Fault identification method for alarm controller of branching system

Abstract

The application discloses a fault identification method for a branch alarm controller and relates to the technical field of fire safety, which comprises the following steps: collecting working current values and working voltage values of each alarm loop in the branch alarm controller, converting the working current values and the working voltage values into digital signals to obtain current data and voltage data, and judging the working states of the alarm loops by using a preset threshold method; establishing a fault feature vector matrix, inputting the fault feature vector matrix into a pre-trained fault diagnosis model, and outputting a fault type judgment result; and according to the fault type judgment result, calling a corresponding processing scheme from a pre-established fault processing database to complete fault identification. The application realizes the full-process intelligentization from fault detection, diagnosis to processing, improves the operation reliability of the branch alarm controller, reduces the maintenance cost, and has important practical significance for improving the overall performance of the fire safety system.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of fire safety, in particular to a fault identification method for a branch alarm controller. BACKGROUND

[0002] With the continuous improvement of social safety needs, the automatic fire alarm system plays an increasingly important role in building fire safety. As the core component of the automatic fire alarm system, the branch alarm controller connects various fire detectors and other fire-fighting equipment through multiple alarm loops, and monitors the fire status in the building in real time. In recent years, with the expansion of building scale and the improvement of intelligent level, the technical requirements of branch alarm controller in system integration, data processing and fault diagnosis are also increasing. The traditional branch alarm controller mainly uses fault detection method based on current and voltage threshold judgment, which sets fixed judgment threshold to identify basic fault types such as short circuit and open circuit in the loop. However, this simple threshold judgment method is difficult to accurately identify complex faults caused by multiple factors, and is prone to misjudgment when the external environment interference is large, affecting the reliability and stability of the system.

[0003] The existing branch alarm controller fault diagnosis technology has the following shortcomings: first, the traditional fault diagnosis method mainly relies on manual experience for fault analysis and processing, which has low diagnosis efficiency and poor accuracy; second, the existing automatic fault diagnosis system generally uses single threshold judgment method, which cannot effectively identify and classify complex fault types, especially when multiple loops fail at the same time, it is difficult to accurately locate the fault source and give the corresponding processing scheme; third, there is a lack of intelligent fault feature extraction and analysis means, which cannot fully utilize the large amount of historical data generated during system operation to optimize the fault diagnosis strategy; finally, the fault diagnosis results of the existing system lack effective correlation mechanism with the fault processing scheme, resulting in slow fault response speed and poor processing effect.

[0004] In view of the above problems, the present application provides a branch alarm controller fault identification method based on deep learning, which can effectively improve the accuracy and automation level of fault diagnosis. SUMMARY

[0005] In view of the problems of low fault identification accuracy, low diagnosis efficiency and inability to handle complex fault conditions in the existing branch alarm controller fault diagnosis technology, the present application is proposed.

[0006] Therefore, the problem to be solved by the present application is how to provide a fault identification method based on deep learning, to realize the rapid and accurate identification of branch alarm controller faults, classification and intelligent recommendation of processing scheme, so as to improve the automation level and accuracy of fault diagnosis.

[0007] To solve the above technical problems, the present application provides the following technical solutions:

[0008] In a first aspect, the embodiments of the present application provide a fault identification method for a branch alarm controller, which comprises: collecting working current values and working voltage values of each alarm loop in the branch alarm controller, converting the working current values and working voltage values into digital signals to obtain current data and voltage data; judging working states of each alarm loop based on the current data and voltage data by using a preset threshold method, wherein the preset threshold comprises an upper limit current threshold, a lower limit current threshold, an upper limit voltage threshold and a lower limit voltage threshold; establishing a fault feature vector matrix according to the working states, wherein the fault feature vector matrix contains current abnormal flag bits and voltage abnormal flag bits of each alarm loop; inputting the fault feature vector matrix into a pre-trained fault diagnosis model, wherein the fault diagnosis model analyzes the fault feature vector matrix by using a deep learning algorithm and outputs a fault type judgment result; and according to the fault type judgment result, retrieving a corresponding processing scheme from a pre-established fault processing database to complete fault identification.

[0009] As a preferred scheme of the fault identification method for the branch alarm controller, the fault diagnosis model performs forward calculation on data segments, the output layer calculates probability distributions of various fault types by using a softmax function, and the class with the highest probability is taken as the fault type judgment result of the data segment; the judgment results of continuous p data segments are counted by voting to obtain a final fault type judgment result, and probability values of various fault types are recorded as reliability indexes of fault judgment.

[0010] As a preferred scheme of the fault identification method for the branch alarm controller, the data segment is obtained by: performing normalization processing on the fault feature vector matrix, mapping the data to the [0, 1] interval by using a min-max method; cutting the normalized data into data segments with a length of 10 according to time sequences, setting an overlap rate between each data segment, and inputting the data segments into the fault diagnosis model in a sliding window manner.

[0011] As a preferred scheme of the fault identification method of the branch alarm controller of the application, wherein: the construction process of the fault feature vector matrix comprises the following steps: a two-dimensional array structure of NXM is adopted to establish the fault feature vector matrix, wherein N is the total number of alarm loops, and M is the number of feature bits of each alarm loop; in the fault feature vector matrix, each row of data corresponds to the feature data of an alarm loop, the first bit of each row of data is a current overload flag bit, which is set to 1 when the alarm loop is in a current overload state, otherwise it is set to 0; the second bit is a current open circuit flag bit, which is set to 1 when the alarm loop is in a current open circuit state, otherwise it is set to 0; the third bit is a voltage overvoltage flag bit, which is set to 1 when the alarm loop is in a voltage overvoltage state, otherwise it is set to 0; the fourth bit is a voltage undervoltage flag bit, which is set to 1 when the alarm loop is in a voltage undervoltage state, otherwise it is set to 0.

[0012] As a preferred scheme of the fault identification method of the branch alarm controller of the application, wherein: further comprising: statistically analyzing the flag bits in the fault feature vector matrix, and calculating the setting frequency of each flag bit of each alarm loop in the continuous sampling period, and the specific formula is as follows:

[0013]

[0014] w i =e -λ(t-i)

[0015] Wherein, F(t) is the setting frequency of the flag bit of the current t sampling point, n is the length of the sampling window, b i is the flag bit value of the i sampling point in the window, w i is the time weight coefficient of the i sampling point in the window, λ is the time attenuation factor, t is the current sampling point, and i is the sampling point in the window.

[0016] When the setting frequency of a certain flag bit exceeds 80% of the preset setting frequency, the fault type corresponding to this flag bit is determined as a steady-state fault; when the setting frequency of a certain flag bit is between 20% and 80% of the preset setting frequency, the fault type corresponding to this flag bit is determined as a fluctuation fault; when the setting frequency of a certain flag bit is less than 20% of the preset setting frequency, the fault type corresponding to this flag bit is determined as a transient fault; the determination results of the steady-state fault, the fluctuation fault and the transient fault are stored in the fault record database, and the fault feature vector matrix is updated in real time, and the timestamp information of the record update is recorded.

[0017] As a preferred scheme of the fault identification method of the branch alarm controller, before the fault feature vector matrix is established, the working state of each alarm loop is judged by using a preset threshold method; the method for judging the working state of each alarm loop is that the current data and voltage data are subjected to data smoothing processing, and data jitter is eliminated by using a sliding average method to obtain smoothed current average value and voltage average value; the current average value and the voltage average value are respectively compared with corresponding threshold values, when the current average value is greater than the upper limit threshold value of the current or less than the lower limit threshold value of the current, the alarm loop is marked as a current abnormal state; when the voltage average value is greater than the upper limit threshold value of the voltage or less than the lower limit threshold value of the voltage, the alarm loop is marked as a voltage abnormal state, and the time duration that the current average value and the voltage average value exceed the corresponding threshold values is recorded; if the time duration exceeds the preset time, the corresponding abnormal state is marked as a persistent fault; if the time duration is less than or equal to the preset time, the corresponding abnormal state is marked as a transient fault; for the alarm loop of the persistent fault, the percentage of the current average value and the voltage average value deviating from the corresponding threshold value is calculated, and the percentage is stored in the fault feature database as a fault degree parameter.

[0018] As a preferred scheme of the fault identification method of the branch alarm controller, the working state of each alarm loop is judged by using a preset threshold method based on the current data and voltage data; the method for obtaining the current data and voltage data is that the working current value and the working voltage value are converted into digital signals by an analog-to-digital conversion circuit; the analog-to-digital conversion circuit includes a signal conditioning circuit, a sample-and-hold circuit and an analog-to-digital conversion chip; the signal conditioning circuit is used for amplifying and filtering the working current value and the working voltage value; the sample-and-hold circuit is used for sampling and holding the analog signal processed by the signal conditioning circuit; the analog-to-digital conversion chip converts the analog signal output by the sample-and-hold circuit into a digital signal to obtain the current data and the voltage data; the current data and the voltage data are respectively stored in a first data buffer area and a second data buffer area.

[0019] In a second aspect, an embodiment of the present application provides a fault identification system for a branch alarm controller, comprising: a collection module configured to collect working current values and working voltage values of each alarm loop in the branch alarm controller, and convert the working current values and working voltage values into digital signals to obtain current data and voltage data; a judgment module configured to judge working states of each alarm loop based on the current data and voltage data using a preset threshold method, wherein the preset threshold comprises an upper current threshold, a lower current threshold, an upper voltage threshold, and a lower voltage threshold; a construction module configured to construct a fault feature vector matrix according to the working states, wherein the fault feature vector matrix comprises current abnormal flag bits and voltage abnormal flag bits of each alarm loop; a fault diagnosis module configured to input the fault feature vector matrix into a pre-trained fault diagnosis model, and the fault diagnosis model analyzes the fault feature vector matrix by a deep learning algorithm to output a fault type judgment result; and a calling module configured to call a corresponding processing scheme from a pre-established fault processing database according to the fault type judgment result to complete fault identification.

[0020] In a third aspect, an embodiment of the present application provides a computer device, comprising a memory and a processor, and the memory stores a computer program, wherein the processor implements any step of the fault identification method for the branch alarm controller when executing the computer program.

[0021] In a fourth aspect, an embodiment of the present application provides a computer readable storage medium, which stores a computer program, wherein the computer program is executed by a processor to implement any step of the fault identification method for the branch alarm controller.

[0022] The present application has the following beneficial effects: the present application collects and digitizes the current and voltage data of the alarm loop, combines multi-dimensional threshold prediction and feature vector matrix construction, provides high-quality structured input for fault diagnosis, uses a deep learning algorithm to intelligently analyze fault features, can identify complex fault patterns, and significantly improves the accuracy of fault diagnosis, and through the pre-established fault processing database, intelligent matching of fault processing schemes is realized to speed up the fault response speed. BRIEF DESCRIPTION OF DRAWINGS

[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiment description. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings. Among them:

[0024] Fig. 1 is a flowchart of the fault identification method for the branch alarm controller.

[0025] Fig. 2 This is a schematic diagram of a wired alarm controller used for a fault identification method in a wired alarm controller. Detailed Implementation

[0026] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0027] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0028] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0029] Example 1

[0030] Reference Figs. 1-2 This is the first embodiment of the present invention, which provides a fault identification method for a split-line alarm controller, including:

[0031] S1: Collect the operating current and operating voltage values ​​of each alarm circuit in the split-line alarm controller, and convert the operating current and operating voltage values ​​into digital signals to obtain current data and voltage data.

[0032] Specifically, the operating current value is acquired through a sampling resistor set in each alarm loop; the operating voltage value is acquired through a voltage divider circuit; the sampling resistor is connected in series in the alarm loop; the voltage divider circuit is connected in parallel in the alarm loop; the digital signal conversion method is to convert the operating current value and the operating voltage value into digital signals through a high-precision analog-to-digital converter circuit.

[0033] It should be noted that the distributed alarm controller includes a power supply module, a main controller module, a communication module, a loop interface module, an operation module, a display module, and an alarm output module.

[0034] Further, the sampling frequency of the analog-digital conversion circuit is 10 kHz, and the resolution is 12 bits, wherein the analog-digital conversion circuit comprises a signal conditioning circuit, a sample-and-hold circuit, and an analog-digital conversion chip, the signal conditioning circuit is configured to amplify and filter the working current value and the working voltage value, the sample-and-hold circuit is configured to sample and hold the signal processed by the signal conditioning circuit, and the analog-digital conversion chip converts the analog signal output by the sample-and-hold circuit into a digital signal to obtain current data and voltage data; the sampling duration of the current data and the voltage data is 1 minute, the sampling interval is 100 microseconds, and the current data and the voltage data are respectively stored in a first data buffer area and a second data buffer area.

[0035] Further, in the signal conditioning circuit, the working current value obtains a voltage drop through a sampling resistor, the resistance of the sampling resistor is 0.1 ohm, the accuracy is 0.1%, and the temperature coefficient is less than 25 ppm / ℃, the working voltage value obtains a sampling voltage through a high-precision resistance voltage division network with a ratio of 100:1, and the overall accuracy of the voltage division network is better than 0.05%; the voltage drop and the sampling voltage are respectively amplified by an operational amplifier, the gain of the operational amplifier is 10 times, the bandwidth is 1 MHz, and the signals are filtered by a second-order Butterworth low-pass filter, and the cutoff frequency of the low-pass filter is 5 kHz; in the sample-and-hold circuit and the analog-digital conversion chip, the conditioned signal is sampled by the sample-and-hold circuit, the sampling time of the sample-and-hold circuit is 1 microsecond, the holding time is 9 microseconds, and a sampling period of 10 microseconds is formed, the output signal of the sample-and-hold circuit is sent to a 12-bit successive approximation analog-digital converter, the reference voltage of the analog-digital converter is 2.5 V, the conversion time is 8 microseconds, the analog signal is quantized into a digital range value of 0-4095 by the analog-digital converter, the digital signal obtained by conversion is transmitted to a microprocessor through an SPI bus, the microprocessor linearly corrects the digital signal to obtain corrected current data and voltage data.

[0036] S2: based on the current data and the voltage data, a preset threshold method is used to judge the working state of each alarm loop.

[0037] Specifically, the current upper threshold is set to 150% of the rated working current of the alarm loop, the current lower threshold is set to 50% of the rated working current of the alarm loop, the voltage upper threshold is set to 120% of the rated working voltage of the alarm loop, and the voltage lower threshold is set to 80% of the rated working voltage of the alarm loop.

[0038] Further, the method for judging the working state of each alarm loop is that the current data and voltage data are subjected to data smoothing processing, and data jitter is eliminated through a sliding average method to obtain smoothed current average value and voltage average value; the current average value and the voltage average value are respectively judged against corresponding threshold values; for an alarm loop with persistent fault, the percentage of the current average value deviating from the corresponding threshold value and the voltage average value is calculated, and the percentage is stored in the fault feature database as a fault degree parameter.

[0039] Further, when the current average value is greater than the upper limit threshold value of the current or less than the lower limit threshold value of the current, the alarm loop is marked as a current abnormal state; when the voltage average value is greater than the upper limit threshold value of the voltage or less than the lower limit threshold value of the voltage, the alarm loop is marked as a voltage abnormal state, and the time duration length of the current average value and the voltage average value exceeding the corresponding threshold value is recorded; if the time duration length exceeding the threshold value is greater than a preset time, the corresponding abnormal state is marked as a persistent fault; if the time duration length exceeding the threshold value is less than or equal to the preset time, the corresponding abnormal state is marked as a transient fault.

[0040] S3: According to the working state, a fault feature vector matrix is established, wherein the fault feature vector matrix comprises current abnormal flag bits and voltage abnormal flag bits of each alarm loop.

[0041] Specifically, the construction process of the fault feature vector matrix comprises the following steps: a two-dimensional array structure of NXM is adopted to establish the fault feature vector matrix, wherein N is the total number of alarm loops, and M is the number of feature bits of each alarm loop; the flag bits in the fault feature vector matrix are subjected to statistical analysis, and the setting frequency of each flag bit of each alarm loop in a continuous sampling period is calculated, and the specific formula is as follows:

[0042]

[0043] w i =e -λ(t-i)

[0044] wherein F(t) is the flag bit setting frequency of the current t sampling point, n is the sampling window length, b i is the flag bit value of the i sampling point in the window, w i is the time weight coefficient of the i sampling point in the window, λ is a time attenuation factor, t is the current sampling point, and i is the sampling point in the window.

[0045] Further, when the setting frequency of a certain flag bit exceeds 80% of the pre-set setting frequency, the fault type corresponding to the flag bit is determined as a steady-state fault; when the setting frequency of a certain flag bit is between 20% and 80% of the pre-set setting frequency, the fault type corresponding to the flag bit is determined as a fluctuation fault; and when the setting frequency of a certain flag bit is less than 20% of the pre-set setting frequency, the fault type corresponding to the flag bit is determined as a transient fault.

[0046] Further, the determination results of the steady-state fault, the fluctuation fault and the transient fault are stored in a fault record database together with time stamp information, and a fault feature vector matrix is updated in real time, and the time stamp information at the time of updating is recorded.

[0047] Specifically, in the fault feature vector matrix, each row of data corresponds to feature data of an alarm loop, the first bit of each row of data is a current overload flag bit, which is set to 1 when the alarm loop is in a current overload state, and is set to 0 otherwise; the second bit is a current open circuit flag bit, which is set to 1 when the alarm loop is in a current open circuit state, and is set to 0 otherwise; the third bit is a voltage overvoltage flag bit, which is set to 1 when the alarm loop is in a voltage overvoltage state, and is set to 0 otherwise; and the fourth bit is a voltage undervoltage flag bit, which is set to 1 when the alarm loop is in a voltage undervoltage state, and is set to 0 otherwise.

[0048] S4: inputting the fault feature vector matrix into a pre-trained fault diagnosis model, the fault diagnosis model analyzing the fault feature vector matrix through a deep learning algorithm and outputting a fault type judgment result.

[0049] Specifically, the fault diagnosis model adopts a five-layer neural network structure, including an input layer, a hidden layer and an output layer; wherein the number of nodes of the input layer matches the feature data of the fault feature vector matrix, and is fixed at 4xN nodes, N being the total number of alarm loops; the hidden layer contains a plurality of neurons, wherein the neurons adopt a ReLU activation function, and the neurons adopt a residual connection structure to avoid gradient disappearance; the output layer contains 8 nodes, respectively corresponding to 8 working states of short circuit fault, open circuit fault, insulation aging, poor contact, abnormal load, power supply fault, communication fault and normal work;

[0050] Further, the fault diagnosis model is trained with 50,000 groups of labeled historical fault data in the training stage, the Adam optimizer is used for parameter optimization, the initial value of the learning rate is set to 0.001, the learning rate is reduced by 50% every 1000 training rounds, the batch size is set to 64, and the dropout rate of 0.3 is used in the training process to prevent overfitting.

[0051] Further, the fault diagnosis model performs forward calculation on the data segments, the output layer calculates the probability distribution of various fault types through a softmax function, and the class with the highest probability is taken as the fault type judgment result of the data segment; the judgment results of continuous p data segments are voted and counted to obtain the final fault type judgment result, and the probability values of various fault types are recorded as the reliability index of fault judgment.

[0052] Specifically, the data segment acquisition method is to normalize the fault feature vector matrix, map the data to the [0, 1] interval using the min-max method; the normalized data is cut into data segments with a length of 10 according to the time sequence, the overlap rate between each data segment is set, and the fault diagnosis model is input in a sliding window manner.

[0053] S5: According to the fault type judgment result, the corresponding processing scheme is retrieved from the pre-established fault processing database, and the fault position and fault type are displayed on the controller display screen.

[0054] Specifically, the fault processing database adopts a relational database structure and includes three main data tables, namely a fault type table, a processing scheme table, and a fault record table; the fault type table stores fault codes, fault type names, fault levels, and fault descriptions; the processing scheme table stores processing steps, required tools, matters needing attention, and expected repair time; and the fault record table stores fault occurrence time, fault duration, fault position, and processing results.

[0055] Further, after obtaining the fault type judgment result, the system retrieves the corresponding record in the fault type table according to the fault type code to obtain fault level information; when the fault level is an emergency level, the system automatically starts an audible and visual alarm and sends a short message to the mobile phone of a preset maintenance personnel for reminding; the system retrieves the corresponding processing scheme from the processing scheme table and sends the processing scheme content to the controller display screen after organizing the content according to a preset format.

[0056] Further, the embodiment further provides a fault identification system for the alarm controller of the branch system, comprising: a collection module, configured to collect working current values and working voltage values of each alarm loop in the alarm controller of the branch system, and convert the working current values and the working voltage values into digital signals to obtain current data and voltage data; a judgment module, configured to judge working states of each alarm loop based on the current data and the voltage data by using a preset threshold method, wherein the preset threshold comprises an upper limit threshold of current, a lower limit threshold of current, an upper limit threshold of voltage and a lower limit threshold of voltage; a construction module, configured to construct a fault feature vector matrix according to the working states, wherein the fault feature vector matrix comprises current abnormal flag bits and voltage abnormal flag bits of each alarm loop; a fault diagnosis module, configured to input the fault feature vector matrix into a pre-trained fault diagnosis model, and the fault diagnosis model analyzes the fault feature vector matrix by using a deep learning algorithm to output a fault type judgment result; and a calling module, configured to call a corresponding processing scheme from a pre-established fault processing database according to the fault type judgment result to complete fault identification.

[0057] The embodiment further provides a computer device suitable for the fault identification method of the alarm controller of the branch system, comprising a memory and a processor; the memory is configured to store computer executable instructions, and the processor is configured to execute the computer executable instructions to realize the fault identification method of the alarm controller of the branch system proposed in the above embodiment.

[0058] The computer device can be a terminal, and the computer device comprises a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operating system and the computer program in the non-volatile storage medium to run. The communication interface of the computer device is configured to perform wired or wireless communication with external terminals. The wireless communication can be achieved through WIFI, operator network, NFC (near field communication) or other technologies. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen. The input device of the computer device can be a touch layer overlaid on the display screen, or can be a key, trackball or touchpad arranged on the shell of the computer device, or can be an external keyboard, touchpad or mouse, etc.

[0059] The embodiment further provides a storage medium having a computer program stored thereon, and the program is executed by a processor to realize the fault identification method of the alarm controller of the branch system proposed in the above embodiment.

[0060] The storage medium proposed in the embodiment belongs to the same inventive concept as the data storage method proposed in the above embodiment, and technical details not described in detail in the embodiment can be referred to the above embodiment, and the embodiment has the same beneficial effects as the above embodiment.

[0061] Embodiment 2

[0062] Referring to Table 1, a second embodiment of the present application is provided, which provides a fault identification method for a branch alarm controller. In order to verify the beneficial effects of the present application, economic benefit calculation and simulation experiments are used for scientific demonstration.

[0063] Specifically, the fire automatic alarm system of a certain large commercial complex is selected as the test object. The complex is divided into 12 fire zones, each zone is equipped with one branch alarm controller, each controller is connected to 8 alarm circuits, and the number of fire detectors on each circuit is different from 20 to 35; the FCP-9000 branch alarm controller is used in the test process, which has a 16-bit high-precision analog-to-digital converter and a sampling frequency of 1 kHz; the experimental test lasts for 30 days, during which the system response data under various fault conditions are simulated and recorded.

[0064] Further, in the data acquisition stage, high-precision current probes and voltage probes are used to collect the working current and voltage signals of each alarm circuit. The measurement range of the current probe is 0-500mA, and the accuracy is ±0.1%; the measurement range of the voltage probe is 0-30V, and the accuracy is ±0.05%. The collected analog signals are amplified and filtered by a signal conditioning circuit, and then converted to digital signals by a 16-bit ADC, with a sampling rate of 1 kHz. The system sets the warning threshold parameters as follows: the upper limit threshold of current is 500mA, the lower limit threshold of current is 5mA, the upper limit threshold of voltage is 28V, and the lower limit threshold of voltage is 20V.

[0065] Further, after normalizing the collected data, the data is divided into data segments with a length of 10 using the sliding window method, and the overlap rate of adjacent data segments is set to 50%. The fault feature vector matrix adopts a two-dimensional array structure of 96x4 (12 controllers x 8 circuits = 96 rows, 4 feature bits = 4 columns). The deep learning model adopts a 5-layer fully connected neural network structure, uses the Adam optimizer, sets the learning rate to 0.001, the batch size to 64, and the training rounds to 100. The fault type determination uses a voting statistics method for 5 consecutive data segments, and the confidence threshold is set to 0.85.

[0066] Specifically, as shown in Table 1, according to the table data analysis, during the 30-day continuous monitoring process, the test results of the six typical fire compartments show that the system has excellent performance. In terms of the number of detector configurations, the number of detectors in each fire compartment is distributed between 156 and 182, among which FD-A04 compartment is configured with the most, reaching 182, and FD-A03 compartment is configured with the least, being 156. This distribution difference is mainly the result of reasonable configuration based on the spatial characteristics and fire protection requirements of each area. In terms of fault identification performance, the system exhibits extremely high accuracy, with the fault identification accuracy of the six compartments all maintaining above 98.8%, among which FD-A04 compartment performs the most outstandingly, with an accuracy of 99.3%, which is mainly due to the high-precision data acquisition equipment and the optimization of deep learning algorithm.

[0067] Table 1 Test data table

[0068]

[0069] Further, in terms of system response performance, the fault response time of each compartment is controlled within the range of 355-365 ms, which is significantly better than the response speed of seconds of traditional systems. Among them, FD-A03 compartment achieves the fastest response time of 355 ms due to the relatively small number of detectors. In terms of system reliability, the availability index of all compartments remains above 99.92%, fully demonstrating the stability of the system. It is particularly noteworthy that the false alarm rate of the system is controlled at a low level, with the lowest being 0.14% (FD-A04 compartment) and the highest being 0.19% (FD-B01 compartment), which is of great significance for reducing system operation and maintenance costs and improving alarm reliability.

[0070] Further, through statistical analysis of the frequency of faults, it is found that during the 30-day test period, the number of faults recorded by each compartment is between 36 and 45, with an average of about 1.3 faults per day per compartment. Among them, FD-A03 compartment records the most faults, reaching 45, while FD-B01 compartment records the least, being 36. This fault distribution rule is consistent with the actual operation of large commercial complexes, indicating that the system can effectively identify and record various fault conditions. In summary, the above data shows that the system has good performance indicators in terms of fault identification accuracy, response timeliness, and operation reliability, fully meeting the actual application requirements and having significant application promotion value.

[0071] It should be noted that the above examples are only used to illustrate the technical solutions of the present application but not limit the present application. Although the present application is described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present application can be modified or equivalently replaced, without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A fault identification method for a distributed alarm controller, characterized in that: include, The operating current and operating voltage values ​​of each alarm circuit in the split-line alarm controller are collected, and the operating current and operating voltage values ​​are converted into digital signals to obtain current data and voltage data. Based on the current and voltage data, the operating status of each alarm circuit is determined by a preset threshold method, wherein the preset thresholds include an upper current threshold, a lower current threshold, an upper voltage threshold, and a lower voltage threshold. Based on the operating state, a fault feature vector matrix is ​​established, wherein the fault feature vector matrix includes the current abnormality flag bit and the voltage abnormality flag bit of each alarm circuit; The fault feature vector matrix is ​​input into a pre-trained fault diagnosis model, which analyzes the fault feature vector matrix using a deep learning algorithm and outputs a fault type judgment result. Based on the fault type determination result, the corresponding processing solution is retrieved from the pre-established fault processing database to complete the fault identification; The process of constructing the fault feature vector matrix includes the following steps: A fault feature vector matrix is ​​established using an N×M two-dimensional array structure, where N is the total number of alarm circuits and M is the number of feature bits for each alarm circuit. In the fault feature vector matrix, each row of data corresponds to the feature data of an alarm circuit. The first bit of each row of data is the current overload flag, which is set to 1 when the alarm circuit is in a current overload state, and set to 0 otherwise; the second bit is the current disconnection flag, which is set to 1 when the alarm circuit is in a current disconnection state, and set to 0 otherwise; the third bit is the voltage overvoltage flag, which is set to 1 when the alarm circuit is in a voltage overvoltage state, and set to 0 otherwise; the fourth bit is the voltage undervoltage flag, which is set to 1 when the alarm circuit is in a voltage undervoltage state, and set to 0 otherwise. Statistical analysis is performed on the flag bits in the fault feature vector matrix to calculate the setting frequency of each flag bit in each alarm circuit within a continuous sampling period. The specific formula is as follows: in, Let n be the bit frequency of the current sampling point t, and n be the sampling window length. This is the flag value of the i-th sampling point within the window. The time weighting coefficient for the i-th sampling point within the window. Here, t is the time decay factor, t is the current sampling point, and i is the sampling point within the window; When the setting frequency of a certain flag exceeds 80% of the preset setting frequency, the fault type corresponding to this flag is determined to be a steady-state fault; when the setting frequency of a certain flag is between 20% and 80% of the preset setting frequency, the fault type corresponding to this flag is determined to be a fluctuating fault; when the setting frequency of a certain flag is less than 20% of the preset setting frequency, the fault type corresponding to this flag is determined to be a transient fault. The determination results of steady-state faults, fluctuation faults, and transient faults, along with timestamp information, are stored in the fault record database, and the fault feature vector matrix is ​​updated in real time, recording the timestamp information at the time of the update.

2. The fault identification method for a split-line alarm controller as described in claim 1, characterized in that: The fault diagnosis model performs forward computation on the data segment, and the output layer calculates the probability distribution of various fault types through the softmax function, taking the category with the highest probability as the fault type judgment result for this data segment. The judgment results of more than p consecutive data segments are voted and statistically analyzed to obtain the final fault type judgment result. At the same time, the probability values ​​of each fault type are recorded as the reliability index of the fault judgment.

3. The fault identification method for a split-line alarm controller as described in claim 2, characterized in that: The method for obtaining the data fragment is as follows: After normalizing the fault feature vector matrix, the min-max method is used to map the data to the [0,1] interval; The normalized data is divided into data segments of length 10 according to the time series, and the overlap rate between each data segment is set. The data segments are then input into the fault diagnosis model in sequence through a sliding window.

4. The fault identification method for the split-line alarm controller as described in claim 3, characterized in that: Before establishing the fault feature vector matrix, the operating status of each alarm circuit must be determined using a preset threshold method; the method for determining the operating status of each alarm circuit is as follows: The current and voltage data are smoothed, and data jitter is eliminated by moving average method to obtain smoothed average current and average voltage. The average current and average voltage are compared with their corresponding thresholds. When the average current is greater than the upper current threshold or less than the lower current threshold, the alarm circuit is marked as having an abnormal current state. When the average voltage is greater than the upper voltage threshold or less than the lower voltage threshold, the alarm circuit is marked as having an abnormal voltage state, and the duration of the time when the average current and average voltage exceed the corresponding threshold is recorded. If the duration of the out-of-time exceeds the preset time, the corresponding abnormal state will be marked as a persistent fault; if the duration of the out-of-time is less than or equal to the preset time, the corresponding abnormal state will be marked as a transient fault. For alarm circuits with persistent faults, calculate the percentage by which the average current and average voltage deviate from the corresponding threshold, and store this percentage as a fault severity parameter in the fault feature database.

5. The fault identification method for a split-line alarm controller as described in claim 4, characterized in that: The operating status of each alarm circuit is determined using a preset threshold method based on the current and voltage data; the method for obtaining the current and voltage data is as follows: The operating current and operating voltage values ​​are converted into digital signals by an analog-to-digital converter (ADC). The ADC includes a signal conditioning circuit, a sample-and-hold circuit, and an ADC chip. The signal conditioning circuit amplifies and filters the operating current and operating voltage values. The sample-and-hold circuit samples and holds the analog signal processed by the signal conditioning circuit. The ADC chip converts the analog signal output from the sample-and-hold circuit into a digital signal to obtain current and voltage data. The current and voltage data are stored in a first data buffer and a second data buffer, respectively.

6. A fault identification system for a split-wire alarm controller, based on the fault identification method for a split-wire alarm controller according to any one of claims 1 to 5, characterized in that: include, The acquisition module is used to acquire the operating current and operating voltage values ​​of each alarm circuit in the split-line alarm controller, and convert the operating current and operating voltage values ​​into digital signals to obtain current data and voltage data. The judgment module uses a preset threshold method to determine the working status of each alarm circuit based on the current data and voltage data. The preset thresholds include an upper current threshold, a lower current threshold, an upper voltage threshold, and a lower voltage threshold. The construction module is used to establish a fault feature vector matrix based on the working state, wherein the fault feature vector matrix includes current abnormality flag bits and voltage abnormality flag bits of each alarm circuit. The fault diagnosis module is used to input the fault feature vector matrix into a pre-trained fault diagnosis model. The fault diagnosis model analyzes the fault feature vector matrix through a deep learning algorithm and outputs the fault type judgment result. The retrieval module is used to retrieve the corresponding processing solution from the pre-established fault processing database based on the fault type judgment result, thereby completing fault identification.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: The steps of the fault identification method of any one of the wired alarm controllers according to claims 1 to 5 are implemented when the processor executes the computer program.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the fault identification method of any one of the wired alarm controllers according to claims 1 to 5.

Images