomaly to be investigated when either of the following conditions is met: a) When the deviation percentage is in an intermediate range, that is, its value is greater than or equal to the first preset threshold but less than the second preset threshold, and the deviation percentage of the maximum amplitude value has not reached a high level, that is, less than the fourth preset threshold. b) When both the deviation percentage and the vibration energy amplitude deviation are at a high level, that is, the deviation percentage is greater than or equal to the second preset threshold, and the deviation percentage of the maximum amplitude value is also greater than or equal to the third preset threshold. This situation indicates that the system state has seriously deteriorated, and both the signal morphology and vibration energy show extreme anomalies. It may involve multiple concurrent faults, and the situation is complex. Therefore, it is classified as an anomaly to be investigated.

[0067] For example, in a specific case, there are three types of faults: internal mechanical condition deterioration (Type A), external installation abnormality (Type B), and anomaly to be investigated (Type C).

[0068] The determination criteria are as follows (let R_dev be the percentage deviation of the cross-correlation values, and A_dev be the percentage deviation of the maximum amplitude):

[0069] Class A (Internal mechanical condition deterioration): R_dev >= the second preset threshold (35%) and A_dev < the third preset threshold (25%).

[0070] Category B (External Installation Exception): R_dev < Second Preset Threshold (35%) and A_dev >= Fourth Preset Threshold (70%).

[0071] Category C (Anomalies to be Detected): Condition C1: First preset threshold (15%) <= R_dev < Second preset threshold (35%) and A_dev < Fourth preset threshold (70%). Or Condition C2: R_dev >= Second preset threshold (35%) and A_dev >= Third preset threshold (25%).

[0072] It should be further explained that the setting of the second, third, and fourth preset thresholds is based on experimental analysis and statistics of typical failure modes: 1. For internal mechanical condition deterioration: In a laboratory environment, different degrees of bearing wear failure were artificially set on multiple compressors, and this detection method was executed. Experimental data shows that when the wear reaches an identifiable level, the percentage deviation of its cross-correlation coefficient usually decreases significantly (≥35%). However, since the fault source is internal, the efficiency of vibration energy transmission to the casing may not increase significantly due to coupling deterioration, and its vibration amplitude deviation percentage is usually <25%. Therefore, the second preset threshold is set to 35%, and the third preset threshold is set to 25%. 2. For external installation abnormalities: In an experimental environment, a loose installation condition was simulated. Experimental data shows that at this time, the electromagnetic-mechanical transmission path may be basically intact, and the percentage deviation of the cross-correlation coefficient is usually small (<35%). However, external impact and resonance cause the casing vibration amplitude to increase sharply, and its deviation percentage is usually ≥70%. Therefore, the fourth preset threshold is set to 70%. The aforementioned thresholds (first preset threshold 15%, second preset threshold 35%, third preset threshold 25%, and fourth preset threshold 70%) together constitute a fault differentiation logic matrix based on experimental data, enabling the system to effectively differentiate the root causes of anomalies according to different combinations of signal characteristics.

[0073] In a specific embodiment of the present invention, triggering the corresponding alarm prompt and generating and outputting abnormal alarm information includes: when the abnormality is determined to be internal mechanical condition deterioration, generating a mechanical wear or looseness type alarm.

[0074] When the anomaly is determined to be an external installation anomaly, an alarm of the type of unstable installation or external impact is generated.

[0075] When an abnormality is determined to be an abnormality to be investigated, a manual intervention diagnostic alarm is generated.

[0076] The abnormality type and alarm type are encapsulated into abnormal alarm information, which is output through a combination of flashing indicator lights and beeping buzzers on the air conditioning panel.

[0077] It should be noted that the system encapsulates the aforementioned anomaly and alarm types into a structured set of anomaly alarm information. This anomaly alarm information is output through the hardware interface on the air conditioning control board: a specific red indicator light on the drive panel flashes continuously at a fixed frequency, while a buzzer emits a specific rhythmic beep. The combination of the flashing indicator light and the beeping buzzer constitutes the final physical alarm output.

[0078] The maximum amplitude refers to the global maximum value of the vibration envelope signal among all data points within the selected time period. Its characteristic attribute is a scalar value, with the unit consistent with the vibration signal, and its function is to characterize the peak intensity of the vibration under that excitation. The health reference amplitude is a reference constant stored in non-volatile memory, representing the maximum amplitude of the vibration envelope signal under healthy conditions. Its setting is based on the same correlation coefficient as the health reference, and is based on the statistical average of multiple health tests. The second preset threshold is a percentage threshold value used to determine whether the cross-correlation coefficient has "decreased significantly," for example, 35%. It is set based on the critical deviation level obtained from experimental statistics that can clearly indicate internal mechanical coupling failure. The third preset threshold is a lenient percentage threshold value used to determine whether the maximum amplitude of vibration "does not exceed" the health reference, for example, 25%. It is set based on allowing a certain normal fluctuation range for the vibration peak value under healthy conditions. The fourth preset threshold is a strict percentage threshold value used to determine whether the maximum amplitude of vibration "significantly exceeds the limit," for example, 70%. It is set based on the significant amplitude increase required to indicate a strong external impact or serious installation problems. Internal mechanical deterioration is a fault type conclusion, specifically referring to the performance degradation of the compressor's internal moving parts due to wear, fatigue, or increased clearance. External installation or environmental interference is a fault type conclusion, specifically referring to abnormal vibrations caused by loose mounting brackets, unstable bases, or external strong winds or impacts on the outdoor unit of the air conditioner.

[0079] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A method for sensing and analyzing operational anomalies in a multimodal sensing household air conditioner, characterized in that, Includes the following steps: The original three-phase current signal and the original high-frequency vibration signal of the casing are obtained when the air conditioner compressor is running. The original three-phase current signal and the original high-frequency vibration signal of the casing are filtered respectively to generate pre-processed current signal and vibration signal. While continuously generating pre-processed current and vibration signals, when the preset triggering conditions are met, the air conditioning compressor is controlled to perform a preset frequency change operation, causing the operating frequency of the air conditioning compressor to undergo a non-steady-state change, generating an active excitation signal for anomaly analysis. During the duration of the active excitation signal, the energy change profiles of the preprocessed current signal and the preprocessed vibration signal are extracted simultaneously to generate current envelope signals and vibration envelope signals, respectively. Calculate the morphological similarity between the current envelope signal and the vibration envelope signal on the time axis, and generate quantified cross-correlation values; The quantified cross-relation values ​​are compared with the pre-stored health benchmark correlation coefficients. Based on the comparison results, it is determined whether there is an electromechanical coupling abnormality in the air conditioner, and an abnormality determination result is generated. If the anomaly determination result indicates the presence of electromechanical coupling anomaly, further analysis of the anomaly characteristics is conducted to distinguish the fault type, and corresponding alarm prompts are triggered, generating and outputting anomaly alarm information. The fault type is internal mechanical condition deterioration, external installation anomaly, or an anomaly to be investigated.

2. The method for sensing and analyzing operational anomalies in a multimodal sensing household air conditioner according to claim 1, characterized in that, The generation of preprocessed current and vibration signals includes: The current waveform of the compressor's three-phase drive circuit is synchronously acquired by a current sensor as the raw three-phase current signal, and anti-aliasing filtering is performed using a low-pass filter to obtain the pre-processed current signal. The vertical acceleration waveform is collected by a vibration sensor attached to the compressor housing as the original high-frequency vibration signal of the housing. The signal is then filtered by a bandpass filter to remove low-frequency vibration interference from the environment, thus obtaining the pre-processed vibration signal.

3. The method for sensing and analyzing operational anomalies in a multimodal sensing household air conditioner according to claim 1, characterized in that, The preset trigger condition for controlling the air conditioner compressor to perform a preset frequency change operation is: continuously monitoring the operating status of the air conditioner, and when the air conditioner is detected to be in a low-load operating state or entering a user-set self-test period, reading the current operating frequency of the compressor and triggering the frequency change operation. The low-load operating state is when the current operating frequency of the compressor is continuously lower than half of its rated frequency.

4. The method for sensing and analyzing operational anomalies in a multimodal sensing household air conditioner according to claim 3, characterized in that, The generation of the active excitation signal for anomaly analysis includes: sending a control command to the compressor's motor driver to control the compressor frequency to linearly decrease from the current operating frequency to a first frequency within a preset time window, and then linearly increase from the first frequency to a second frequency, forming a V-shaped frequency change process. The non-steady-state operating state of the compressor generated by the V-shaped frequency change process constitutes the active excitation signal.

5. The method for sensing and analyzing operational anomalies in a multimodal sensing household air conditioner according to claim 1, characterized in that, The generation of current envelope signals and vibration envelope signals respectively includes: The preprocessed three-phase current signal is transformed into current components in a two-phase orthogonal stationary coordinate system in real time by Clarke transform. The instantaneous magnitude of the current space vector is calculated, and a current envelope signal with low phase delay that directly reflects the magnitude of the instantaneous synthesized electromagnetic torque is generated. The preprocessed vibration signal is subjected to Hilbert transform to construct an analytic signal, and the instantaneous amplitude of the analytic signal is calculated to generate a vibration envelope signal with low phase delay that can be accurately aligned with real mechanical impact events.

6. The method for sensing and analyzing operational anomalies in a multimodal sensing household air conditioner according to claim 1, characterized in that, The calculation of the similarity in shape between the current envelope signal and the vibration envelope signal on the time axis generates quantified cross-correlation values, including: On the time axis, based on the triggering time of the active excitation signal, the current envelope signal reflecting the instantaneous synthesized electromagnetic torque and the vibration envelope signal reflecting the instantaneous impact energy are aligned. Extract the current envelope data sequence and vibration envelope data sequence within the same time period from the start of the frequency change operation to the frequency recovery and stabilization. The Pearson correlation coefficient was used to calculate the degree of linear correlation between the current envelope data sequence and the vibration envelope data sequence. The calculation results were used as numerical values ​​to quantitatively evaluate the integrity of the electromagnetic-mechanical dynamic transmission coordination of the compressor.

7. The method for sensing and analyzing operational anomalies in a multimodal sensing household air conditioner according to claim 1, characterized in that, The pre-stored health baseline correlation coefficient is stored in non-volatile memory, and the health baseline correlation coefficient is obtained by averaging the cross-correlation values ​​calculated multiple times under the same active excitation conditions during the air conditioner's factory commissioning or initial installation learning phase.

8. The method for sensing and analyzing operational anomalies in a multimodal sensing household air conditioner according to claim 7, characterized in that, The process of determining whether the air conditioner has an electromechanical coupling abnormality based on the comparison results and generating an abnormality determination result includes: Calculate the percentage deviation of the current cross-correlation value from the health baseline correlation coefficient; The deviation percentage is compared with a first preset threshold. If the deviation percentage is greater than or equal to the first preset threshold, it is determined that the air conditioner has an electromechanical coupling abnormality, and an abnormal judgment result with abnormal content is generated. If the deviation percentage is less than the first preset threshold, it is determined that the air conditioner does not have an electromechanical coupling abnormality, and an abnormal judgment result with normal content is generated.

9. The method for sensing and analyzing operational anomalies in a multimodal sensing household air conditioner according to claim 8, characterized in that, The further analysis of abnormal features to distinguish fault types includes: When the anomaly determination result is that there is an electromechanical coupling anomaly, the maximum amplitude of the vibration envelope signal during the active excitation signal is obtained, and the maximum amplitude is compared with the preset health benchmark amplitude. If the percentage deviation of the cross-correlation values ​​is greater than or equal to the second preset threshold, and the percentage deviation of the maximum value is less than the third preset threshold, then the abnormality is determined to be the deterioration of the internal mechanical state, the physical basis of which is the softening of the electromagnetic-mechanical dynamic coupling stiffness of the compressor. If the percentage deviation of the cross-correlation values ​​is less than the second preset threshold, but the percentage deviation of the maximum value is greater than or equal to the fourth preset threshold, then the abnormality is determined to be an external installation abnormality. The physical basis is that the electromagnetic-mechanical dynamic coupling stiffness is intact, but there is an external excitation source unrelated to the driving force. If the percentage deviation of the cross-correlation values ​​is greater than or equal to the first preset threshold and less than the second preset threshold, and the percentage deviation of the maximum value is less than the fourth preset threshold, or if the percentage deviation of the cross-correlation values ​​is greater than or equal to the second preset threshold and the percentage deviation of the maximum value is greater than or equal to the third preset threshold, then the abnormality is determined to be an abnormality to be investigated.

10. The method for sensing and analyzing operational anomalies in a multimodal sensing household air conditioner according to claim 9, characterized in that, The triggering of the corresponding alarm prompt, generating and outputting abnormal alarm information includes: When the abnormality is determined to be a deterioration of the internal mechanical condition, an alarm of mechanical wear or loosening type is generated. When the anomaly is determined to be an external installation anomaly, an alarm of the type of unstable installation or external impact is generated. When an abnormality is determined to be an abnormality to be investigated, a manual intervention diagnostic alarm is generated. The abnormality type and alarm type are encapsulated into abnormal alarm information, which is output through a combination of flashing indicator lights and beeping buzzers on the air conditioning panel.

Images