Air conditioner compressor fault diagnosis method and system based on acoustic feature recognition

Abstract

The invention provides an air conditioner compressor fault diagnosis method and system based on acoustic feature recognition. The method comprises the following steps: firstly, acquiring an operation sound signal in the operation process of an air conditioner compressor, then, identifying an acoustic characteristic mode related to a compressor fault based on the operation sound signal, then, matching the identified acoustic characteristic mode with a pre-stored compressor fault characteristic library to obtain a matching result, and finally, according to the matching result, determining the fault of the compressor according to the matching result. Outputting a real-time fault diagnosis report of the air conditioner compressor; according to the technical scheme, the defect that traditional fault codes need to be translated by professionals is overcome, real-time, automatic and operable output of diagnosis information of the air conditioner compressor is achieved, non-intrusive monitoring of the running state of the compressor is achieved, and system shutdown accidents caused by progressive faults are avoided.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of air conditioning system fault diagnosis, and in particular to an air conditioner compressor fault diagnosis method and system based on acoustic feature recognition. BACKGROUND

[0002] In the current air conditioner compressor operation process, the high-pressure refrigerant circulating condition forms a sustained alternating load on the mechanical parts, which easily causes progressive failures such as impeller wear and cavitation; the traditional contact type vibration sensor destroys the sealing of the equipment due to installation and cannot capture the fluid acoustic features, resulting in missed detection of key fault signals.

[0003] The current mainstream scheme adopts a fault diagnosis technology based on multi-channel vibration signal spectrum analysis, vibration sensors are deployed at multiple points on the compressor shell to collect vibration spectra, and the energy distribution characteristics of the impeller rotating harmonics are extracted and input into a pre-trained recognition model to complete classification; this scheme realizes automatic diagnosis, but has inherent defects; the vibration sensor can only indirectly capture mechanical conduction signals and cannot directly obtain the wideband acoustic features induced by refrigerant pulsation, resulting in missed detection of fluid faults such as cavitation; the speed self-adaptive module only mechanically adjusts the sampling rate, and does not establish the coupling relationship between acoustic resonance and impeller dynamics response under variable frequency conditions, so the misjudgment rate significantly increases when the speed suddenly changes; the multi-sensor drilling installation destroys the sealing structure of the compressor, interferes with the stability of the refrigerant circulation, and is difficult to apply to in-service equipment modification. SUMMARY

[0004] The present application provides an air conditioner compressor fault diagnosis method and system based on acoustic feature recognition, to solve the fluid acoustic feature capture distortion problem caused by indirect detection of vibration sensors in the prior art, and the problem of increased fault misjudgment rate under variable frequency conditions caused by the lack of dynamic condition collaborative analysis.

[0005] In a first aspect, the present application provides an air conditioner compressor fault diagnosis method based on acoustic feature recognition, comprising:

[0006] obtaining an operating sound signal in the operation process of an air conditioner compressor;

[0007] based on the operating sound signal, identifying an acoustic feature mode related to compressor fault;

[0008] matching the identified acoustic feature mode with a pre-stored compressor fault feature library to obtain a matching result;

[0009] outputting a real-time fault diagnosis report of the air conditioner compressor according to the matching result.

[0010] Optionally, obtaining an operating sound signal in the operation process of an air conditioner compressor comprises:

[0011] decomposing the operating sound signal into a plurality of discrete frequency components;

[0012] from the plurality of discrete frequency components, screening out target frequency components directly related to the rotating movement of the impeller of the compressor;

[0013] tracking the continuous variation process of sound signal intensity on the target frequency components to generate a tracking result;

[0014] extracting a deviation pattern that repeatedly appears in the tracking result;

[0015] converting the deviation pattern into the acoustic feature pattern.

[0016] Optionally, extracting the deviation pattern that repeatedly appears in the tracking result comprises:

[0017] marking all time points at which the sound signal intensity exceeds a reference value from the tracking result;

[0018] grouping and counting the marked time points to generate an interval distribution between each group of time points;

[0019] locating a cluster of time points within a fixed interval range in the interval distribution;

[0020] combining the sound signal intensity variation segment corresponding to the cluster of time points into the deviation pattern.

[0021] Optionally, converting the deviation pattern into the acoustic feature pattern comprises:

[0022] defining a segmentation unit length according to the fixed interval range;

[0023] segmenting the deviation pattern into a plurality of continuous pattern units according to the segmentation unit length;

[0024] locating an ascending direction feature point position and a descending direction feature point position of sound signal intensity in each pattern unit;

[0025] recording the intensity value corresponding to the ascending direction feature point position and the intensity value corresponding to the descending direction feature point position in each pattern unit;

[0026] combining the intensity values of all pattern units into an ordered set as the acoustic feature pattern.

[0027] Optionally, matching the identified acoustic feature pattern with a pre-stored compressor fault feature library to obtain a matching result, comprising:

[0028] sequentially selecting each feature item in the compressor fault feature library, wherein the feature item contains a standard feature point position sequence and a corresponding intensity value range of the standard feature point position sequence;

[0029] For the currently selected feature item, each feature point position in the acoustic feature mode is sequentially calculated with the position offset of the standard feature point position sequence, and the cumulative number of position offsets exceeding the allowable deviation value is counted;

[0030] When the cumulative number is lower than a preset threshold, the current feature item is recorded as a candidate matching item;

[0031] The first recorded candidate matching item is selected as the matching result.

[0032] Optionally, according to the matching result, an real-time fault diagnosis report of the air conditioner compressor is output, including:

[0033] Extracting the fault type identifier contained in the feature item in the matching result;

[0034] Separating the position description word in the fault type identifier from the fault behavior description word into independent fields;

[0035] Converting the intensity difference value parameter in the feature item into a severity description word;

[0036] Generating a diagnosis text unit in the order of the position description word, the fault behavior description word and the severity description word;

[0037] Obtaining the current system time, combining the diagnosis text unit with the current system time to form a time-related record;

[0038] Based on the time-related record as a basic unit, outputting an air conditioner compressor fault diagnosis report.

[0039] Optionally, converting the intensity difference value parameter in the feature item into a severity description word includes:

[0040] Obtaining a pre-set critical classification boundary sequence, wherein the critical classification boundary sequence contains a plurality of boundary values arranged in ascending order;

[0041] Calculating the reference interval between the intensity difference value parameter and each boundary value in the critical classification boundary sequence;

[0042] Determining the boundary value position identifier of the minimum reference interval;

[0043] Taking the predetermined description word corresponding to the boundary value position identifier as the severity description word.

[0044] In a second aspect, the application provides an air conditioner compressor fault diagnosis system based on acoustic feature recognition, comprising:

[0045] an acquisition module configured to acquire an operating sound signal of an air conditioner compressor during operation;

[0046] an identification module configured to identify an acoustic feature pattern associated with compressor fault based on the operating sound signal;

[0047] a matching module configured to match the identified acoustic feature pattern with a pre-stored compressor fault feature library to obtain a matching result;

[0048] an output module configured to output a real-time fault diagnosis report of the air conditioner compressor according to the matching result.

[0049] In a third aspect, the application provides a computing device comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement the air conditioner compressor fault diagnosis method based on acoustic feature recognition as described in the first aspect.

[0050] In a fourth aspect, the application provides a computer storage medium storing a computer program, wherein the computer program is executed by a computer to implement the air conditioner compressor fault diagnosis method based on acoustic feature recognition as described in the first aspect.

[0051] The application solves the limitation of traditional vibration sensors that need to damage the equipment seal installation by directly acquiring the operating sound signal of the air conditioner compressor and identifying the acoustic feature pattern associated with the fault. The scheme matches the acoustic feature pattern with the fault feature library and outputs the diagnosis report, realizes the non-invasive monitoring of the compressor operating state, ensures the integrity of the refrigeration system, and completely captures the fluid acoustic fault features under the high-pressure refrigerant working condition. The finally generated real-time diagnosis report effectively guides the operation and maintenance decision, avoids the gradual fault causing system shutdown accidents.

[0052] Further, for the rotating fault mode specific to the air conditioner compressor impeller, the target frequency component related to the impeller movement is accurately separated from the operating sound signal, and the repeated deviation mode is extracted by tracking the continuous change process of the sound intensity. The scheme creatively converts the deviation mode into an acoustic feature mode, breaking through the bottleneck of traditional frequency spectrum analysis that is difficult to distinguish normal vibration from fluid noise interference. This process enhances the sensitivity to impeller wear, cavitation and other faults, especially under the working condition of sudden change of refrigerant pressure, ensuring the reliability of acoustic feature extraction and the early identification ability of compressor faults; at the same time, a man-machine cooperative diagnosis report generation mechanism is constructed, the position description word and the fault behavior description word of the fault type identification are analyzed, and the severity level converted by the intensity difference parameter is combined to form a diagnosis text unit that can be directly understood. Finally, the diagnosis text and the system time are bound to generate a structured report entity, which not only retains the original accuracy of acoustic feature data, but also meets the needs of operation and maintenance personnel to quickly locate faults, overcoming the defects of traditional fault codes that need to be translated by professionals, and realizing the real-time, automatic and operable output of air conditioner compressor diagnosis information.

[0053] These aspects or other aspects of the present application will be more apparent in the following description of embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0054] In order to more clearly illustrate the technical solutions in the application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0055] Figure 1 A flow chart of an air conditioner compressor fault diagnosis method based on acoustic feature recognition provided by the present application is shown;

[0056] Figure 2 A structural schematic diagram of an air conditioner compressor fault diagnosis system based on acoustic feature recognition provided by the present application is shown;

[0057] Figure 3 A structural schematic diagram of a computing device provided by the present application is shown. DETAILED DESCRIPTION

[0058] In order for those skilled in the art to better understand the technical solutions of the present application, the technical solutions in the present application will be described clearly and completely below in conjunction with the drawings in the present application.

[0059] In some of the processes described in this specification and in the accompanying drawings, multiple operations are described in a particular, sequential order. However, it should be understood that, unless otherwise specifically noted, these operations can be performed in any order, including simultaneously, and that the ordering of operations should not be understood to imply that there are no actions between operations, unless otherwise contextually indicated. Additionally, some of the processes described in this specification and in the accompanying drawings can include more, fewer or other steps. Additionally, although this description can sometimes use an order of describing processes as the order in which they happen, this is not the exclusive order. For instance, blocks can occur in a different order, can occur in parallel, or can occur sequentially. Also, descriptions of “first” and “second” are used to distinguish different messages, devices, modules, etc., and do not necessarily imply a temporal sequence or a different type.

[0060] The technical solutions in the present application will be described clearly and completely below in conjunction with the accompanying drawings in the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative work fall within the scope of protection of the present application.

[0061] Figure 1 A flowchart of an air conditioner compressor fault diagnosis method based on acoustic feature recognition is provided in the present application, as shown in Figure 1 The method comprises the following steps:

[0062] Step 101, obtaining the running sound signal in the running process of the air conditioner compressor.

[0063] In this step, the running sound signal refers to the sound data generated when the air conditioner compressor is running in a normal or fault state. These data exist in the form of sound waves and contain information such as frequency distribution and amplitude variation, which are used for subsequent acoustic feature extraction and fault analysis. Specifically, this signal is composed of vibration sound waves generated by the mechanical parts of the compressor (such as impeller, bearing) during movement, and its characteristics may reflect potential faults, such as abnormal friction or imbalance.

[0064] In this embodiment, first, an acoustic sensor (such as a condenser microphone) is installed at a designated position on the air conditioner compressor shell to capture the original sound waves in a non-invasive manner; the analog signal output by the sensor is gain-adjusted by a preamplifier, and then passes through an anti-aliasing filter (such as a Butterworth low-pass filter) to remove high-frequency noise interference; then, the signal is input to an analog-to-digital converter (ADC) for sampling and quantization, converting continuous sound waves into discrete digital sequences; finally, the digital sequence is transmitted to the central processing unit through a data acquisition interface (such as USB or Bluetooth), and stored as a time series data file (such as WAV format) for subsequent feature recognition.

[0065] For example, in a test scenario of an air conditioner compressor fault diagnosis system, an engineer fixes a high-sensitivity microphone near the air inlet of the compressor shell to capture the sound of the impeller rotating; after amplification and filtering, the sensor signal is digitized by the embedded ADC module at a reasonable sampling rate and transmitted to the cloud server through Wi-Fi to be stored as an audio file. This file serves as the basis for subsequent acoustic feature pattern recognition, ensuring that the starting data of the fault diagnosis process is complete and reliable.

[0066] Step 102, based on the operating sound signal, identify the acoustic feature pattern related to compressor fault.

[0067] In this step, the acoustic feature pattern is a repetitive abnormal data sequence that quantitatively represents the compressor fault, containing feature point position and intensity value information.

[0068] In this embodiment, first, apply the Fast Fourier Transform (FFT) algorithm to the input operating sound signal (digital time series data from step 101) for spectral decomposition, converting it into multiple discrete frequency components, each containing a frequency value and a corresponding amplitude value; then, based on the impeller rotation characteristics of the compressor (such as the pre-set impeller fundamental frequency range), use a band-pass filtering algorithm to filter out the target frequency components from the discrete frequency components, ensuring that only the frequency band directly related to the rotating motion is retained; then, on the target frequency components, use envelope detection techniques (such as Hilbert transform) to track the continuous changes in sound signal intensity, generating a time series of tracking results, which are represented as a sequence of intensity time data points; after that, apply pattern recognition algorithms (such as peak detection and clustering analysis) to the tracking results to identify recurring deviation patterns, including marking time points where intensity exceeds dynamic benchmarks, grouping interval distribution, locating fixed interval clusters and combining them into deviation segments; finally, use segmentation and feature point positioning methods (such as sliding window segmentation and local extremum detection) to convert the deviation pattern into a standardized acoustic feature pattern, including defining segmentation units, segmentation pattern units, locating feature point positions and recording intensity values, forming an ordered set. The entire process is executed in series: the output of FFT decomposition (discrete frequency components) is used as the input for screening, the target frequency components after screening are used for envelope detection to generate tracking results, the tracking results are used as the input of the pattern recognition algorithm to extract the deviation pattern, and the deviation pattern is finally converted into the acoustic feature pattern for subsequent fault matching.

[0069] For example, in the test scenario of air conditioner compressor fault diagnosis system, based on the WAV format audio file acquired and stored in step 101, the engineer first loads the file and applies the FFT algorithm to decompose the signal to obtain the frequency spectrum data containing multiple frequency components; then, according to the impeller design parameters of the compressor (such as the base frequency corresponding to the rated speed), the target frequency component (for example, the frequency band concentrated in 100-500Hz) is filtered out using a digital filter; then, envelope detection is performed on the frequency band to generate a curve whose intensity changes with time; then, the algorithm automatically identifies the repeatedly appearing intensity surges (such as abnormal peaks every 0.5 seconds) in the curve and extracts them as deviation patterns; finally, the deviation patterns are segmented into unit sequences, the rising point and falling point intensity values of each unit are located, and they are combined into acoustic feature patterns (such as an ordered set of [intensity 1, intensity 2,...]). This feature pattern will be used as input for the fault feature library matching in step 103.

[0070] Step 103, match the identified acoustic feature pattern with the pre-stored compressor fault feature library to obtain a matching result.

[0071] In this step, the matching result is the diagnostic entry output after the successful comparison of the acoustic feature pattern with the fault feature library entry, which contains the fault type identification and intensity parameters.

[0072] In this embodiment, the matching process is realized by sequentially traversing the fault feature library: first, load a feature entry (such as the "impeller imbalance" entry), which contains a standard feature point position sequence (such as a time point sequence and the corresponding intensity value range); then, use the dynamic time warping (DTW) algorithm to align each feature point position in the acoustic feature pattern with the standard sequence point by point, and calculate the position offset |pi-tj| of each point; then, count the number of feature points whose offset exceeds the allowed deviation value (such as the preset timing tolerance), and record it as the cumulative number; if the cumulative number is less than the preset threshold (such as no more than 10% of the total number of feature points), record the feature entry as a candidate matching entry; finally, select the first candidate entry that meets the conditions as the matching result output (such as matching the most common fault type in the library first).

[0073] Step 104, output the real-time fault diagnosis report of the air conditioner compressor according to the matching result.

[0074] In this embodiment, the fault type identifier in the matching results is first parsed, and string segmentation technology is used to separate location descriptors and fault behavior descriptors. Then, the intensity difference parameter is extracted and converted into severity descriptors using a boundary value mapping algorithm. The preset ascending critical grade boundary sequence is loaded, the absolute distance between the intensity difference and each boundary value is calculated, the boundary value location identifier corresponding to the minimum distance is located, and the associated pre-stored descriptive word library is called to output the text label. Then, a diagnostic text unit is synthesized in the order of "location-behavior-severity". Simultaneously, the system clock timestamp is obtained in real time, and the text unit and the timestamp are combined into a structured time-related record. Finally, based on this record, a readable diagnostic report containing the time, fault location, anomaly type, and severity level is generated.

[0075] As another practicable embodiment, according to step 102, identifying an acoustic characteristic pattern related to a compressor fault based on the operating sound signal includes:

[0076] Step 201: decompose the running sound signal into multiple discrete frequency components.

[0077] In this step, discrete frequency components refer to independent frequency components separated from the running sound signal after mathematical transformation. Each component contains a specific frequency value and its corresponding amplitude, which is used to characterize the acoustic energy distribution in different frequency bands.

[0078] In this embodiment, the operating sound signal (time domain digital sequence) obtained in step 101 is first read, and FFT is applied to convert it from the time domain to the frequency domain to generate spectrum data in complex form; then the amplitude of each frequency component is calculated (the complex modulus length is taken) to obtain a set of discrete frequency point amplitude pairs, that is, discrete frequency components.

[0079] Step 202 : Filter out a target frequency component directly related to the rotational motion of the impeller of the compressor from the plurality of discrete frequency components.

[0080] In this step, the target frequency component is a specific frequency band subset extracted from the discrete frequency components that is strongly associated with the physical motion of the impeller, and its change directly reflects the impeller rotation state (such as fundamental frequency, harmonics or sideband frequency).

[0081] In this embodiment, the characteristic frequency band range is first determined, and based on the impeller design parameters of the compressor (such as the rated speed), the impeller rotation fundamental frequency (speed / 60) and its main harmonic frequencies (such as 2 times the fundamental frequency, 3 times the fundamental frequency) are calculated, and the preset bandwidth is expanded to form a target frequency band interval; then frequency band filtering is performed, and a digital bandpass filter (such as an FIR filter) is used to process the discrete frequency components, retaining the components that fall within the target frequency band interval, eliminating irrelevant frequency band components, and finally outputting the target frequency components.

[0082] Step 203, track the continuous variation process of the sound signal intensity on the target frequency component to generate a tracking result.

[0083] In this step, the tracking result is the continuous trajectory data of the acoustic signal intensity on the target frequency component changing over time, which is expressed as a time sequence amplitude curve, used to capture the dynamic abnormal characteristics in the rotating process of the compressor impeller.

[0084] In this embodiment, first, inverse fast Fourier transform (IFFT) is performed on the target frequency components screened in step 202 to reconstruct them into time domain sub-signals; then, the Hilbert transform is applied to each sub-signal to calculate the instantaneous envelope and extract the continuous trajectory of the amplitude changing over time; finally, the envelope curves of multiple target frequency components are superimposed along the time axis to generate a comprehensive tracking result.

[0085] Step 204, extract the deviation pattern that repeatedly appears in the tracking result.

[0086] In this step, the deviation pattern is a periodic or regular intensity anomaly segment identified from the tracking result, which is expressed as a repeated amplitude mutation feature in a specific time interval, used to represent the acoustic fingerprint of the compressor fault.

[0087] In this embodiment, first, the moving average intensity is calculated as a baseline according to the tracking result; then, all mutation points where the intensity exceeds the dynamic baseline are detected and the time stamps are recorded; finally, interval statistical analysis is performed on the time stamp sequence to locate the mutation point cluster in a fixed time interval, and the intensity change segment corresponding to the time interval is intercepted and combined into a repetitive deviation pattern.

[0088] Step 205, convert the deviation pattern into the acoustic feature pattern.

[0089] In this embodiment, the conversion process uses gradient extremum positioning and spatiotemporal binding packaging technology, first defines the unit length based on the repetition interval of the deviation pattern; then, the continuous deviation pattern is divided into independent pattern units according to the unit length; in each unit, the rising direction feature point (maximum positive gradient position) and the falling direction feature point (maximum negative gradient position) are located by gradient analysis algorithm, and the intensity values corresponding to the two points are recorded simultaneously; finally, the position-intensity value pairs are packaged in order to generate an ordered set of acoustic feature patterns.

[0090] As another embodiment, according to step 204, extracting the deviation pattern that repeatedly appears in the tracking result, comprises:

[0091] Step 301, mark all time points where the sound signal intensity exceeds the baseline value from the tracking result.

[0092] In this step, the reference value refers to the dynamic reference baseline of the sound signal intensity of the air conditioner compressor under a specific speed condition, which is automatically generated from the historical normal sound intensity fluctuation interval corresponding to the current speed.

[0093] In this embodiment, first, the dynamic envelope baseline generation algorithm (based on Hilbert envelope demodulation and sliding window statistics) is called to generate an adaptive reference curve in combination with the current speed value of the compressor. Real-time amplitude comparison operations are performed on the sound intensity continuous change curve in the tracking result, and all intensity points exceeding the reference line are scanned. Time stamp accurate recording technology (through high-precision clock interrupt service) is used to mark the occurrence time of each out-of-limit point.

[0094] Step 302, grouping statistics is performed on the sequence of marked time points to generate the interval distribution between each group of time points.

[0095] In this step, the adjacent interval refers to the absolute value of the time difference between two consecutive time points in the fault feature trigger point sequence, which is used to represent the periodicity of the air conditioner compressor impeller rotation abnormal event. The interval distribution set is a statistical data structure generated by grouping and aggregating adjacent intervals, which contains the mapping relationship of the frequency of interval occurrence in each interval segment, and directly reflects the periodicity or divergence characteristics of the impeller rotation movement in the acoustic abnormality level.

[0096] In this embodiment, after arranging the sequence of marked time points in chronological order, the interval value between each two adjacent points is first calculated to generate the original interval set. Based on the grouping interval division of the theoretical rotation period of the compressor impeller, a plurality of grouping interval segments centered on the theoretical period value are designed. Interval value interval matching statistics are performed to distribute each interval value into the corresponding grouping interval segment. Finally, the frequency count values of the same interval segment are merged to output the interval distribution set containing the interval boundary and the frequency.

[0097] Step 303, locating the time point cluster within the fixed interval range in the interval distribution.

[0098] In this step, the fixed interval range refers to the reasonable deviation interval allowed by the theoretical rotation period of the air conditioner compressor impeller under a specific speed condition, which is determined by the thermal expansion coefficient of the impeller material and the speed control accuracy; the time point cluster refers to an ordered set of feature trigger points that satisfy the interval consistency constraint on the continuous time axis, which embodies the periodic acoustic abnormal event chain caused by local damage of the impeller (such as single blade cavitation).

[0099] In this embodiment, based on the interval distribution set, fixed interval matching is performed: first, the theoretical rotation period is calculated based on the real-time speed value of the compressor, and the period allowed deviation interval (such as ±10%) is set according to the working condition; all interval data segments falling within this interval are retrieved in the interval distribution set; the original time point records corresponding to these interval data segments are extracted; and the adjacent points are merged according to the time continuity to generate the time point cluster.

[0100] Step 304, combine the sound signal intensity change segments corresponding to the time point cluster into the deviation pattern.

[0101] In this step, the sound signal intensity change segment is a sound intensity curve data segment centered on a specific time point, with a length equal to an integer multiple of the theoretical rotation period of the air conditioner compressor impeller (ensuring coverage of the complete mechanical motion period), containing the complete sound intensity fluctuation pattern of the leading and trailing transition zones.

[0102] In this embodiment, for each time point in the time point cluster, a sound intensity change curve segment of the theoretical rotation period length is symmetrically intercepted centered on the point; signal mutations at the segment connection are eliminated through endpoint smoothing transition technology (cosine window function weighting); all curve segments are connected in sequence to form a continuous waveform; finally, amplitude normalization alignment is performed (scaling subsequent segments based on the first segment) to ensure rotation period consistency, and the smooth periodic acoustic anomaly profile is output as the deviation pattern.

[0103] As another embodiment, according to step 205, converting the deviation pattern into the acoustic feature pattern comprises:

[0104] Step 401, define a segmentation unit length according to the fixed interval range.

[0105] In this step, the segmentation unit length is a standard time span established according to the central trend statistics of the fixed interval range, which is used as a reference scale for equally dividing the continuous deviation pattern, and its physical meaning is equivalent to the typical value of the actual rotation period of the air conditioner compressor impeller within the diagnosis period, ensuring that each segmentation unit contains the acoustic response period generated by one impeller rotation process.

[0106] In this embodiment, first, the minimum and maximum boundary points are extracted from the fixed interval range, and the arithmetic mean value is calculated as the central reference value; the reference value is determined as the segmentation unit length as the core constant; based on this length, an equidistant segmentation point sequence is established on the deviation pattern time axis; finally, a standard time slice unit set is generated with the segmentation points as boundary markers, providing a unified scale reference for subsequent feature extraction.

[0107] Step 402, segment the deviation pattern into multiple continuous pattern units according to the segmentation unit length.

[0108] In this step, the continuous pattern unit is an equal-length time interval segment strictly divided on the deviation pattern time axis according to the segmentation unit length, and each unit corresponds to a sound response segment of one complete rotation period of the air conditioner compressor impeller, with a time span equal to the segmentation unit length (allowing insufficient length for the last segment), ensuring that the unit contains complete mechanical motion acoustic features without period truncation.

[0109] In this embodiment, the reading step 401 defines a constant value of the segmentation unit length; locates the starting point of the time axis of the deviation pattern waveform; starts from the starting point and gradually cuts off equal time periods in the segmentation unit length; independently marks the residual section at the end that cannot be completely segmented as a special unit; and finally outputs a set of time segments arranged in strict sequence from the time starting point, each segment carrying the acoustic response data of the corresponding rotation period.

[0110] Step 403, locate the rising direction feature point position and the falling direction feature point position of the sound signal intensity in each of the mode units.

[0111] In this step, the rising direction feature point position refers to the local maximum value corresponding to the time stamp when the sound intensity curve slope changes from positive to negative, representing the acoustic response vertex of the sudden rise of the inlet pressure in the single cycle rotation of the air compressor impeller; the falling direction feature point position is the local minimum value time stamp when the slope changes from negative to positive, reflecting the bottom point of the acoustic energy released by the exhaust pressure, and the two together constitute the key turning marks of the mechanical-fluid coupling response in the impeller rotation period.

[0112] In this embodiment, the time-sound intensity curve of each mode unit is subjected to continuous slope calculation to generate a first derivative sequence, the derivative sequence is scanned to locate the zero-crossing points where the sign changes from positive to negative and record them as the rising feature point positions; the zero-crossing points where the sign changes from negative to positive are marked as the falling feature point positions; false inflection point interference is excluded through local interval extremum verification, and finally the accurate time coordinate pair set of the rising and falling feature points in each mode unit is output.

[0113] Step 404, record the intensity value corresponding to the rising direction feature point position and the intensity value corresponding to the falling direction feature point position in each of the mode units.

[0114] In this step, the intensity value refers to the instantaneous physical quantity measurement result of the sound signal intensity at the time corresponding to the feature point position, directly reflecting the acoustic energy release intensity of the air compressor impeller at the key phase points of the rotation period (such as the maximum compression pressure time and the minimum pressure relief time), and is the original scalar data unit for quantifying the degree of mechanical fault abnormality.

[0115] In this embodiment, the rising direction feature point position and the falling direction feature point position coordinate set located in each mode unit are traversed; the instantaneous intensity values of the sound intensity curve at these specific time coordinate points are directly read; a binding record structure of feature point type identification and intensity value is established; and finally a data set containing all the feature point intensity values and their type labels in all the mode units is generated, preserving the original measurement accuracy without any conversion processing.

[0116] Step 405, combine the intensity values of all the mode units into an ordered set as the acoustic feature pattern.

[0117] In the embodiment, the unit priority queue is established according to the sequence of the mode units on the time axis; the intensity value set (containing the rising feature point intensity and the falling feature point intensity) corresponding to each unit is extracted from the head of the queue in turn; the intensity value pairs of each unit are spliced in sequence according to the unit time sequence; and finally a one-dimensional intensity value sequence is generated as the acoustic feature mode.

[0118] As another embodiment, the identified acoustic feature mode is matched with the compressor fault feature library pre-stored according to step 103, and a matching result is obtained, including:

[0119] Step 501, each feature item in the compressor fault feature library is selected in sequence, wherein the feature item contains a standard feature point position sequence and an intensity value range corresponding to the standard feature point position sequence.

[0120] In the embodiment, the retrieval pointer is initialized to locate the address of the first item of the fault feature library, the standard feature point position sequence data (the array of theoretical time points) and the intensity value range set (the list of sound intensity lower limit-upper limit value pairs) of the item are loaded, the pointer is sequentially jumped to the next item address after the data loading of the current item is completed, and the loop operation is performed; the traversal is ended when the pointer points to the library termination marker, and an item access queue arranged in the original order is generated for subsequent matching calls.

[0121] Step 502, for the currently selected feature item, the position offset amount of each feature point position in the acoustic feature mode from the standard feature point position sequence is calculated in turn, and the cumulative number of times that the position offset amount exceeds the allowable deviation value is counted.

[0122] In this step, the position offset amount refers to the absolute difference between the actual time stamp of a single feature point in the acoustic feature mode and the theoretical time value of the index point corresponding to the standard position sequence of the fault feature item, which directly quantifies the time sequence drift anomaly of the acoustic event in the rotating period of the compressor impeller of the air conditioner; the allowable deviation value is a pre-set dynamic time tolerance threshold, which is determined jointly by the bearing gap tolerance and the speed control accuracy; the cumulative number of times is counted by an accumulator in real time to count the total number of times that the position offset amount exceeds the tolerance, which reflects the mismatching degree of the current feature item and the measured acoustic mode.

[0123] In the embodiment, for the selected feature item (such as the impeller cavitation), the index point forced alignment operation of the acoustic feature mode and the standard position sequence is performed; the absolute value difference between the actual position and the standard position of each index position point is calculated in real time to generate a position offset amount data set; the size relationship between each value in the position offset amount data set and the allowable deviation value is compared synchronously, and the accumulator is driven to count incrementally when the out-of-tolerance condition is triggered; after the full index traversal is completed, the final accumulator value is output as the cumulative out-of-tolerance frequency.

[0124] Step 503: When the accumulated number of times is lower than a preset threshold, the current feature entry is recorded as a candidate matching entry.

[0125] In this step, the preset threshold is a critical criterion dynamically generated based on the ratio of the total number of feature points in the acoustic feature pattern, reflecting the tolerance of the air-conditioning compressor fault diagnosis system to abnormal events in the impeller rotation cycle; the candidate matching entry refers to the temporary record of the feature entry identification and offset summary when the cumulative deviation frequency reaches the standard, including the fault type code and the distribution snapshot of the key deviation abnormal point, which serves as the intermediate result carrier of the matching process.

[0126] In this embodiment, the cumulative frequency data of out-of-tolerance generated by the current feature entry matching process is obtained, and the floating threshold value preset according to the ratio of the total number of feature points is read; a scalar comparison operation between the frequency and the threshold is performed; when the frequency is lower than the threshold, the out-of-tolerance position index information in the current entry identifier and position offset set is extracted; a snapshot structure containing the fault type code and the abnormal point distribution is packaged and generated; and the structure is written into the candidate matching entry cache for temporary storage.

[0127] Step 504: Select the first recorded candidate matching entry as the matching result.

[0128] In this embodiment, the status register bit of the candidate matching entry cache is monitored; when the status register bit indicates that it is not empty, the hardware interrupt mechanism is triggered to generate a short-circuit output flag; the candidate entry data structure (including the fault type code and the out-of-tolerance point bitmap) stored at the first address of the cache is read through the direct memory access (DMA) controller; and the entry content is mapped into a standardized matching result format.

[0129] As another embodiment, according to step 104, outputting a real-time fault diagnosis report of the air-conditioning compressor based on the matching result includes:

[0130] Step 601: extract the fault type identifier contained in the feature entry in the matching result.

[0131] In this step, the fault type identifier is a predefined fault semantic code in the compressor fault feature library entry. It is a composite string formed by concatenating the location descriptor (mechanical component area) and the fault behavior descriptor (damage type) with a separator. For example, "impeller|cavitation" represents a cavitation fault in the impeller area. This serves as the core classification basis for the diagnostic report.

[0132] In this embodiment, the fault type identification field is located from the matching result data structure output from step 504; the string format of the field is parsed to identify the separator position; the string is cut into the front position description word and the back fault behavior description word with the separator as the boundary; and they are stored as independent string variables respectively to complete the deconstruction and extraction of the fault type identification.

[0133] Step 602, separate the location description word in the fault type identification from the fault behavior description word into independent fields.

[0134] In this step, the location description word refers to the semantic unit in the fault type identification that represents the position of the compressor mechanical component (such as "impeller" pointing to the rotating component area); the fault behavior description word refers to the semantic unit in the fault type identification that describes the damage mechanism (such as "bearing wear" representing the friction loss behavior). The separation process is an atomic operation that splits the composite string into two independent semantic units.

[0135] In this embodiment, the fault type identification string content is directly read; the position of the predefined separator symbol (such as the vertical bar "|") in the string is identified; the front character substring is cut off as the location description word with the separator symbol as the boundary; the rear character substring is cut off as the fault behavior description word; and the two substrings are stored as independent text fields for subsequent calls.

[0136] Step 603, convert the intensity difference value parameter in the feature item into a severity description word.

[0137] In this embodiment, the pre-stored ordered boundary value array is first called, and the sequence is arranged in ascending order; then the absolute difference value operation is performed on the intensity difference value parameter and each boundary value in the boundary sequence; then the minimum value corresponding to the boundary value index position is found by scanning the interval set; finally, the corresponding severity description word is output according to the boundary value index position retrieved from the pre-set description word dictionary.

[0138] Step 604, obtain the current system time, combine the diagnostic text unit with the current system time to form a time-related record.

[0139] In this step, the current system time refers to the precise time stamp at the time of triggering the diagnostic operation, including date, hour, minute, second and millisecond information; the time-related record is a structured data entity generated by binding the diagnostic text unit with the current time stamp, used to mark the occurrence time of the air conditioner compressor fault event, ensuring that the fault diagnosis report has traceability and time sequence analysis capability.

[0140] In this embodiment, the real-time clock (RTC) interface of the operating system is called to obtain the current UTC standard time string; the key-value binding technique is used to bind the diagnostic text unit as the value body and the current time as the key name; the JSON format record entity containing the time stamp and the diagnostic text is generated through data encapsulation, realizing the permanent association storage of the fault event and the occurrence time.

[0141] Step 605, based on the time-related record as the basic unit, output the air conditioner compressor fault diagnosis report.

[0142] In this embodiment, first, a compressor device unique identification field is added in the time correlation record; then, all time correlation records are arranged in ascending order of timestamp to generate an event timeline; then, the device identification and event sequence are encapsulated as a JSON object, and a report generation timestamp and a check code are attached, and an air conditioner compressor fault diagnosis report package is output.

[0143] As a further embodiment, according to step 603, converting the intensity difference value parameter in the feature item into a severity descriptor includes:

[0144] Step 701, obtaining a pre-set critical classification boundary sequence, wherein the critical classification boundary sequence contains a plurality of boundary values arranged in ascending order.

[0145] In this step, the critical classification boundary sequence is a pre-defined sound intensity anomaly degree quantification scale, which is formed by arranging the intensity difference value thresholds corresponding to different fault levels of the air conditioner compressor in ascending order, for example, the boundary value sequence [low threshold, medium threshold, high threshold] corresponds to the watershed of the light, moderate and severe fault levels respectively.

[0146] In this embodiment, the pre-stored configuration module in the compressor fault feature library is called to load the ordered array of boundary values bound to the current diagnosis fault type (such as impeller cavitation); the strict ascending order property of the sequence is verified (to ensure that the latter is greater than the former); the boundary value sequence that passes the verification is loaded into the memory buffer area as the reference scale for the intensity difference value parameter conversion.

[0147] Step 702, calculating the reference interval between the intensity difference value parameter and each boundary value in the critical classification boundary sequence.

[0148] In this step, the reference interval is the absolute difference scalar of the intensity difference value parameter and a single boundary value, which is used to quantify the degree of deviation of the current fault state from each critical point of the classification. The calculation process is to traverse each element of the boundary sequence and independently perform the absolute value operation of the arithmetic difference between the intensity difference value parameter and the boundary value.

[0149] In this embodiment, first, read each boundary value of the critical classification boundary sequence in the storage order; then, perform subtraction operation between the current boundary value and the intensity difference value parameter, and take the absolute value of the result; finally, store the absolute difference value corresponding to each boundary value in sequence as the reference interval set, preserving the original calculation precision.

[0150] Step 703, determining the position identifier of the boundary value with the minimum reference interval.

[0151] In this step, the boundary value position identifier is the sequence index number corresponding to the minimum value element in the reference interval set, which directly maps a specific boundary value position in the critical classification boundary sequence, used to lock the closest fault classification critical point of the intensity difference value parameter. The position identifier represents the sequential ranking of the boundary value in the sequence in integer form (the first item is 0, the second item is 1, and so on).

[0152] In this embodiment, first, all elements in the set are traversed to locate the minimum numerical reference interval through a linear comparison process; then the subscript sequence number of the minimum value in the set is recorded (such as the first subscript 0, the second subscript 1); finally, the subscript number is output as the boundary value position identifier, which strictly corresponds to the index of the critical classification boundary sequence.

[0153] Step 704, taking the predetermined descriptive word corresponding to the boundary value position identifier as the severity descriptive word.

[0154] In this step, the predetermined descriptive word is a set of fault degree semantic labels pre-stored in the fault feature library, stored in order of index (such as index 0→"mild", index 1→"moderate", index 2→"severe"), which has a physical meaning of text description of the fault state level corresponding to the abnormal shift amount of the air conditioner compressor sound intensity.

[0155] In this embodiment, first, the descriptive word mapping table bound to the current fault type (such as impeller cavitation) in the fault feature library is called; then the boundary value position identifier is used as a search key value to locate the corresponding predetermined descriptive word in the mapping table; finally, the matching descriptive word text is extracted as the final severity descriptive word, completing the conversion of the intensity difference value to the semantic label.

[0156] Figure 2 A structure diagram of an air conditioner compressor fault diagnosis system based on acoustic feature recognition is provided for the present application, as shown in Figure 2 The system comprises:

[0157] An acquisition module 21 is configured to acquire a running sound signal during the running of an air conditioner compressor.

[0158] An identification module 22 is configured to identify an acoustic feature pattern related to a compressor fault based on the running sound signal.

[0159] A matching module 23 is configured to match the identified acoustic feature pattern with a pre-stored compressor fault feature library to obtain a matching result.

[0160] An output module 24 is configured to output a real-time fault diagnosis report of the air conditioner compressor according to the matching result.

[0161] Figure 2The air conditioner compressor fault diagnosis system based on acoustic feature recognition can execute Figure 1 The air conditioner compressor fault diagnosis method based on acoustic feature recognition of the embodiment has the implementation principle and technical effects which will not be repeated. The specific operation manner of each module and unit of the air conditioner compressor fault diagnosis system based on acoustic feature recognition in the above embodiment has been described in detail in the embodiment related to the method, and will not be described in detail here.

[0162] In one possible design, Figure 2 The air conditioner compressor fault diagnosis system based on acoustic feature recognition of the embodiment can be implemented as a computing device, such as a computer. Figure 3 As shown, the computing device can include a storage component 31 and a processing component 32.

[0163] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32.

[0164] The processing component 32 is configured to execute the above Figure 1 The air conditioner compressor fault diagnosis method based on acoustic feature recognition of the embodiment.

[0165] The processing component 32 can include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component can also be one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components, for executing the above method.

[0166] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component can be realized by any type of volatile or non-volatile storage device or their combination, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0167] Of course, the computing device can also include other components, such as input / output interface, display component, communication component, etc.

[0168] The input / output interface provides an interface between the processing component and the peripheral interface module, which can be an output device, an input device, etc.

[0169] The communication component is configured to facilitate wired or wireless communication between the computing device and other devices.

[0170] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform, and the computing device can be a cloud server, and the processing component, the storage component, and the like can be a basic server resource rented or purchased from the cloud computing platform.

[0171] The embodiment of the application further provides a computer storage medium storing a computer program, and the computer program can implement the above-mentioned Figure 1 The embodiment shown in the figure is an air conditioner compressor fault diagnosis method based on acoustic feature recognition.

[0172] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-mentioned system, device and unit can refer to the corresponding processes in the foregoing method embodiments, and will not be described here.

[0173] The device embodiments described above are only schematic, and the units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiment scheme according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0174] Through the description of the foregoing embodiments, those skilled in the art can clearly understand that each embodiment can be realized by means of software and a necessary general hardware platform, and of course, it can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0175] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the application, and not to limit them; although the application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for some technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the application.

Claims

1. A method for diagnosing air-conditioning compressor faults based on acoustic feature recognition, characterized in that: include: Acquire the operating sound signal of the air-conditioning compressor during operation; identifying an acoustic characteristic pattern associated with a compressor failure based on the operating sound signal; Matching the identified acoustic characteristic pattern with a pre-stored compressor fault characteristic library to obtain a matching result; According to the matching result, a real-time fault diagnosis report of the air-conditioning compressor is output.

2. The method according to claim 1, characterized in that Based on the operating sound signal, an acoustic characteristic pattern associated with a compressor fault is identified, including: Decomposing the operating sound signal into a plurality of discrete frequency components; screening out a target frequency component directly related to the rotational motion of the impeller of the compressor from the plurality of discrete frequency components; Tracking a continuous change in the intensity of the sound signal on the target frequency component to generate a tracking result; extracting recurring deviation patterns in the tracking results; The deviation pattern is converted into the acoustic characteristic pattern.

3. The method according to claim 2, characterized in that Extracting recurring deviation patterns in the tracking results, including: Marking all time points at which the sound signal strength exceeds a baseline value from the tracking results; Performing group statistics on the marked time point sequence to generate interval distribution between each group of time points; locating a cluster of time points within a fixed interval in the interval distribution; The sound signal intensity variation segments corresponding to the time point clusters are combined into the deviation pattern.

4. The method according to claim 2, characterized in that Converting the deviation pattern into the acoustic characteristic pattern comprises: Defining a segmentation unit length according to the fixed interval range; dividing the deviation pattern into a plurality of continuous pattern units according to the division unit length; Locating a rising direction feature point position and a falling direction feature point position of the sound signal intensity in each of the pattern units; Recording the intensity value corresponding to the position of the characteristic point in the ascending direction and the intensity value corresponding to the position of the characteristic point in the descending direction in each of the pattern units; The intensity values ​​of all mode units are combined into an ordered set as the acoustic characteristic pattern.

5. The method according to claim 1, wherein The identified acoustic feature pattern is matched with the pre-stored compressor fault feature library to obtain matching results, including: Sequentially selecting each feature entry in the compressor fault feature library, wherein the feature entry includes a standard feature point position sequence and an intensity value range corresponding to the standard feature point position sequence; For the currently selected feature entry, calculate the position offset of each feature point position in the acoustic feature pattern and the standard feature point position sequence in sequence, and count the cumulative number of times the position offset exceeds the allowable deviation value; When the cumulative number of times is lower than a preset threshold, the current feature entry is recorded as a candidate matching entry; The first recorded candidate matching entry is selected as the matching result.

6. The method according to claim 1, characterized in that Based on the matching results, a real-time fault diagnosis report of the air-conditioning compressor is output, including: Extract the fault type identifier contained in the feature entry in the matching result; Separating the location description word and the fault behavior description word in the fault type identifier into independent fields; converting the intensity difference parameter in the feature entry into a severity descriptor; Generate a diagnostic text unit in the order of location descriptor, fault behavior descriptor, and severity descriptor; Obtaining the current time of the system, and combining the diagnostic text unit with the current time of the system to form a time-related record; Based on the time-related records as basic units, an air-conditioning compressor fault diagnosis report is output.

7. The method according to claim 6, characterized in that Converting the intensity difference parameter in the feature entry into a severity descriptor includes: Obtaining a preset critical grading boundary sequence, wherein the critical grading boundary sequence comprises a plurality of boundary values ​​arranged in ascending order; Calculating a reference distance between the intensity difference parameter and each boundary value in the critical classification boundary sequence; Determine a boundary value position marker where the reference spacing is a minimum value; The predetermined descriptor corresponding to the boundary value position identifier is used as the severity descriptor.

8. An air-conditioning compressor fault diagnosis system based on acoustic feature recognition, characterized in that: include: An acquisition module is used to acquire an operating sound signal of the air-conditioning compressor during operation; an identification module, configured to identify an acoustic characteristic pattern associated with a compressor fault based on the operating sound signal; A matching module is used to match the identified acoustic characteristic pattern with a pre-stored compressor fault characteristic library to obtain a matching result; The output module is used to output a real-time fault diagnosis report of the air-conditioning compressor according to the matching result.

9. A computing device, characterized in that It includes a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement an air-conditioning compressor fault diagnosis method based on acoustic feature recognition as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that A computer program is stored, and when the computer program is executed by a computer, an air-conditioning compressor fault diagnosis method based on acoustic feature recognition as claimed in any one of claims 1 to 7 is implemented.

Images