A diagnostic correction method for multi-serial port data communication

Abstract

The present invention discloses a method for diagnosing and correcting multi-serial port data communications, belonging to the field of serial port communications. The method comprises: Step S1: generating several different serial port test instructions, determining whether the other end device is operating normally based on the operating status signal of the other end device, evaluating the fault type, and outputting a serial port fault data set and a serial port normal data set; constructing a random forest network model A for diagnosing multi-serial port data communication faults and a diagnostic model B for diagnosing device operating faults; during operation, the other end device inputs real-time time series status signals into the diagnostic model B, and, in combination with the random forest network model A, diagnoses whether a serial port in the communication module has a fault. The present invention is used for diagnosing and identifying multi-serial port communications faults. By combining the diagnostic model for device operating faults with the random forest network model, the method makes the fault diagnosis highly adaptable to different communication environments and fault types, effectively improving the accuracy of multi-serial port correction.

Description

Technical Field

[0001] The present invention relates to the field of serial port communications, and in particular to a diagnosis and correction method for multi-serial port data communications. Background Art

[0002] As a mature and stable communication method, serial communication plays a vital role in numerous industrial and scientific applications. Due to its widespread use in diverse environments, timely fault detection and resolution are crucial for maintaining stable system operation. Fault diagnosis is not only a necessary skill for technical support personnel in their daily work, but also an integral part of system maintenance. By diagnosing serial port faults, you can quickly locate the problem, significantly reducing system downtime and improving work efficiency.

[0003] In multi-serial data communication systems, due to the large number of serial ports and complex data transmission, data transmission errors, data loss, and communication delays often occur, seriously affecting system stability and reliability. Existing diagnostic and correction methods are often inaccurate and unable to quickly and accurately locate and resolve problems, resulting in high system maintenance costs and low efficiency. Summary of the Invention

[0004] In response to the above-mentioned deficiencies in the prior art, the present invention provides a method for diagnosing and correcting multi-serial port data communications. Based on the combination of a device fault diagnosis model and a random forest regression model, the method can diagnose and locate multi-serial port data communication faults, facilitating accurate and rapid fault correction.

[0005] In order to achieve the above-mentioned object of the invention, the technical solution adopted by the present invention is:

[0006] A method for diagnosing and correcting multi-serial port data communications is provided, comprising:

[0007] Step S1: Generate several different serial port test instructions, and send the serial port test instructions to a communication module composed of multiple serial ports. The device at one end of the communication module generates a test signal based on the serial port test instruction, and the test signal is sent to the device at the other end through the communication module.

[0008] Step S2: Collect the transmission signal received by the other end device, and determine whether the other end device is operating normally based on the status signal of the other end device, evaluate the fault type, and output the serial port fault data set and the serial port normal data set;

[0009] Step S3: Constructing a random forest network model for diagnosing multi-serial port data communication faults A and diagnostic models for equipment operation failures B ; Random Forest Network Model A and diagnostic models BThe decision function is connected in series; the random forest network model is trained using the serial port fault data set and the serial port normal data set. A Conduct training;

[0010] Step S4: During operation, the other end device inputs the real-time timing status signal into the diagnosis model B to diagnose whether the device has a fault. If the device has a fault, the transmission signal output by each serial port in the current communication module is collected and input into the random forest network model. A Check whether the serial port in the communication module is faulty; if the device is not faulty, continue to collect real-time timing status signals;

[0011] If there is a faulty serial port in the communication module, the faulty serial port number will be sent to the staff, the backup serial port in the communication module will be activated, a retransmission instruction will be generated, and the transmission signal will be sent to the other end device through the backup serial port;

[0012] If there is no faulty serial port in the communication module, a misjudgment signal is generated and sent to the staff.

[0013] Furthermore, step S2 specifically includes:

[0014] Step S21: Collect the transmission signal received by the other end device, and collect the timing state signal of the other end device based on the transmission signal to generate the timing state signal matrix , t For timing, R is the set of real numbers, T is the time step, D is the state signal dimension;

[0015] Step S22: Use multi-head self-attention to capture the long-range dependency of the temporal state signal and output the attention value;

[0016] ;

[0017] in, are the query vector, key vector, and value vector obtained by linear transformation of the time series state signal, are the weight matrices corresponding to the query vector, key vector, and value vector, respectively. is the latitude of the key vector;

[0018] Step S23: Get the state signal matrix Each timing status signal The attention value is used as the output representation of each temporal state signal , i is the number of the timing status signal;

[0019] Step S24: Injecting timing information into each timing state signal through position coding ;

[0020] ;

[0021] in, u Timing status signal The latitude index of ;

[0022] Step S25: Setting the timing information threshold and attention threshold , using timing information And the output representation , calculate the timing failure coefficient of the other end device during operation and operating state failure coefficient ;

[0023] Step S26: Setting the threshold of the timing failure coefficient and the threshold value of the operating state failure coefficient , assess whether the equipment fails during operation;

[0024] like and , it is determined that the equipment operation fails, and step S27 is executed;

[0025] Otherwise, it is determined that the equipment is operating normally;

[0026] Step S27: collecting the transmission signal output by each serial port, extracting the start bit, data bit, parity bit, and stop bit in the transmission signal, generating a standard transmission signal under normal operating conditions of the other end device, and extracting the start bit, data bit, parity bit, and stop bit of the standard transmission signal;

[0027] Step S28: Calculating a difference coefficient between the transmission signal and the standard transmission signal based on the start bit, data bit, check bit, and stop bit;

[0028] Step S29: Setting the serial port difference coefficient threshold , evaluate whether the serial port is faulty;

[0029] like , then determine the serial port u Failure occurs, otherwise, the serial port u No faults occurred;

[0030] Step S210: extracting transmission signals and difference coefficients of all serial port faults based on different serial port test instructions as a serial port fault data set;

[0031] The transmission signals and difference coefficients corresponding to all serial ports without faults are extracted as the normal serial port data set.

[0032] Furthermore, the timing failure coefficient and operating state failure coefficient The calculation method is:

[0033] ;

[0034] in, I is the number of timing status signals.

[0035] Furthermore, the coefficient of variation is calculated as;

[0036] ;

[0037] in, They are the start bit, data bit, check bit and stop bit of the transmission signal respectively. They are the start bit, data bit, check bit and stop bit of the standard transmission signal respectively. For the u The difference coefficient corresponding to the transmission signal output by the serial port, u is the serial port number, are the weight coefficients of the start bit, data bit, check bit and stop bit in the transmission signal respectively.

[0038] Furthermore, the diagnostic model B includes a diagnostic function that outputs a timing fault coefficient and an operating state fault coefficient based on the timing state signal of the other end device. ;

[0039] ;

[0040] in, represents the output representation function based on the timing state signal, is the output function of the timing information, represents the mapping between the output representation and the operating state fault coefficient, represents the mapping between timing information and timing fault coefficients, Fault tags for equipment operation, Normal label for device operation.

[0041] Furthermore, the random forest network model A The training method is:

[0042] Randomly extract with replacement from the serial port fault data set and the serial port normal data set nThe transmission signal and the corresponding difference coefficient are grouped together, and the start bit, data bit, check bit and stop bit of the transmission signal are used as input features, and the difference coefficient is used as the output feature to establish a training data set;

[0043] In the random forest network model A Build a random forest regression model in the decision tree, input the training data set into the random forest regression model, modify the parameters of the random forest regression model until the random forest regression model converges, and output the trained random forest network model A ;

[0044] The objective function for the random forest regression model to converge is;

[0045] ;

[0046] in, L is the difference in prediction accuracy during training, is the prediction accuracy threshold, i is the training data number in the training data set, is the predicted value of the output, is the true coefficient of variation.

[0047] Random Forest Network Model A The output logic of the combiner is: if M decision trees predict M There is a predicted coefficient of variation among the coefficients of variation , then it is determined that the serial port has a fault, otherwise, the serial port has no fault. m is the number of the decision tree.

[0048] The beneficial effects of the present invention are as follows: the present invention is used to diagnose and identify faults in multi-serial port communications, enabling comprehensive diagnosis of device faults and serial port faults, preventing device faults from affecting serial port fault diagnosis, and accurately locating serial port faults by referencing the fault results generated during device operation, thereby preventing the system's self-repair capabilities from affecting diagnosis. The present invention combines a diagnostic model for device operation faults with a random forest network model, which is used to diagnose serial port faults. The diagnostic model serves as a priori condition, and a decision function is constructed as the startup function of the random forest network model. This makes fault diagnosis highly adaptable to different communication environments and fault types, effectively improving the accuracy of multi-serial port correction. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 The present invention is a flowchart of a method for diagnosing and correcting multi-serial port data communications.

[0050] Figure 2 This is the schematic diagram for fault diagnosis. DETAILED DESCRIPTION

[0051] The specific embodiments of the present invention are described below to facilitate understanding of the present invention by those skilled in the art. However, it should be clear that the present invention is not limited to the scope of the specific embodiments. For those skilled in the art, as long as various changes are within the spirit and scope of the present invention as defined and determined by the appended claims, these changes are obvious, and all inventions and creations utilizing the concepts of the present invention are protected.

[0052] like Figure 1 As shown, a diagnosis and correction method for multi-serial port data communication includes:

[0053] Step S1: Generate several different serial port test instructions, and send the serial port test instructions to a communication module composed of multiple serial ports. The device at one end of the communication module generates a test signal based on the serial port test instruction, and the test signal is sent to the device at the other end through the communication module.

[0054] Step S2: Collect the transmission signal received by the other end device, and determine whether the other end device is operating normally based on the status signal of the other end device, evaluate the fault type, and output the serial port fault data set and the serial port normal data set.

[0055] Serial communication works by sending and receiving data bit by bit. Each data bit consists of a start bit, data bits, a parity bit, and a stop bit. The start and stop bits mark the beginning and end of a data frame, the data bits transmit the actual data, and the parity bit verifies the data's accuracy. Serial communication configuration parameters primarily include the baud rate, which indicates the data transmission rate.

[0056] Multi-serial port data communication is typically used to control different devices. Devices on one end (such as microcontrollers and control modules) generate different control signals. During a single control process, these signals are sent to devices on the other end (such as voice or video equipment) via a multi-serial port communication module. During multi-serial port data communication, faults may occur in the communication module or in the device on the other end. Therefore, it is necessary to diagnose the fault location and then correct it.

[0057] Step S2 specifically includes the following steps:

[0058] Step S21: Collect the transmission signal received by the other end device, and collect the timing state signal of the other end device based on the transmission signal, and generate the timing state signal matrix , t For timing, R is the set of real numbers, T is the time step, D is the state signal dimension;

[0059] Step S22: Use multi-head self-attention to capture the long-range dependency of the temporal state signal and output the attention value;

[0060] ;

[0061] in, are the query vector, key vector, and value vector obtained by linear transformation of the time series state signal, are the weight matrices corresponding to the query vector, key vector, and value vector, respectively. is the latitude of the key vector;

[0062] Step S23: Get the state signal matrix Each timing status signal The attention value is used as the output representation of each temporal state signal , i is the number of the timing status signal;

[0063] Step S24: Injecting timing information into each timing state signal through position coding ;

[0064] ;

[0065] in, u Timing status signal The latitude index of ;

[0066] Step S25: Setting the timing information threshold and attention threshold , using timing information And the output representation , calculate the timing failure coefficient of the other end device during operation and operating state failure coefficient ;

[0067] ;

[0068] in, I is the number of timing status signals;

[0069] Timing Failure Coefficient Characterizes the timing error of the status signal of the other end device during operation, such as the error in the duration of the voice device playback, the error in the time when the voice device starts playing, etc.; Operation status failure coefficient Indicates errors in the status signal of the other end device during operation, such as incorrect volume or tone of the voice device.

[0070] Step S26: Setting the threshold of the timing failure coefficient and the threshold value of the operating state failure coefficient , assess whether the equipment fails during operation;

[0071] like and , it is determined that the equipment operation fails, and step S27 is executed;

[0072] Otherwise, it is determined that the equipment is operating normally;

[0073] Threshold of timing failure coefficient and the threshold value of the operating state failure coefficient It is used as a standard to determine whether the other end device has timing faults and operating status faults during operation.

[0074] Step S27: collecting the transmission signal output by each serial port, extracting the start bit, data bit, parity bit, and stop bit in the transmission signal, generating a standard transmission signal under normal operating conditions of the other end device, and extracting the start bit, data bit, parity bit, and stop bit of the standard transmission signal;

[0075] Step S28: Calculating a difference coefficient between the transmission signal and the standard transmission signal based on the start bit, data bit, check bit, and stop bit;

[0076] ;

[0077] in, They are the start bit, data bit, check bit and stop bit of the transmission signal respectively. They are the start bit, data bit, check bit and stop bit of the standard transmission signal respectively. For the u The difference coefficient corresponding to the transmission signal output by the serial port, u is the serial port number, are the weight coefficients of the start bit, data bit, check bit and stop bit in the transmission signal respectively;

[0078] The start bit of the serial port output transmission signal indicates that before sending valid data, no configuration is required. A low-level start bit of 1 bit logic "0" will be automatically generated to indicate the start of serial port data transmission. After that, valid data will be sent. Since it does not require configuration and the data complexity is low, the probability of error is small.

[0079] The data bit determines the number of valid data bits transmitted during the communication process. The data bits are usually 5, 6, 7, or 8 bits and need to be configured accordingly.

[0080] The function of the check bit is that the serial data communication process is susceptible to external interference, which may cause data deviation. Therefore, a check bit is added after the valid data to solve this problem. The check mode needs to be configured. The check modes include odd check, even check, 0 check and 1 check.

[0081] The stop bit marks the end of a frame of data and can be 1 bit, 1.5 bits, or 2 bits of logic "1" high level. It needs to be configured according to your needs. Each device has its own clock, and there may be slight asynchrony during the transmission process. The stop bit not only indicates the end of the transmission, but also provides an opportunity to correct clock synchronization.

[0082] This embodiment determines the weight coefficient according to the data complexity of the start bit, data bit, check bit and stop bit during the serial communication process. Generally, .

[0083] During serial communication, the serial port has self-repair capabilities. A small amount of data errors can be repaired through the self-retransmission mechanism. However, for faults that cannot be self-repaired, inspection and troubleshooting are required.

[0084] Step S29: Setting the serial port difference coefficient threshold , evaluate whether the serial port is faulty;

[0085] like , then determine the serial port u Failure occurs, otherwise, the serial port u No faults occurred;

[0086] When the difference coefficient is small, the serial port can perform self-repair and no fault report is required. When the difference coefficient is large, the serial port cannot perform self-repair and a fault report is required.

[0087] Step S210: extracting transmission signals and difference coefficients of all serial port faults based on different serial port test instructions as a serial port fault data set;

[0088] The transmission signals and difference coefficients corresponding to all serial ports without faults are extracted as the normal serial port data set.

[0089] Step S3: Figure 2 As shown in the figure, a random forest network model is constructed for diagnosing multi-serial port data communication faults. A and diagnostic models for equipment operation failures B ; Random Forest Network Model A and diagnostic models B The decision function is connected in series; the random forest network model is trained using the serial port fault data set and the serial port normal data set. A Conduct training.

[0090] Diagnostic model B includes a diagnostic function that outputs a timing fault coefficient and an operating state fault coefficient based on the timing state signal of the equipment operation. ;

[0091] ;

[0092] in, Represents an output representation function based on the time series state signal. The output representation function drives the execution of steps S21-S23, and uses the multi-head self-attention mechanism to output the attention value of the time series state signal. is the output function of the timing information, driving the execution of step S24 to output the timing information. represents the mapping between the output representation and the operating state fault coefficient, represents the mapping between timing information and timing fault coefficients, Fault tags for equipment operation, For the normal label of the device operation, the formula for calculating the timing fault coefficient and the operating state fault coefficient in step S25 is constructed in the mapping;

[0093] The decision function is based on the diagnostic function The output label type determines whether to start the random forest network model A Diagnose the serial port fault of the communication module; if the diagnostic function Output fault label to start random forest network model A Diagnose the serial port fault of the communication module, collect the transmission signal of each serial port in the communication module, and input it into the trained random forest network model A In the process, the faulty serial port in the communication module is diagnosed;

[0094] Random Forest Network Model A The training method is:

[0095] Randomly extract with replacement from the serial port fault data set and the serial port normal data set n The transmission signal and the corresponding difference coefficient are grouped together, and the start bit, data bit, check bit and stop bit of the transmission signal are used as input features, and the difference coefficient is used as the output feature to establish a training data set;

[0096] In the random forest network model A Build a random forest regression model in the decision tree, input the training data set into the random forest regression model, modify the parameters of the random forest regression model until the random forest regression model converges, and output the trained random forest network model A ;

[0097] The objective function for the random forest regression model to converge is;

[0098] ;

[0099] in, L is the difference in prediction accuracy during training, is the prediction accuracy threshold, i is the training data number in the training data set, is the predicted value of the output, is the true coefficient of variation.

[0100] Random Forest Network Model A The output logic of the combiner is: if M decision trees predict M There is a predicted coefficient of variation among the coefficients of variation , then it is determined that the serial port has a fault, otherwise, the serial port has no fault. m is the number of the decision tree.

[0101] Step S4: During operation, the other end device inputs the real-time timing status signal into the diagnosis model B to diagnose whether the device has a fault. If the device has a fault, the transmission signal output by each serial port in the current communication module is collected and input into the random forest network model. A Check whether the serial port in the communication module is faulty; if the device is not faulty, continue to collect real-time timing status signals;

[0102] If there is a faulty serial port in the communication module, the faulty serial port number will be sent to the staff, the backup serial port in the communication module will be activated, a retransmission instruction will be generated, and the transmission signal will be sent to the other end device through the backup serial port;

[0103] The present invention is used to diagnose and identify faults in multi-serial port communications, enabling comprehensive diagnosis of both device and serial port faults, preventing device faults from affecting serial port fault diagnosis. By referencing the fault results generated during device operation, the present invention accurately locates serial port faults and prevents the system's self-repair capabilities from impacting diagnosis. The present invention combines a diagnostic model for device operational faults with a random forest network model. The random forest network model is used to diagnose serial port faults. The diagnostic model serves as a priori condition, and a decision function is constructed as the startup function of the random forest network model. This makes fault diagnosis highly adaptable to different communication environments and fault types, effectively improving the accuracy of multi-serial port corrections.

Claims

1. A diagnostic correction method for multi-serial port data communication, characterized in that: include: Step S1: Generate several different serial port test instructions, and send the serial port test instructions to a communication module composed of multiple serial ports. The device at one end of the communication module generates a test signal based on the serial port test instruction, and the test signal is sent to the device at the other end through the communication module. Step S2: Collect the transmission signal received by the other end device, and determine whether the other end device is operating normally based on the status signal of the other end device, evaluate the fault type, and output the serial port fault data set and the serial port normal data set; Step S3: Constructing a random forest network model for diagnosing multi-serial port data communication faults A and diagnostic models for equipment operation failures B ; Random Forest Network Model A and diagnostic models B They are connected in series through decision functions; The random forest network model is trained using the serial port fault data set and the serial port normal data set. A Conduct training; Step S4: During operation, the other end device inputs the real-time timing status signal into the diagnosis model B to diagnose whether the device has a fault. If the device has a fault, the transmission signal output by each serial port in the current communication module is collected and input into the random forest network model. A Check whether the serial port in the communication module is faulty; if the device is not faulty, continue to collect real-time timing status signals; If there is a faulty serial port in the communication module, the faulty serial port number will be sent to the staff, the backup serial port in the communication module will be activated, a retransmission instruction will be generated, and the transmission signal will be sent to the other end device through the backup serial port; If there is no faulty serial port in the communication module, a false positive signal is generated and sent to the staff; The step S2 specifically includes: Step S21: Collect the transmission signal received by the other end device, and collect the timing state signal of the other end device based on the transmission signal to generate the timing state signal matrix , t For timing, R is the set of real numbers, T is the time step, D is the state signal dimension; Step S22: Use multi-head self-attention to capture the long-range dependency of the temporal state signal and output the attention value; ； in, are the query vector, key vector, and value vector obtained by linear transformation of the time series state signal, are the weight matrices corresponding to the query vector, key vector, and value vector, respectively. is the latitude of the key vector; Step S23: Get the state signal matrix Each timing status signal The attention value is used as the output representation of each temporal state signal , i is the number of the timing status signal; Step S24: Injecting timing information into each timing state signal through position coding ; ； in, u Timing status signal The latitude index of ; Step S25: Setting the timing information threshold and attention threshold , using timing information And the output representation , calculate the timing failure coefficient of the other end device during operation and operating state failure coefficient ; Step S26: Setting the threshold of the timing failure coefficient and the threshold value of the operating state failure coefficient , assess whether the equipment fails during operation; like and , it is determined that the equipment operation fails, and step S27 is executed; Otherwise, it is determined that the equipment is operating normally; Step S27: collecting the transmission signal output by each serial port, extracting the start bit, data bit, parity bit, and stop bit in the transmission signal, generating a standard transmission signal under normal operating conditions of the other end device, and extracting the start bit, data bit, parity bit, and stop bit of the standard transmission signal; Step S28: Calculating a difference coefficient between the transmission signal and the standard transmission signal based on the start bit, data bit, check bit, and stop bit; Step S29: Setting the serial port difference coefficient threshold , evaluate whether the serial port is faulty; like , then determine the serial port u Failure occurs, otherwise, the serial port u No faults occurred; Step S210: extracting transmission signals and difference coefficients of all serial port faults based on different serial port test instructions as a serial port fault data set; The transmission signals and difference coefficients corresponding to all serial ports without faults are extracted as the normal serial port data set.

2. The multi-serial port data communication diagnosis and correction method according to claim 1, characterized in that: The timing failure coefficient and operating state failure coefficient The calculation method is: ； in, I is the number of timing status signals.

3. The multi-serial port data communication diagnosis and correction method according to claim 2, characterized in that: The calculation method of the coefficient of difference is: ； in, They are the start bit, data bit, check bit and stop bit of the transmission signal respectively. They are the start bit, data bit, check bit and stop bit of the standard transmission signal respectively. For the u The difference coefficient corresponding to the transmission signal output by the serial port, u is the serial port number, are the weight coefficients of the start bit, data bit, check bit and stop bit in the transmission signal respectively.

4. The multi-serial port data communication diagnosis and correction method according to claim 3, characterized in that: The diagnostic model B includes a diagnostic function based on the timing state signal output of the other end device operating timing fault coefficient and the operating state fault coefficient ; ； in, represents the output representation function based on the timing state signal, is the output function of the timing information, represents the mapping between the output representation and the operating state fault coefficient, represents the mapping between timing information and timing fault coefficients, Fault tags for equipment operation, Normal label for device operation.

5. The multi-serial port data communication diagnosis and correction method according to claim 4, characterized in that: The random forest network model A The training method is: Randomly extract with replacement from the serial port fault data set and the serial port normal data set n The transmission signal and the corresponding difference coefficient are grouped together, and the start bit, data bit, check bit and stop bit of the transmission signal are used as input features, and the difference coefficient is used as the output feature to establish a training data set; In the random forest network model A Build a random forest regression model in the decision tree, input the training data set into the random forest regression model, modify the parameters of the random forest regression model until the random forest regression model converges, and output the trained random forest network model A ; The objective function for the random forest regression model to converge is; ； in, L is the difference in prediction accuracy during training, is the prediction accuracy threshold, i is the training data number in the training data set, is the predicted value of the output, is the true coefficient of variation; Random Forest Network Model A The output logic of the combiner is: if M decision trees predict M There is a predicted coefficient of variation among the coefficients of variation , then it is determined that the serial port has a fault, otherwise, the serial port has no fault. m is the number of the decision tree.

Images