Intelligent power supply unit fault recording and diagnosis method, device, system and storage medium

Abstract

The application relates to a smart power supply unit fault recording and diagnosis method, system and device and a storage medium. The method comprises the following steps: an electric parameter of a smart power supply unit is synchronously sampled through a data acquisition circuit, and the obtained sampling data is stored in a preset ring waveform storage area in real time; a cycle of the sampling data is identified through an edge analysis engine, and a steady-state waveform feature baseline of the smart power supply unit is established; when it is detected that the sampling data of the current cycle deviates from the steady-state waveform feature baseline, a freezing operation is performed on the ring waveform storage area, and deviated waveform identification is performed on the sampling data to obtain deviated recording and wave data; a deviated waveform feature of the deviated recording and wave data is extracted through the edge analysis engine, the deviated waveform feature is subjected to fault identification according to a preset fault feature library, and fault diagnosis information is obtained. The application can realize intelligent diagnosis of power supply abnormalities, and can convert a complex fault troubleshooting process into a process flow.

Description

Technical Field

[0001] This invention relates to the technical field of power supply units, and in particular to a method, device, system, and storage medium for fault recording and diagnosis of intelligent power supply units. Background Technology

[0002] Industrial power supply lines are complex in structure and diverse in equipment. When equipment malfunctions or restarts abnormally, maintenance personnel often struggle to quickly determine whether the fault originates from the power supply unit itself, the output line, or the downstream load equipment. Traditional intelligent power supply units typically have threshold alarm functions such as overcurrent, overvoltage, and undervoltage, issuing alarms when electrical parameters exceed set values. However, these alarms only record switching events and limit-crossing information, lacking analog voltage and current waveform data before and after the fault, thus failing to provide a basis for in-depth analysis. Existing equipment only provides raw data records, requiring specialized engineers to manually interpret waveforms on-site with instruments such as oscilloscopes. This not only heavily relies on personnel skills but also results in low analysis efficiency and difficulty in quickly locating the fault. Furthermore, traditional power supply units can only passively respond to faults that have already occurred and cannot provide early warnings for gradual problems such as line aging and contact oxidation. Summary of the Invention

[0003] The main objective of this invention is to provide a method, device, system, and storage medium for fault recording diagnosis of intelligent power supply units. By integrating a high-precision waveform acquisition data acquisition circuit and an edge analysis engine within the power supply unit, it enables automatic waveform recording, intelligent diagnosis, and root cause localization of power supply anomalies, transforming the complex fault investigation process into an intelligent processing flow.

[0004] To achieve the above objectives, the present invention provides a method for fault recording diagnosis of intelligent power supply units, comprising: The data acquisition circuit synchronously samples the electrical parameters of the intelligent power supply unit and stores the obtained sampled data in a preset ring waveform storage area in real time. The sampling data is periodically identified by the edge analysis engine, and a steady-state waveform characteristic baseline of the intelligent power supply unit is established. When the sampled data of the current period is detected to deviate from the steady-state waveform characteristic baseline, the circular waveform storage area is controlled to perform a freeze operation, and the sampled data is subjected to waveform deviation identification to obtain deviation waveform recording data; The edge analysis engine extracts the deviation waveform features of the deviation waveform data, and performs fault identification on the deviation waveform features according to the preset fault feature library to obtain fault diagnosis information.

[0005] Furthermore, the step of synchronously sampling the electrical parameters of the intelligent power supply unit through the data acquisition circuit and storing the obtained sampling data in a preset ring waveform storage area in real time includes: The data acquisition circuit samples the voltage and current of the multiple output terminals of the intelligent power supply unit according to a preset sampling frequency to obtain load electrical data. Add the corresponding channel identifier and sampling time to each of the load electrical data to obtain the sampled data; Obtain the write pointer of the circular waveform storage area, and write the sampled data sequentially into the storage unit pointed to by the write pointer; After each sampled data is written, the write pointer is moved forward by one memory cell, and when it reaches the end address of the circular waveform storage area, the write position is reset to the starting address to cyclically overwrite the initial sampled data.

[0006] Furthermore, the step of periodically identifying the sampled data through an edge analysis engine and establishing a steady-state waveform characteristic baseline for the intelligent power supply unit includes: Based on a preset detection cycle, the edge analysis engine extracts the extreme values ​​of voltage fluctuations and current fluctuations of the sampled data from the ring waveform storage area to form a waveform feature vector; The waveform feature vector is compared with the standard steady-state vector in the preset steady-state feature table. When all the calculated difference values ​​are less than the steady-state threshold of the preset steady-state feature table, the waveform feature vector is stored in the steady-state feature table. When the waveform feature vectors stored in the steady-state feature table reach a preset number of learning samples, statistical calculations and benchmark construction are performed based on all the waveform feature vectors to obtain the steady-state waveform feature baseline.

[0007] Further, when the sampled data of the current period is detected to deviate from the steady-state waveform characteristic baseline, the circular waveform storage area is controlled to perform a freeze operation, and the sampled data is subjected to waveform deviation identification to obtain deviation waveform recording data, including: Extract the sampling data of the current period from the ring waveform storage area and monitor whether a protection action signal is received; If the protection action signal is received, a freeze command is sent to the ring waveform storage area; If the protection action signal is not received, the deviation amplitude between the sampled data of the current period and the steady-state waveform characteristic baseline is detected. When the deviation amplitude exceeds the preset freeze threshold, the freeze command is sent to the ring waveform storage area. In response to the freeze command, the circular waveform storage area is frozen, and the sampled data is subjected to waveform deviation identification to obtain deviation waveform recording data.

[0008] Further, the step of freezing the circular waveform storage area in response to the freeze command and performing waveform deviation identification on the sampled data to obtain deviation waveform recording data includes: In response to the freeze command, the data writing operation in the ring waveform storage area is stopped, and the fault reference storage bit is identified from the storage location in the ring waveform storage area according to the freeze command; The deviation is calculated based on the preset recording time parameters and the sampling period of the sampled data to obtain the forward offset and the backward offset. The read start position is determined forward from the fault reference storage bit based on the forward offset, and the read end position is determined backward from the fault reference storage bit based on the backward offset. Starting from the read start position, the sampled data at each of the storage positions are traversed sequentially until the read end position is reached. All the read sampled data are then integrated to obtain the offset waveform data.

[0009] Further, the step of extracting the deviation waveform features of the deviation waveform data through the edge analysis engine, and performing fault identification on the deviation waveform features according to a preset fault feature library to obtain fault diagnosis information includes: The deviation waveform data is input into the diagnostic model of the edge analysis engine, and the feature extraction layer in the diagnostic model is used to extract features from the deviation waveform data to obtain a set of feature parameters. The feature recognition layer of the diagnostic model matches the feature parameter set with a preset fault feature library to obtain the fault type. The location determination layer of the diagnostic model extracts the corresponding current amplitude parameters and waveform feature parameters from the feature parameter set according to the fault type, and performs location identification on the current amplitude parameters and waveform feature parameters according to preset positioning conditions to obtain the fault location. The fault type and the fault location are combined for diagnosis to generate fault diagnosis information.

[0010] Further, the step of matching the feature parameter set with a preset fault feature library through the feature recognition layer of the diagnostic model to obtain the fault type includes: Extract combined feature parameters, including short-circuit current rise slope parameter, zero-sequence current feature parameter, and motor starting impulse waveform parameter, from the feature parameter set; Obtain type feature templates from the fault feature library, and match the combined feature parameters with the type feature templates; If all parameters of the combined feature parameters match the type feature template, then the fault type of the type feature template is identified; otherwise, type feature templates that match any two parameters of the combined feature parameters are selected and set as candidate templates. The matching degree of all candidate templates is calculated according to the preset matching rules, the candidate template with the highest matching degree is selected, and the fault type of the candidate template is identified.

[0011] The present invention also provides an intelligent power supply unit fault recording diagnostic device, applied to the intelligent power supply unit fault recording diagnostic method described in any one of the above, comprising: The acquisition module is used to synchronously sample the electrical parameters of the intelligent power supply unit through the data acquisition circuit, and store the obtained sampled data in a preset ring waveform storage area in real time. An analysis module is used to perform periodic identification on the sampled data through an edge analysis engine and establish a steady-state waveform characteristic baseline of the intelligent power supply unit. The association module is used to control the circular waveform storage area to perform a freeze operation when it is detected that the sampled data of the current period deviates from the steady-state waveform characteristic baseline, and to identify the deviation waveform of the sampled data to obtain the deviation waveform recording data. The processing module is used to extract the deviation waveform features of the deviation waveform data through the edge analysis engine, and to identify faults in the deviation waveform features according to a preset fault feature library to obtain fault diagnosis information.

[0012] This invention also provides an intelligent power supply unit fault recording and diagnostic system, comprising: Memory, used to store programs; A processor is configured to execute the program to implement the various steps of the intelligent power supply unit fault recording diagnostic method as described in any of the preceding claims.

[0013] The present invention also provides a storage medium storing computer instructions for causing a computer to perform any of the methods described above.

[0014] The present invention provides a method, device, system, and storage medium for fault recording diagnosis of intelligent power supply units, which has the following beneficial effects: By synchronously sampling voltage and current through a high-precision data acquisition circuit and combining it with an edge analysis engine to perform routine monitoring and establish a steady-state waveform characteristic baseline, this invention can automatically capture and save complete waveform data when a fault occurs. This allows for the generation of diagnostic information including fault type and location, shifting from passively recording events to proactively diagnosing root causes. This significantly reduces reliance on the technical expertise of maintenance personnel and effectively shortens the mean time to recovery. Secondly, by extracting characteristic parameters such as the short-circuit current rise slope, zero-sequence current characteristics, and motor starting impact waveform from the deviation waveform data and matching them with a fault feature database, this invention can accurately distinguish the location of the fault. This avoids the limitations of traditional equipment, which can only identify power supply anomalies but cannot pinpoint the specific fault location, reducing the blindness of on-site troubleshooting. Furthermore, through long-term trend analysis of waveform characteristics, this invention can identify gradual-type hidden dangers in advance, transforming the traditional passive response maintenance model into proactive prevention and improving the operational reliability of the power supply system. Attached Figure Description

[0015] Figure 1 This is a flowchart of a fault recording diagnosis method for an intelligent power supply unit provided by the present invention; Figure 2 This invention also provides a structural diagram of an intelligent power supply unit fault recording diagnostic device; Figure 3 The present invention also provides a structural diagram of an intelligent power supply unit fault recording and diagnostic system.

[0016] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0018] The present invention will now be further described in conjunction with the accompanying drawings and specific embodiments.

[0019] Reference Figure 1 As shown, the present invention provides 1. a method for fault recording diagnosis of an intelligent power supply unit, characterized in that it includes: Step S1: The electrical parameters of the intelligent power supply unit are synchronously sampled through the data acquisition circuit, and the obtained sampled data is stored in the preset ring waveform storage area in real time. Specifically, the intelligent power supply unit is equipped with a high-sampling-rate data acquisition circuit, which is connected to each output terminal of the power supply unit. When the power supply unit enters normal operation, the data acquisition circuit synchronously samples the instantaneous voltage and current values ​​of each output terminal according to a preset sampling frequency (e.g., once per microsecond or millisecond). After each sampling, the data acquisition circuit adds a corresponding channel identifier and sampling timestamp to the sampled value of each output terminal, forming a sampled data. This sampled data is sent to a preset circular waveform storage area in chronological order of sampling time. The circular waveform storage area physically represents a continuous memory space, which maintains a write pointer. When sampled data arrives, the current position of the write pointer is obtained, and the data is written to the memory cell pointed to by the write pointer. After the write operation is completed, the write pointer automatically moves forward one memory cell. When the write pointer moves beyond the end address of the circular waveform storage area, the write pointer is reset to the beginning address of the storage area, thereby achieving cyclic storage of new sampled data and overwriting of the oldest sampled data. In this way, the circular waveform storage area always stores continuous sampled data from the most recent period.

[0020] Step S2: The sampling data is periodically identified using an edge analysis engine, and a steady-state waveform characteristic baseline of the intelligent power supply unit is established; Specifically, during normal operation of the power supply unit, the edge analysis engine extracts sampled data from the ring waveform storage area according to a preset detection cycle (e.g., every few seconds or minutes). For the extracted sampled data, the edge analysis engine calculates the extreme values ​​of voltage fluctuations (including maximum and minimum voltage values) and current fluctuations (including maximum and minimum current values) for each output channel within that cycle, and combines these statistical values ​​into a waveform feature vector. The edge analysis engine then performs a difference calculation on each waveform feature vector generated in the current cycle and the standard steady-state vectors already stored in the steady-state feature table. The steady-state feature table is empty during the initial system startup. Upon initial startup or a load change, the edge analysis engine stores the first waveform feature vector determined to be in a steady state in this table as the initial standard. In subsequent detection cycles, if the difference between the current waveform feature vector and all stored vectors in the steady-state feature table is less than a preset steady-state threshold, the current cycle is determined to be a steady-state operating cycle, and the current waveform feature vector is stored in the steady-state feature table. As the operating time accumulates, the number of waveform feature vectors stored in the steady-state feature table gradually increases. Once the number of stored waveform feature vectors reaches the preset number of learning samples, the edge analysis engine performs statistical calculations on all waveform feature vectors in the table to obtain the average range of voltage fluctuation and average range of current fluctuation for each output channel under steady-state operation. The statistical results are then used as the baseline of the steady-state waveform characteristics of the intelligent power supply unit under the current load condition.

[0021] Step S3: When it is detected that the sampled data of the current period deviates from the steady-state waveform characteristic baseline, the circular waveform storage area is controlled to perform a freeze operation, and the sampled data is subjected to waveform deviation identification to obtain deviation waveform recording data; Specifically, during the continuous operation of the power supply unit, the edge analysis engine extracts the sampling data of the current cycle from the ring waveform storage area according to a predetermined monitoring cycle and performs deviation detection on the data. In practice, the edge analysis engine compares the instantaneous voltage and current values ​​of each output channel in the current cycle's sampling data with the normal fluctuation range of the corresponding channel in the steady-state waveform characteristic baseline. If the voltage or current value of any channel exceeds the normal fluctuation range specified by the baseline, it is determined that the sampling data of the current cycle deviates from the steady-state waveform characteristic baseline, and the sampling data of that cycle is marked as candidate abnormal data. Simultaneously, the edge analysis engine continuously monitors whether it receives protection action signals from external inputs of the power supply unit. These signals could be, for example, circuit breaker trip signals or residual current device (RCD) action signals. When a protection action signal is received, regardless of whether the current sampling data deviates from the baseline, the waveform recording process is immediately triggered. After determining that an anomaly has occurred or receiving a protection action signal, the edge analysis engine sends a freeze command to the ring waveform storage area. Upon receiving a freeze command, the circular waveform storage area immediately stops moving its write pointer, preventing subsequent sampled data from overwriting existing data in the storage area. This ensures the complete preservation of the continuous sampled data frame sequence stored before the freeze command was issued. After the freeze operation is complete, the edge analysis engine identifies the location of the faulty sampled data frame that triggered the freeze command from the frozen circular waveform storage area. Using this location as a reference, it extracts a continuous sampled data frame sequence, including the sequence from the first preset duration before the fault occurred to the second preset duration after the fault occurred. This sequence is then used as the offset waveform recording data.

[0022] Step S4: Extract the deviation waveform features of the deviation waveform data through the edge analysis engine, and identify the faults of the deviation waveform features according to the preset fault feature library to obtain fault diagnosis information.

[0023] Specifically, after obtaining the deviation waveform recording data, the edge analysis engine calls its internally integrated diagnostic model to process the data. The diagnostic model analyzes the voltage and current waveforms of each path in the deviation waveform recording data, extracting waveform feature parameters that can characterize fault characteristics, including but not limited to short-circuit current rise slope, zero-sequence current feature parameters, and motor starting impulse waveform parameters, and combines these parameters to form a feature parameter set. The diagnostic model matches the extracted feature parameter set with the baseline feature templates of various faults stored in a pre-defined fault feature library. The fault feature library predefines feature templates for various fault types, such as load-side short-circuit fault templates, motor stall overcurrent fault templates, poor line contact fault templates, and lightning surge fault templates, etc. Each template contains the standard feature range of that type of fault in multiple feature dimensions. By comparing the matching degree of each parameter in the feature parameter set with each type of fault template, the diagnostic model identifies the fault type to which the current deviation waveform recording data belongs. After determining the fault type, the diagnostic model further extracts parameters related to the fault location from the feature parameter set based on the location determination rules corresponding to that fault type. These parameters include, for example, the current amplitude at a specific moment and waveform distortion characteristics. The model compares these parameters with preset location thresholds to infer the specific location of the fault, determining whether it is an internal fault of the power supply unit, a fault in the middle section of the output line, or a fault at the load device port. After completing fault type identification and fault location inference, the edge analysis engine integrates the fault type and fault location information to generate fault diagnosis information containing complete diagnostic conclusions. This diagnostic information can be displayed locally or pushed to the maintenance terminal via the network, providing direct evidence for fault troubleshooting.

[0024] This invention provides an intelligent power supply unit fault recording and diagnostic method. It synchronously samples voltage and current using a high-precision data acquisition circuit and combines this with an edge analysis engine to perform routine monitoring and establish a steady-state waveform characteristic baseline. This allows for the automatic capture and saving of complete waveform data when a fault occurs, generating diagnostic information including fault type and location. This represents a shift from passively recording events to proactively diagnosing root causes, significantly reducing reliance on the technical expertise of maintenance personnel and effectively shortening the mean time to recovery. Secondly, by extracting characteristic parameters such as the short-circuit current rise slope, zero-sequence current characteristics, and motor starting impulse waveform from the deviation waveform data and matching them with a fault feature database, this invention can accurately distinguish the location of the fault, avoiding the limitations of traditional equipment that can only identify power supply anomalies but cannot pinpoint the specific fault point, thus reducing the blindness of on-site troubleshooting. Furthermore, through long-term trend analysis of waveform characteristics, this invention can identify gradual-type hidden dangers in advance, transforming the traditional passive response maintenance mode into proactive prevention and improving the operational reliability of the power supply system.

[0025] 2. The intelligent power supply unit fault recording diagnosis method according to claim 1, characterized in that, the step of synchronously sampling the electrical parameters of the intelligent power supply unit through a data acquisition circuit and storing the obtained sampling data in a preset ring waveform storage area in real time includes: The data acquisition circuit samples the voltage and current of the multiple output terminals of the intelligent power supply unit according to a preset sampling frequency to obtain load electrical data. Specifically, after the intelligent power supply unit is powered on, the data acquisition circuit starts operating according to a pre-set sampling frequency. For each output terminal, the data acquisition circuit simultaneously reads the voltage and current values ​​at each sampling moment, and quantizes the read values ​​to form digital quantities that can be recognized by the processor. These digital quantities constitute the original load electrical data, reflecting the power supply status of each output terminal at that moment. After each sampling, the data acquisition circuit generates a set of load electrical data containing the voltage and current values ​​of all output terminals, and this set of data corresponds one-to-one with the sampling moment.

[0026] Add the corresponding channel identifier and sampling time to each of the load electrical data to obtain the sampled data; Obtain the write pointer of the circular waveform storage area, and write the sampled data sequentially into the storage unit pointed to by the write pointer; Specifically, after assembling the sampled data, the current write pointer of the circular waveform storage area is obtained. This write pointer holds the address information of the next free storage unit. The first sampled data is copied to the storage unit pointed to by the write pointer, completing the physical storage of the data. The next sampled data to be written is obtained, and the above write operation is repeated. Throughout the writing process, the chronological order of the sampled data is followed, ensuring that the data collected earlier is stored in the storage unit with the lower address, and the data collected later is stored in the storage unit with the higher address. In this way, the sampled data stored in the circular waveform storage area maintains a consistent correspondence with the time axis. The write operation continues until all the sampled data in the current batch is stored in the corresponding storage unit.

[0027] After each sampled data is written, the write pointer is moved forward by one memory cell, and when it reaches the end address of the circular waveform storage area, the write position is reset to the starting address to cyclically overwrite the initial sampled data.

[0028] The method provided in this embodiment ensures the consistency of electrical parameters across all channels in the time dimension by simultaneously acquiring instantaneous voltage and current values ​​at each output terminal through a synchronous sampling mechanism in the data acquisition circuit. By adding channel identifiers and sampling times to the load electrical data, each sampled data entry has a clear source identifier and time stamp, ensuring data traceability during storage and retrieval. By employing write pointers to manage write operations in the circular waveform storage area, ordered storage and automatic cyclic overwriting of sampled data are achieved, continuously retaining complete waveform data from the most recent time period within a limited memory space. A pointer reset mechanism automatically restores the write position to the starting address, enabling cyclic use of the circular storage area and ensuring immediate freezing of the storage area in the event of a fault.

[0029] 3. The intelligent power supply unit fault recording diagnosis method according to claim 1, characterized in that, the step of periodically identifying the sampled data through an edge analysis engine and establishing a steady-state waveform characteristic baseline of the intelligent power supply unit includes: Based on a preset detection cycle, the edge analysis engine extracts the extreme values ​​of voltage fluctuations and current fluctuations of the sampled data from the ring waveform storage area to form a waveform feature vector; The waveform feature vector is compared with the standard steady-state vector in the preset steady-state feature table. When all the calculated difference values ​​are less than the steady-state threshold of the preset steady-state feature table, the waveform feature vector is stored in the steady-state feature table. Specifically, after the intelligent power supply unit enters normal operation, the edge analysis engine starts a timer according to a preset detection cycle. At the end of each detection cycle, the edge analysis engine reads all the sampled data stored in that cycle from the ring waveform storage area. For each output channel, the edge analysis engine traverses all sampled data frames in that cycle, finding the maximum and minimum voltage values, as well as the maximum and minimum current values, for that channel. For power supply units with multiple output channels, the edge analysis engine repeats the above operation until the extreme values ​​of voltage and current fluctuations for each output channel have been extracted. After the extreme value extraction is completed, the edge analysis engine arranges and combines the maximum and minimum voltage values, maximum and minimum current values ​​of each channel according to a fixed channel order, forming a multi-dimensional data set, which is the waveform feature vector for the current detection cycle.

[0030] When the waveform feature vectors stored in the steady-state feature table reach a preset number of learning samples, statistical calculations and benchmark construction are performed based on all the waveform feature vectors to obtain the steady-state waveform feature baseline.

[0031] Specifically, after each new waveform feature vector is added to the steady-state feature table, the edge analysis engine checks whether the total number of vectors stored in the current table has reached the preset number of learning samples. If the total number of vectors has not reached the number of learning samples, the edge analysis engine continues to wait for waveform feature vectors that meet the conditions to be added in subsequent detection cycles. When the total number of vectors reaches the number of learning samples for the first time, the edge analysis engine starts the baseline establishment procedure. The edge analysis engine reads all stored waveform feature vectors from the steady-state feature table in sequence and processes the maximum voltage, minimum voltage, maximum current, and minimum current for each output channel separately. For each type of extreme value data, the edge analysis engine collects the extreme value of the corresponding channel from all waveform feature vectors to form a dataset of that extreme value. By performing statistical calculations on each dataset, the central tendency value and dispersion value of the extreme value are obtained, thereby determining the normal fluctuation range of the maximum and minimum voltage values ​​and the normal fluctuation range of the maximum and minimum current values ​​for each output channel under steady-state operation. After completing the statistical calculations for all channels, the edge analysis engine integrates the voltage fluctuation range and current fluctuation range of each channel into a steady-state waveform feature baseline.

[0032] The method provided in this embodiment achieves continuous tracking and feature condensation of the operating state by setting a preset detection cycle, enabling the edge analysis engine to periodically extract the extreme values ​​of voltage and current fluctuations from the sampled data and form waveform feature vectors. By calculating the difference between the current waveform feature vector and the standard steady-state vectors in the steady-state feature table, and storing all vectors with differences less than the steady-state threshold in the steady-state feature table, the stored samples are all feature vectors highly consistent with historical steady-state states, eliminating random fluctuation interference and improving baseline purity. When the number of vectors in the table reaches a preset number of learning samples, statistical calculations and benchmark construction are performed on all vectors to obtain the voltage and current fluctuation range of each output channel under steady-state operation, forming a steady-state waveform feature baseline. This baseline accurately reflects the normal electrical parameter fluctuation characteristics of the power supply unit.

[0033] 4. The intelligent power supply unit fault recording diagnosis method according to claim 1, characterized in that, when it is detected that the sampled data of the current period deviates from the steady-state waveform characteristic baseline, the ring waveform storage area is controlled to perform a freeze operation, and the sampled data is subjected to waveform deviation identification to obtain deviation recording data, including: Extract the sampling data of the current period from the ring waveform storage area and monitor whether a protection action signal is received; Among them, the protection action signal refers to the switching signal from outside or inside the power supply unit. This signal is generated when the air switch trips, the residual current device operates, or other protection devices are triggered, and is used to indicate that a protection event requiring immediate response has occurred.

[0034] If the protection action signal is received, a freeze command is sent to the ring waveform storage area; The freeze command refers to the control command sent by the edge analysis engine to the ring waveform storage area. This command is used to instruct the ring waveform storage area to stop the cyclic overwrite operation and retain the currently stored data.

[0035] If the protection action signal is not received, the deviation amplitude between the sampled data of the current period and the steady-state waveform characteristic baseline is detected. When the deviation amplitude exceeds the preset freeze threshold, the freeze command is sent to the ring waveform storage area. Specifically, after the edge analysis engine completes monitoring of the protection action signal, if no state change is detected at the input port, it enters the regular deviation detection process. It retrieves the sampled data for the current cycle from the working buffer and, for each output channel, compares the instantaneous voltage and current values ​​of that channel with the normal fluctuation range of the corresponding channel in the steady-state waveform characteristic baseline. For voltage values, the edge analysis engine determines whether the current instantaneous voltage value exceeds the range defined by the minimum and maximum voltage values ​​in the baseline; for current values, it determines whether the current instantaneous current value exceeds the range defined by the minimum and maximum current values ​​in the baseline. When the voltage or current value of a channel exceeds the corresponding range, the edge analysis engine calculates the difference between the exceeded value and the range boundary value and uses this difference as the instantaneous deviation amplitude of that channel. The edge analysis engine statistically analyzes the deviation amplitudes at all sampling moments within the current cycle and takes the maximum or cumulative value as the overall deviation amplitude for the current cycle. The calculated overall deviation amplitude is compared with a preset freeze threshold. If the overall deviation amplitude exceeds the freeze threshold, it is determined that the sampling data of the current period has significantly deviated from the normal operating state, meeting the conditions for waveform recording triggering. The edge analysis engine generates a freeze command and sends it to the control logic of the ring waveform storage area through the memory controller, triggering the storage area to enter the freeze state.

[0036] In response to the freeze command, the circular waveform storage area is frozen, and the sampled data is subjected to waveform deviation identification to obtain deviation waveform recording data.

[0037] The method provided in this embodiment triggers a freeze command by monitoring the protection action signal and detecting the deviation amplitude of the sampled data from the steady-state waveform characteristic baseline. This enables the system to respond to the immediate action of external protection devices and autonomously identify abnormal changes in electrical parameters, ensuring the reliability of waveform recording startup under various fault scenarios. When the deviation amplitude exceeds a preset freeze threshold, a freeze command is sent, distinguishing the waveform recording triggering condition from the fluctuation range of normal operation, avoiding false triggering caused by normal fluctuations, and ensuring timely capture of genuine anomalies. In response to the freeze command, a freeze operation is performed on the ring waveform storage area, and deviation waveform identification is performed on the sampled data. This ensures that the continuous sampled data stored before the fault occurs is completely preserved, and deviation waveform recording data containing the complete waveforms before and after the fault is obtained through identification and extraction.

[0038] 5. The intelligent power supply unit fault recording diagnosis method according to claim 4, characterized in that, in response to the freeze command, freezing the ring waveform storage area and performing deviation waveform identification on the sampled data to obtain deviation recording data, includes: In response to the freeze command, the data writing operation in the ring waveform storage area is stopped, and the fault reference storage bit is identified from the storage location in the ring waveform storage area according to the freeze command; Here, "storage location" refers to the storage address of each storage unit in the ring waveform storage area, and each storage address holds a sampled data. The fault reference storage bit refers to the specific storage location within the ring waveform storage area of ​​the sampled data frame that triggered the freeze instruction.

[0039] The deviation is calculated based on the preset recording time parameters and the sampling period of the sampled data to obtain the forward offset and the backward offset. Specifically, the preset recording time parameters are read to obtain the specific values ​​of the recording duration before and after the fault. The sampling period of the data acquisition circuit is obtained, which reflects the time interval between two adjacent sampled data frames. The recording duration before the fault is divided by the sampling period to obtain a value representing the number of sampled data frames collected in the time period before the fault occurred. Based on the one-to-one correspondence between storage units and sampled data frames, this value is the number of storage units that need to be traced backward from the fault reference storage position. This value is rounded down and used as the forward offset. Similarly, the edge analysis engine divides the recording duration after the fault by the sampling period to obtain the number of sampled data frames collected after the fault occurred. This value is rounded down and used as the backward offset. If the calculated forward offset is greater than the number of frames actually stored before the fault reference storage position, the number of frames actually existing before that is used as the final forward offset; if the backward offset exceeds the remaining space from the end of the storage area to the fault reference storage position, the actual available space after that is used as the final backward offset.

[0040] The read start position is determined forward from the fault reference storage bit based on the forward offset, and the read end position is determined backward from the fault reference storage bit based on the backward offset. Starting from the read start position, the sampled data at each of the storage positions are traversed sequentially until the read end position is reached. All the read sampled data are then integrated to obtain the offset waveform data.

[0041] Specifically, the edge analysis engine initiates a data read operation, reading the address value of the starting position. The edge analysis engine uses this address as the current access address and retrieves the sampled data frame stored in the memory cell at that address. The retrieved sampled data frame contains the instantaneous voltage and current values, channel identifiers, and sampling timestamps of each output channel at that moment, and this data frame is temporarily stored in the analysis buffer. After completing the reading of the current address, the current address is incremented by the length of one memory cell to obtain the address of the next storage location. During the address increment process, it is continuously checked whether the current address exceeds the end address of the circular waveform storage area. If the current address exceeds the end address, the edge analysis engine resets it to the starting address of the storage area, realizing a circular mapping of addresses. The edge analysis engine repeats the above read operation, appending the sampled data from each storage location to the end of the analysis buffer. After each read, the edge analysis engine compares the current address with the address of the end read position. When the current address equals the end read position, it indicates that the last data frame to be extracted has been read. At this time, the edge analysis engine performs the final read operation, storing the sampled data frame at that position into the analysis buffer. After all reads are completed, the data frames stored in the analysis buffer are arranged in the order they were read. The sampling timestamps of these data frames increment continuously, forming a deviation waveform data. The first frame of this deviation waveform data corresponds to the sampling time corresponding to the first preset duration before the fault occurred, and the last frame corresponds to the sampling time corresponding to the second preset duration after the fault occurred. The middle frame contains the fault trigger time corresponding to the fault reference storage bit.

[0042] The method provided in this embodiment stops data writing in response to a freeze command and identifies the fault reference storage bit, ensuring accurate positioning of the sampled data frame corresponding to the fault trigger moment and providing a precise reference point for waveform extraction. Forward and backward offsets are calculated based on the waveform recording time parameters and sampling period, converting the time requirement into the number of storage unit offsets, thus ensuring the accuracy of the extracted waveform in the time dimension. The start and end positions of the read operation are determined from the fault reference storage bit based on the offsets. Through circular address mapping in the circular storage area, the continuous data region covering the time period before and after the fault is completely located within a limited space. Starting from the start position, each storage location is traversed sequentially until the end position, integrating all read sampled data to obtain offset waveform recording data containing the complete waveform from before to after the fault.

[0043] 6. The intelligent power supply unit fault recording diagnosis method according to claim 1, characterized in that, the step of extracting the deviation waveform features of the deviation recording data through the edge analysis engine, and performing fault identification on the deviation waveform features according to a preset fault feature library to obtain fault diagnosis information, includes: The deviation waveform data is input into the diagnostic model of the edge analysis engine, and the feature extraction layer in the diagnostic model is used to extract features from the deviation waveform data to obtain a set of feature parameters. Specifically, after obtaining the offset waveform data, the edge analysis engine passes the data sequence as input to the diagnostic model. Upon receiving the offset waveform data, the diagnostic model sends it to the input of the feature extraction layer. The feature extraction layer then analyzes the voltage and current waveform sequences of each output channel in the offset waveform data item by item. For the voltage waveform sequence, the feature extraction layer calculates parameters such as the voltage drop rate, voltage dip depth, and voltage recovery time at the moment of fault occurrence; for the current waveform sequence, it calculates parameters such as the current rise rate, current peak value, and current oscillation frequency at the moment of fault occurrence. For different types of potential faults, the feature extraction layer also extracts specific waveform features, including the short-circuit current rise slope, zero-sequence current characteristics during ground faults, impulse waveform characteristics during motor startup, intermittent waveform distortion characteristics caused by poor line contact, and peak voltage characteristics caused by lightning surges. For each extracted feature parameter, the feature extraction layer quantizes and encodes it according to a preset format. After all feature parameters are extracted, the feature extraction layer arranges and combines these parameters in a fixed order to form a multi-dimensional feature parameter set.

[0044] The feature recognition layer of the diagnostic model matches the feature parameter set with a preset fault feature library to obtain the fault type. The location determination layer of the diagnostic model extracts the corresponding current amplitude parameters and waveform feature parameters from the feature parameter set according to the fault type, and performs location identification on the current amplitude parameters and waveform feature parameters according to preset positioning conditions to obtain the fault location. The fault location refers to the identified location where the fault occurred, including types such as internal faults in the power supply unit, faults in the middle section of the output line, and faults at the port of the load equipment.

[0045] Specifically, the location determination layer reads the fault type. Based on this fault type, the layer queries the internally stored location rule table for the corresponding location conditions. The location condition table predefines which current amplitude parameters and waveform characteristic parameters need to be extracted from the feature parameter set for each fault type, and the fault locations corresponding to different value ranges of these parameters. Based on the query results, the location determination layer extracts the specified current amplitude parameters and waveform characteristic parameters from the feature parameter set. For example, for a short-circuit fault, the extracted parameters include the short-circuit current peak value, current rise slope, and voltage drop depth. After extraction, the location determination layer compares the extracted current amplitude parameters and waveform characteristic parameters with the preset threshold ranges in the location conditions. Taking an internal power supply unit fault as an example, the location conditions stipulate that an internal fault is determined when the current amplitude is within a specific range and the waveform has certain characteristics; taking a mid-section fault in the output line as an example, the location conditions stipulate that a line fault is determined when the current amplitude attenuates to a certain level and the waveform exhibits reflection characteristics; taking a load device port fault as an example, the location conditions stipulate that a port fault is determined when the current amplitude is high and the waveform exhibits typical load characteristics. The location determination layer determines which location category the current fault parameters match by comprehensively comparing multiple parameters. After the comparison is complete, the location determination layer uses the successfully matched location category as the identified fault location.

[0046] The fault type and the fault location are combined for diagnosis to generate fault diagnosis information.

[0047] The method provided in this embodiment transforms off-track waveform data into a set of feature parameters through a feature extraction layer, achieving a quantitative conversion from the original waveform to key feature parameters. A feature recognition layer matches the feature parameter set with a fault feature database, enabling automatic fault type determination and avoiding the subjectivity of manual analysis. A location determination layer extracts corresponding current amplitude parameters and waveform feature parameters based on the fault type, and combines this with location conditions for accurate location identification, distinguishing between three types of fault locations: inside the power supply unit, in the middle of the output line, and at the load device port. Integrating the fault type and fault location generates fault diagnosis information, providing maintenance personnel with complete fault conclusions and reducing blind spot investigations and inter-departmental buck-passing. 7. The intelligent power supply unit fault recording diagnosis method according to claim 6, characterized in that, the step of matching the feature parameter set with a preset fault feature library through the feature recognition layer of the diagnostic model to obtain the fault type includes: Extract combined feature parameters, including short-circuit current rise slope parameter, zero-sequence current feature parameter, and motor starting impulse waveform parameter, from the feature parameter set; Obtain type feature templates from the fault feature library, and match the combined feature parameters with the type feature templates; If all parameters of the combined feature parameters match the type feature template, then the fault type of the type feature template is identified; otherwise, type feature templates that match any two parameters of the combined feature parameters are selected and set as candidate templates. Specifically, after completing the item-by-item comparison of all types of feature templates, the feature recognition layer obtains the number of successful matches between each template and the current combined feature parameters. The feature recognition layer first traverses all templates and checks whether there is a template with a number of successful matches equal to three. If such a template exists, it indicates that the current fault waveform completely matches the baseline characteristics of a certain type of fault. The feature recognition layer immediately uses the fault type identifier code corresponding to the template as the recognition result and skips all subsequent matching steps.

[0048] If no perfectly matching template is found after the initial traversal, the feature recognition layer initiates a candidate template filtering process. The feature recognition layer iterates through all templates again, selecting those with two successful matches. For each template that meets the criteria, the feature recognition layer records its template identifier and stores it in a temporary candidate list. During the filtering process, the feature recognition layer also records which two parameters in each candidate template match successfully, and the relative values ​​of these two parameters to the template's standard range. After traversing all templates, the feature recognition layer uses the temporary candidate list as the candidate template set. If the candidate template set is empty, it means that no template achieves a two-parameter match; the feature recognition layer then writes a matching failure flag to the output register, awaiting further processing or default classification.

[0049] The matching degree of all candidate templates is calculated according to the preset matching rules, the candidate template with the highest matching degree is selected, and the fault type of the candidate template is identified.

[0050] The method provided in this embodiment extracts short-circuit current rise slope parameters, zero-sequence current characteristic parameters, and motor starting impulse waveform parameters as combined characteristic parameters, providing a clear basis for fault type identification. By matching the combined characteristic parameters item by item with type feature templates in the fault feature library, a systematic comparison of various fault types is achieved. When all parameters successfully match a certain template, the fault type is directly identified, simplifying the judgment process; when no fully matching template exists, any two templates with successfully matched parameters are selected as candidates, avoiding missed judgments due to parameter fluctuations. The matching degree of the candidate templates is calculated according to preset matching rules, and the highest matching degree is selected to identify the fault type. Even in cases of partial matching, the closest fault category can still be determined through quantitative comparison, ensuring the accuracy and reliability of fault type identification.

[0051] Reference Figure 2 As shown, the present invention also provides an intelligent power supply unit fault recording diagnostic device, applied to any of the above-described intelligent power supply unit fault recording diagnostic methods, comprising: The acquisition module is used to synchronously sample the electrical parameters of the intelligent power supply unit through the data acquisition circuit, and store the obtained sampled data in a preset ring waveform storage area in real time. An analysis module is used to perform periodic identification on the sampled data through an edge analysis engine and establish a steady-state waveform characteristic baseline of the intelligent power supply unit. The association module is used to control the circular waveform storage area to perform a freeze operation when it is detected that the sampled data of the current period deviates from the steady-state waveform characteristic baseline, and to identify the deviation waveform of the sampled data to obtain the deviation waveform recording data. The processing module is used to extract the deviation waveform features of the deviation waveform data through the edge analysis engine, and to identify faults in the deviation waveform features according to a preset fault feature library to obtain fault diagnosis information.

[0052] Reference Figure 3 As shown, the present invention also provides an intelligent power supply unit fault recording and diagnostic system, comprising: Memory, used to store programs; A processor is configured to execute the program to implement the various steps of the intelligent power supply unit fault recording diagnostic method as described in any of the preceding claims.

[0053] In this embodiment, the processor and memory can be connected via a bus or other means. The memory may include volatile memory, such as random access memory; the memory may also include non-volatile memory, such as read-only memory, flash memory, hard disk, or solid-state drive. The processor may be a general-purpose processor, such as a central processing unit, digital signal processor, application-specific integrated circuit, or one or more integrated circuits configured to implement embodiments of the present invention.

[0054] The present invention also provides a storage medium storing computer instructions for causing a computer to perform the method according to any one of the preceding claims.

[0055] It should be noted that those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the system and each module described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0056] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural or procedural transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A fault recording diagnosis method for intelligent power supply units, characterized in that, include: The data acquisition circuit synchronously samples the electrical parameters of the intelligent power supply unit and stores the obtained sampled data in a preset ring waveform storage area in real time. The sampling data is periodically identified by the edge analysis engine, and a steady-state waveform characteristic baseline of the intelligent power supply unit is established. When the sampled data of the current period is detected to deviate from the steady-state waveform characteristic baseline, the circular waveform storage area is controlled to perform a freeze operation, and the sampled data is subjected to waveform deviation identification to obtain deviation waveform recording data; The edge analysis engine extracts the deviation waveform features of the deviation waveform data, and performs fault identification on the deviation waveform features according to the preset fault feature library to obtain fault diagnosis information.

2. The intelligent power supply unit fault recording diagnosis method according to claim 1, characterized in that, The step of synchronously sampling the electrical parameters of the intelligent power supply unit through a data acquisition circuit and storing the obtained sampling data in a preset ring waveform storage area in real time includes: The data acquisition circuit samples the voltage and current of the multiple output terminals of the intelligent power supply unit according to a preset sampling frequency to obtain load electrical data. Add the corresponding channel identifier and sampling time to each of the load electrical data to obtain the sampled data; Obtain the write pointer of the circular waveform storage area, and write the sampled data sequentially into the storage unit pointed to by the write pointer; After each sampled data is written, the write pointer is moved forward by one memory cell, and when it reaches the end address of the circular waveform storage area, the write position is reset to the starting address to cyclically overwrite the initial sampled data.

3. The intelligent power supply unit fault recording diagnosis method according to claim 1, characterized in that, The step of periodically identifying the sampled data using an edge analysis engine and establishing a steady-state waveform characteristic baseline for the intelligent power supply unit includes: Based on a preset detection cycle, the edge analysis engine extracts the extreme values ​​of voltage fluctuations and current fluctuations of the sampled data from the ring waveform storage area to form a waveform feature vector; The waveform feature vector is compared with the standard steady-state vector in the preset steady-state feature table. When all the calculated difference values ​​are less than the steady-state threshold of the preset steady-state feature table, the waveform feature vector is stored in the steady-state feature table. When the waveform feature vectors stored in the steady-state feature table reach a preset number of learning samples, statistical calculations and benchmark construction are performed based on all the waveform feature vectors to obtain the steady-state waveform feature baseline.

4. The intelligent power supply unit fault recording diagnosis method according to claim 1, characterized in that, When the sampled data of the current period is detected to deviate from the steady-state waveform characteristic baseline, the ring waveform storage area is controlled to perform a freeze operation, and the sampled data is subjected to waveform deviation identification to obtain deviation waveform recording data, including: Extract the sampling data of the current period from the ring waveform storage area and monitor whether a protection action signal is received; If the protection action signal is received, a freeze command is sent to the ring waveform storage area; If the protection action signal is not received, the deviation amplitude between the sampled data of the current period and the steady-state waveform characteristic baseline is detected. When the deviation amplitude exceeds the preset freeze threshold, the freeze command is sent to the ring waveform storage area. In response to the freeze command, the circular waveform storage area is frozen, and the sampled data is subjected to waveform deviation identification to obtain deviation waveform recording data.

5. The intelligent power supply unit fault recording diagnosis method according to claim 4, characterized in that, In response to the freeze command, the circular waveform storage area is frozen, and deviation waveform identification is performed on the sampled data to obtain deviation waveform recording data, including: In response to the freeze command, the data writing operation in the ring waveform storage area is stopped, and the fault reference storage bit is identified from the storage location in the ring waveform storage area according to the freeze command; The deviation is calculated based on the preset recording time parameters and the sampling period of the sampled data to obtain the forward offset and the backward offset. The read start position is determined forward from the fault reference storage bit based on the forward offset, and the read end position is determined backward from the fault reference storage bit based on the backward offset. Starting from the read start position, the sampled data at each of the storage positions are traversed sequentially until the read end position is reached. All the read sampled data are then integrated to obtain the offset waveform data.

6. The intelligent power supply unit fault recording diagnosis method according to claim 1, characterized in that, The step involves extracting the deviation waveform features from the deviation waveform recording data using the edge analysis engine, and then performing fault identification on the deviation waveform features based on a preset fault feature library to obtain fault diagnosis information, including: The deviation waveform data is input into the diagnostic model of the edge analysis engine, and the feature extraction layer in the diagnostic model is used to extract features from the deviation waveform data to obtain a set of feature parameters. The feature recognition layer of the diagnostic model matches the feature parameter set with a preset fault feature library to obtain the fault type. The location determination layer of the diagnostic model extracts the corresponding current amplitude parameters and waveform feature parameters from the feature parameter set according to the fault type, and performs location identification on the current amplitude parameters and waveform feature parameters according to preset positioning conditions to obtain the fault location. The fault type and the fault location are combined for diagnosis to generate fault diagnosis information.

7. The intelligent power supply unit fault recording diagnosis method according to claim 6, characterized in that, The step of matching the feature parameter set with a preset fault feature library through the feature recognition layer of the diagnostic model to obtain the fault type includes: Extract combined feature parameters, including short-circuit current rise slope parameter, zero-sequence current feature parameter, and motor starting impulse waveform parameter, from the feature parameter set; Obtain type feature templates from the fault feature library, and match the combined feature parameters with the type feature templates; If all parameters of the combined feature parameters match the type feature template, then the fault type of the type feature template is identified; otherwise, type feature templates that match any two parameters of the combined feature parameters are selected and set as candidate templates. The matching degree of all candidate templates is calculated according to the preset matching rules, the candidate template with the highest matching degree is selected, and the fault type of the candidate template is identified.

8. An intelligent power supply unit fault recording diagnosis device, characterized in that, The fault recording diagnostic method for intelligent power supply units according to any one of claims 1-7 includes: The acquisition module is used to synchronously sample the electrical parameters of the intelligent power supply unit through the data acquisition circuit, and store the obtained sampled data in a preset ring waveform storage area in real time. An analysis module is used to perform periodic identification on the sampled data through an edge analysis engine and establish a steady-state waveform characteristic baseline of the intelligent power supply unit. The association module is used to control the circular waveform storage area to perform a freeze operation when it is detected that the sampled data of the current period deviates from the steady-state waveform characteristic baseline, and to identify the deviation waveform of the sampled data to obtain the deviation waveform recording data. The processing module is used to extract the deviation waveform features of the deviation waveform data through the edge analysis engine, and to identify faults in the deviation waveform features according to a preset fault feature library to obtain fault diagnosis information.

9. An intelligent power supply unit fault recording diagnosis system, characterized in that, include: Memory, used to store programs; A processor is used to execute the program to implement the various steps of the intelligent power supply unit fault recording diagnosis method as described in any one of claims 1-7.

10. A storage medium, characterized by The computer contains computer instructions for causing the computer to perform the method according to any one of claims 1 to 7.

Images