Constant-temperature and constant-humidity air conditioner abnormity monitoring method and system fused with state covariant analysis

Abstract

The constant-temperature and constant-humidity air conditioner abnormity monitoring method comprises the steps that different working conditions are divided, and a reference covariant matrix and a vibration quantity coefficient threshold value of the different working conditions are established; calculating a correlation coefficient of the current and the vibration quantity in real time, and establishing a dynamic covariant matrix; according to the dynamic covariant matrix and a reference covariant matrix corresponding to the current working condition, calculating a covariant deviation degree, and if the covariant deviation degree is greater than a set covariant deviation degree threshold value or a correlation coefficient of the current and the vibration quantity is greater than a corresponding vibration quantity coefficient threshold value, judging that the equipment state is abnormal; judging whether an environment state is abnormal or not according to the workshop environment temperature and humidity at the current moment; inputting the workshop environment temperature and humidity in a set period into an isolated forest algorithm to output an unsupervised abnormal score; and air conditioner abnormity grading judgment is conducted. According to the invention, the probability of missing report and false report is greatly reduced, and more comprehensive and reliable abnormity diagnosis is realized.

Description

Technical Field

[0001] This invention belongs to the field of air conditioning monitoring technology, specifically relating to a method and system for monitoring anomalies in constant temperature and humidity air conditioning systems based on fusion state covariance analysis. Background Technology

[0002] In modern industrial production and equipment management, the stability of constant temperature and humidity environments plays a crucial role in the quality of key equipment and products, especially in scenarios such as power industry workshops, precision manufacturing workshops, data centers, and laboratories. Taking power industry workshops as an example, high-voltage substations, relay protection devices, and electronic control equipment are extremely sensitive to changes in environmental temperature and humidity. Excessively high or low temperatures may lead to abnormal heat dissipation and electrical insulation failure; excessive humidity can easily cause corrosion of components, short circuits, or safety hazards; and excessively low humidity increases the risk of electrostatic discharge.

[0003] Traditional constant temperature and humidity air conditioning monitoring systems are mostly based on single-variable threshold alarms (such as temperature and humidity thresholds, excessive current, and temperature exceeding limits) and simple static models. They cannot capture the normal fluctuations of state variables of the equipment under different operating conditions. Single-variable analysis ignores dynamic coupling relationships. Many early faults are manifested as abnormal changes in these correlations rather than the excess of the absolute value of a certain parameter, thus resulting in missed alarms. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides a method and system for monitoring anomalies in constant temperature and humidity air conditioning systems based on fusion state covariance analysis.

[0005] The present invention adopts the following technical solution.

[0006] The first aspect of this invention proposes a method for monitoring anomalies in constant temperature and humidity air conditioning systems based on fusion state covariance analysis, comprising: The operation of the air conditioning equipment is divided into different working conditions based on the current of the air conditioning equipment. A reference covariance matrix and vibration coefficient threshold are established for each working condition based on the historical state variables of the air conditioning equipment under each working condition. The state variables include the current, voltage and vibration of the air conditioning equipment. Real-time acquisition of workshop ambient temperature and humidity, as well as various state variables of air conditioning equipment; calculation of the correlation coefficient between current and vibration; and fitting of all state variables through a GARCH model to obtain the conditional standard deviation and original residual of each state variable. A dynamic covariance matrix is ​​then established based on the conditional standard deviation and original residual. The current operating condition is determined by the current in the state variables. The covariance deviation is calculated based on the dynamic covariance matrix and the reference covariance matrix corresponding to the current operating condition. If the covariance deviation is greater than the set covariance deviation threshold or the correlation coefficient between the current and the vibration is greater than the corresponding vibration coefficient threshold, the equipment is judged to be in an abnormal state. Determine whether an environmental anomaly has occurred based on the current workshop ambient temperature and humidity. The workshop ambient temperature and humidity within a set period are input into the isolated forest algorithm to output unsupervised anomaly scores; based on whether equipment status anomalies occur, whether environmental status anomalies occur, and the unsupervised anomaly scores, air conditioning anomaly classification is determined.

[0007] Preferably, the step of dividing the operation of the air conditioning equipment into different operating conditions based on the current of the air conditioning equipment, and establishing a reference covariance matrix and vibration coefficient threshold for each operating condition based on the historical state variables of the air conditioner's normal operation under each operating condition, specifically involves: The ratio of current to rated current is used as the load factor. Different ranges are set for each operating condition. The operating condition corresponding to the range of load factor is the current operating condition. The average value of each historical state variable under each operating condition is obtained. The conditional standard deviation and original residual of each average value are obtained by fitting the GARCH model. The historical dynamic covariance matrix is ​​established based on the conditional standard deviation and original residual as the benchmark covariance matrix for the corresponding operating condition. Based on the historical state variables of the air conditioner's normal operation under each operating condition, establish the benchmark covariance matrix and vibration coefficient threshold for different operating conditions; Obtain all historical currents and historical vibrations for each operating condition of the air conditioner, solve for the correlation coefficients between all corresponding historical currents and vibrations, and calculate the average and standard deviation of all historical current and vibration correlation coefficients. The average minus three times the standard deviation is the vibration coefficient threshold.

[0008] Preferably, the step of establishing a dynamic covariant matrix based on the conditional standard deviation and the original residuals specifically involves:

[0009] in, It is a dynamic covariant matrix; for t A diagonal matrix composed of the conditional standard deviations of the state variables at each time step; for t The time-varying correlation matrix at time t is a matrix composed of the correlation coefficients of every two state variables; , All are attenuation coefficients. To extract the diagonal elements of the matrix, create a diagonal matrix. for t The unstandardized dynamic covariance matrix at time t. for t The unstandardized dynamic covariance matrix at time -1; is the unconditional covariance matrix, and is the covariance matrix of the standardized residual vectors of the historical data; fort The standardized residual vector at time -1 is a vector composed of the standardized residuals of each state variable. The standardized residual is the original residual divided by the corresponding conditional standard deviation. This is the transpose symbol.

[0010] Preferably, recursive calculation t The unstandardized dynamic covariance matrix at time t and t The unstandardized dynamic covariance matrix at time -1 ,when t When it is 0 ， Set as unconditional covariance matrix .

[0011] Preferably, the step of calculating the covariance deviation based on the dynamic covariance matrix and the reference covariance matrix corresponding to the current operating condition specifically involves:

[0012] in, This is a vectorized operation, meaning that the columns of the matrix are stacked together to form a column vector; It is a dynamic covariant matrix; This is the baseline covariance matrix corresponding to the current operating condition; for The inverse covariance matrix.

[0013] Preferably, the step of determining whether an environmental anomaly has occurred based on the current workshop ambient temperature and humidity specifically involves: Calculate the rate of change of the current workshop ambient temperature and humidity, as well as the difference between the temperature and humidity and the historical temperature and humidity corresponding to the normal operation of the air conditioner under the current working conditions. If the current workshop ambient temperature is greater than the set temperature threshold, or the rate of temperature change is greater than the set temperature change rate threshold, or the difference between the current temperature and the historical temperature is greater than the set temperature difference threshold; or the humidity is greater than the set humidity threshold, or the rate of humidity change is greater than the set humidity change rate threshold, or the difference between the current humidity and the historical humidity is greater than the set humidity difference threshold; then an abnormal environmental condition is determined to have occurred.

[0014] Preferably, the step of inputting the workshop ambient temperature and humidity within a set period into the Isolation Forest algorithm to output unsupervised anomaly scores specifically involves: Based on the workshop ambient temperature and humidity within a set period, calculate multiple feature data. Traverse each isolated tree in the isolated forest algorithm, calculate the path length of all feature data from the root node to the leaf node for each isolated tree, calculate the average path length of all isolated trees, and substitute it into the following formula to obtain the anomaly score. :

[0015] in, For the first period within the set period i A collection of feature data for a sliding window; Indicates the path length of an isolated tree; This is the average path length of all isolated trees; For normalization term, This represents the number of training samples used when training the Isolation Forest algorithm. For the first The harmonic number of the term.

[0016] Preferably, the feature data specifically includes: The set period is divided into multiple sliding windows according to the set sliding window length and step size. For each sliding window, the feature data includes the average temperature, average humidity, temperature variance, humidity variance, range, and correlation coefficient between temperature and humidity.

[0017] Preferably, the step of classifying air conditioning anomalies based on whether equipment status anomalies have occurred, whether environmental status anomalies have occurred, and unsupervised anomaly scores is as follows: If the unsupervised anomaly score is greater than the set first anomaly threshold and less than or equal to the set second threshold, then a minor unsupervised anomaly is determined to have occurred; if the unsupervised anomaly score is greater than the set second anomaly threshold, then a severe unsupervised anomaly is determined to have occurred. If the abnormal equipment status, abnormal environmental status, minor unsupervised abnormality, and severe unsupervised abnormality are identified, and none of the above abnormalities occur, then the air conditioner is considered to be normal. If it is determined that only one of the following abnormalities has occurred: abnormal equipment status, abnormal environmental status, or minor unsupervised abnormality, then it is considered a minor air conditioning abnormality. If it is determined that both equipment status abnormality and unsupervised minor abnormality occur simultaneously, or both environmental status abnormality and unsupervised minor abnormality occur simultaneously, then it is considered a common air conditioning abnormality. If it is determined that both equipment status abnormality and environmental status abnormality occur simultaneously, or if an unsupervised serious abnormality occurs, then it is considered a serious air conditioning abnormality.

[0018] The second aspect of this invention proposes a constant temperature and humidity air conditioning anomaly monitoring system based on the fusion state covariance analysis method described in the first aspect of this invention, comprising a benchmark setting module, a dynamic covariance matrix calculation module, an equipment state anomaly judgment module, an environmental state anomaly judgment module, an unsupervised anomaly score judgment module, and an anomaly classification module, specifically: Reference setting module: used to divide the operation of air conditioning equipment into different working conditions according to the current of the air conditioning equipment, and to establish a reference covariance matrix and vibration coefficient threshold for different working conditions based on the historical state variables of the air conditioner's normal operation under each working condition; the state variables include current, voltage and vibration. Data acquisition module: used to collect real-time ambient temperature and humidity in the workshop, as well as various status variables of the air conditioning equipment; Dynamic covariance matrix calculation module: used to calculate the correlation coefficient between current and vibration, and to fit all state variables through the GARCH model to obtain the conditional standard deviation and original residual of each state variable, and to establish the dynamic covariance matrix based on the conditional standard deviation and original residual; Equipment status anomaly judgment module: It is used to judge the current working condition by the current in the status variable, calculate the covariance deviation degree according to the dynamic covariance matrix and the reference covariance matrix corresponding to the current working condition, and judge the equipment status anomaly if the covariance deviation degree is greater than the set covariance deviation degree threshold or the correlation coefficient between the current and the vibration is greater than the corresponding vibration coefficient threshold. Environmental anomaly judgment module: used to determine whether an environmental anomaly has occurred based on the current workshop ambient temperature and humidity; Unsupervised anomaly score judgment module: Used to input the workshop ambient temperature and humidity within a set period into the isolated forest algorithm to output unsupervised anomaly scores; Anomaly Classification Module: Based on whether equipment status anomalies have occurred, whether environmental status anomalies have occurred, and unsupervised anomaly scores, the module classifies air conditioning anomalies.

[0019] The beneficial effects of this invention are that, compared with the prior art, Based on current-based division of operating conditions and the establishment of a benchmark covariance matrix and vibration coefficient threshold for each operating condition, this method effectively avoids the limitations of the traditional single threshold method and adapts to various operating conditions. It utilizes a GARCH model to construct a dynamic covariance matrix to calculate the covariance deviation, capturing the time-varying correlation characteristics between state variables. The correlation coefficient between current and vibration can be used to determine whether mechanical transmission failure (such as bearing wear) has occurred, enhancing the sensitivity for identifying latent anomalies. By combining anomaly detection based on the correlation coefficient between covariance deviation current and vibration, anomaly detection based on temperature and humidity, and unsupervised detection using isolated forests, it achieves collaborative diagnosis of equipment state anomalies, environmental anomalies, and irregular anomalies, significantly reducing the probability of missed and false alarms and realizing a more comprehensive and reliable anomaly diagnosis. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating the abnormal monitoring method for constant temperature and humidity air conditioning based on fusion state covariance analysis of the present invention. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of this invention. The embodiments described in this application are merely some embodiments of this invention, and not all embodiments. Based on the spirit of this invention, all other embodiments obtained by those skilled in the art without creative effort are within the protection scope of this invention.

[0022] like Figure 1 As shown, Embodiment 1 of the present invention proposes a method for monitoring anomalies in constant temperature and humidity air conditioning systems based on fusion state covariance analysis, including: The operation of the air conditioning equipment is divided into different working conditions based on the current of the air conditioning equipment. A reference covariance matrix and vibration coefficient threshold are established for each working condition based on the historical state variables of the air conditioner's normal operation. The state variables include current, voltage and vibration. It should be noted that the air conditioning equipment includes an air conditioning compressor, a frequency converter, and a fan. The state variables may also include other relevant parameters such as the equipment's heat output and airflow organization parameters. Real-time acquisition of workshop ambient temperature and humidity, as well as various state variables of air conditioning equipment; calculation of the correlation coefficient between current and vibration; and fitting of all state variables through a GARCH model to obtain the conditional standard deviation and original residual of each state variable. A dynamic covariance matrix is ​​then established based on the conditional standard deviation and original residual. The current operating condition is determined by the current in the state variables. The covariance deviation is calculated based on the dynamic covariance matrix and the reference covariance matrix corresponding to the current operating condition. If the covariance deviation is greater than the set covariance deviation threshold or the correlation coefficient between the current and the vibration is greater than the corresponding vibration coefficient threshold, the equipment is judged to be in an abnormal state. Determine whether an environmental anomaly has occurred based on the current workshop ambient temperature and humidity. The workshop ambient temperature and humidity within a set period are input into the isolated forest algorithm to output unsupervised anomaly scores; based on whether equipment status anomalies occur, whether environmental status anomalies occur, and the unsupervised anomaly scores, air conditioning anomaly classification is determined.

[0023] In this preferred embodiment, the step of dividing the operation of the air conditioning equipment into different operating conditions based on the current of the air conditioning equipment, and establishing a reference covariance matrix and vibration coefficient threshold for each operating condition based on the historical state variables of the air conditioning equipment's normal operation under each operating condition, specifically involves: The ratio of current to rated current is used as the load factor. Different ranges are set for each operating condition. The operating condition corresponding to the range of load factor is the current operating condition. The average value of each historical state variable under each operating condition is obtained. The conditional standard deviation and original residual of each average value are obtained by fitting the GARCH model. The historical dynamic covariance matrix is ​​established based on the conditional standard deviation and original residual as the benchmark covariance matrix for the corresponding operating condition. Based on the historical state variables of the air conditioner's normal operation under each operating condition, establish the benchmark covariance matrix and vibration coefficient threshold for different operating conditions; Obtain all historical currents and historical vibrations for each operating condition of the air conditioner, solve for the correlation coefficients between all corresponding historical currents and vibrations, and calculate the average and standard deviation of all historical current and vibration correlation coefficients. The average minus three times the standard deviation is the vibration coefficient threshold.

[0024] If the load rate is greater than or equal to 0 and less than 5%, the system is in a shutdown / standby state; if the load rate is greater than or equal to 5% and less than 30%, the system is in a low load state; if the load rate is greater than or equal to 30% and less than 70%, the system is in a normal load state; if the load rate is greater than or equal to 70%, the system is in a high load state. In this preferred embodiment, the step of establishing the dynamic covariance matrix based on the conditional standard deviation and the original residuals specifically involves:

[0025] in, It is a dynamic covariant matrix; for t A diagonal matrix composed of the conditional standard deviations of the state variables at each time step; for t The time-varying correlation matrix at time t is a matrix composed of the correlation coefficients of every two state variables; , All are attenuation coefficients. To extract the diagonal elements of the matrix, create a diagonal matrix. for t The unstandardized dynamic covariance matrix at time t. for t The unstandardized dynamic covariance matrix at time -1; is the unconditional covariance matrix, and is the covariance matrix of the standardized residual vectors of the historical data; for t The standardized residual vector at time -1 is a vector composed of the standardized residuals of each state variable. The standardized residual is the original residual divided by the corresponding conditional standard deviation. This is the transpose symbol.

[0026] In this preferred embodiment, recursive calculation is used. t The unstandardized dynamic covariance matrix at time t and tThe unstandardized dynamic covariance matrix at time -1 ,when t When it is 0 ， Set as unconditional covariance matrix .

[0027] In this preferred embodiment, the calculation of the covariance deviation based on the dynamic covariance matrix and the reference covariance matrix corresponding to the current operating condition specifically involves:

[0028] in, This is a vectorized operation, meaning that the columns of the matrix are stacked together to form a column vector; It is a dynamic covariant matrix; This is the baseline covariance matrix corresponding to the current operating condition; for The inverse covariance matrix.

[0029] In this preferred embodiment, the step of determining whether an environmental anomaly has occurred based on the current workshop ambient temperature and humidity specifically involves: Calculate the rate of change of the current workshop ambient temperature and humidity, as well as the difference between the temperature and humidity and the historical temperature and humidity corresponding to the normal operation of the air conditioner under the current working conditions. If the current workshop ambient temperature is greater than the set temperature threshold, or the rate of temperature change is greater than the set temperature change rate threshold, or the difference between the current temperature and the historical temperature is greater than the set temperature difference threshold; or the humidity is greater than the set humidity threshold, or the rate of humidity change is greater than the set humidity change rate threshold, or the difference between the current humidity and the historical humidity is greater than the set humidity difference threshold; then an abnormal environmental condition is determined to have occurred.

[0030] In this preferred embodiment, the step of inputting the workshop ambient temperature and humidity within a set period into the Isolation Forest algorithm to output unsupervised anomaly scores specifically involves: When training the Isolation Forest algorithm, isolated trees are generated through recursive random partitioning. The recursive random partitioning process is as follows: Multiple feature data points are calculated based on the workshop ambient temperature and humidity from historical data to form the training set; one feature data point is randomly selected, and a split point is randomly generated. The range of the split point is between the minimum and maximum values ​​of the selected feature data in the historical data. Based on the comparison with the split point, all historical data are divided into two subsets: the selected feature data in the left subset is less than the split point, and the selected feature data in the right subset is greater than or equal to the split point. For each subset, the above steps are repeated starting from randomly selecting one feature data point until the newly partitioned subset contains only one historical data point, or the number of partitions reaches a set depth threshold. This recursive random partitioning process is repeated to generate multiple isolated trees. Based on the workshop ambient temperature and humidity within a set period, calculate multiple feature data. Traverse each isolated tree in the isolated forest algorithm, calculate the path length of all feature data from the root node to the leaf node for each isolated tree, calculate the average path length of all isolated trees, and substitute it into the following formula to obtain the anomaly score. :

[0031] in, For the first period within the set period i A collection of feature data for a sliding window; Indicates the path length of an isolated tree; This is the average path length of all isolated trees; For normalization term, This represents the number of training samples used when training the Isolation Forest algorithm. For the first The harmonic number of the term.

[0032] In this preferred embodiment, the feature data specifically includes: The set period is divided into multiple sliding windows according to the set sliding window length and step size. For each sliding window, the feature data includes the average temperature, average humidity, temperature variance, humidity variance, range, and correlation coefficient between temperature and humidity.

[0033] In this preferred embodiment, the step of classifying air conditioning anomalies based on whether equipment status anomalies have occurred, whether environmental status anomalies have occurred, and unsupervised anomaly scores is specifically as follows: If the unsupervised anomaly score is greater than the set first anomaly threshold and less than or equal to the set second threshold, then a minor unsupervised anomaly is determined to have occurred; if the unsupervised anomaly score is greater than the set second anomaly threshold, then a severe unsupervised anomaly is determined to have occurred. If the abnormal equipment status, abnormal environmental status, minor unsupervised abnormality, and severe unsupervised abnormality are identified, and none of the above abnormalities occur, then the air conditioner is considered to be normal. If it is determined that only one of the following abnormalities has occurred: abnormal equipment status, abnormal environmental status, or minor unsupervised abnormality, then it is considered a minor air conditioning abnormality. If it is determined that both equipment status abnormality and unsupervised minor abnormality occur simultaneously, or both environmental status abnormality and unsupervised minor abnormality occur simultaneously, then it is considered a common air conditioning abnormality. If it is determined that both equipment status abnormality and environmental status abnormality occur simultaneously, or if an unsupervised serious abnormality occurs, then it is considered a serious air conditioning abnormality.

[0034] Embodiment 2 of the present invention proposes a constant temperature and humidity air conditioning anomaly monitoring system based on the fusion state covariance analysis method described in Embodiment 1 of the present invention, including a benchmark setting module, a dynamic covariance matrix calculation module, an equipment state anomaly judgment module, an environmental state anomaly judgment module, an unsupervised anomaly score judgment module, and an anomaly classification module, characterized in that: Reference setting module: used to divide the operation of air conditioning equipment into different working conditions according to the current of the air conditioning equipment, and to establish a reference covariance matrix and vibration coefficient threshold for different working conditions based on the historical state variables of the air conditioner's normal operation under each working condition; the state variables include current, voltage and vibration. Data acquisition module: used to collect real-time ambient temperature and humidity in the workshop, as well as various status variables of the air conditioning equipment; Dynamic covariance matrix calculation module: used to calculate the correlation coefficient between current and vibration, and to fit all state variables through the GARCH model to obtain the conditional standard deviation and original residual of each state variable, and to establish the dynamic covariance matrix based on the conditional standard deviation and original residual; Equipment status anomaly judgment module: It is used to judge the current working condition by the current in the status variable, calculate the covariance deviation degree according to the dynamic covariance matrix and the reference covariance matrix corresponding to the current working condition, and judge the equipment status anomaly if the covariance deviation degree is greater than the set covariance deviation degree threshold or the correlation coefficient between the current and the vibration is greater than the corresponding vibration coefficient threshold. Environmental anomaly judgment module: used to determine whether an environmental anomaly has occurred based on the current workshop ambient temperature and humidity; Unsupervised anomaly score judgment module: Used to input the workshop ambient temperature and humidity within a set period into the isolated forest algorithm to output unsupervised anomaly scores; Anomaly Classification Module: Based on whether equipment status anomalies have occurred, whether environmental status anomalies have occurred, and unsupervised anomaly scores, the module classifies air conditioning anomalies.

[0035] It should be noted that, preferably in this embodiment, the calculation and judgment of the environmental state anomaly judgment module are relatively simple, and are directly calculated at the edge. After the data acquisition module collects data, it directly judges whether an environmental state anomaly has occurred through the environmental state anomaly judgment module. Then, the environmental state anomaly judgment module sends the status of environmental state anomaly to the cloud or remote monitoring center. The data acquisition module also collects the workshop ambient temperature and humidity and the various state variables of the air conditioning equipment and sends them to the cloud or remote monitoring center. The dynamic covariance matrix calculation module, the equipment state anomaly judgment module, the unsupervised anomaly score judgment module, and the anomaly classification module are deployed in the cloud or remote monitoring center for subsequent calculation and judgment. Dual protocol redundancy transmission is used for data transmission. The data acquisition module transmits the workshop ambient temperature and humidity and the various state variables of the air conditioning equipment through Modbus RTU polling. The environmental state anomaly judgment module transmits whether an environmental state anomaly has occurred through the MQTT protocol. When a transmission fails, the constant temperature and humidity air conditioning anomaly monitoring system does not completely stop working.

[0036] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A method for monitoring anomalies in constant temperature and humidity air conditioning systems based on fusion state covariance analysis, characterized in that, include: Based on the current of the air conditioning equipment, the operation of the air conditioning equipment is divided into different working conditions. Based on the historical state variables of the normal operation of the air conditioning equipment under each working condition, a benchmark covariance matrix and vibration coefficient threshold are established for different working conditions. State variables include the current, voltage, and vibration of the air conditioning equipment; Real-time acquisition of workshop ambient temperature and humidity, as well as various state variables of air conditioning equipment; calculation of the correlation coefficient between current and vibration; and fitting of all state variables through a GARCH model to obtain the conditional standard deviation and original residual of each state variable. A dynamic covariance matrix is ​​then established based on the conditional standard deviation and original residual. The current operating condition is determined by the current in the state variables. The covariance deviation is calculated based on the dynamic covariance matrix and the reference covariance matrix corresponding to the current operating condition. If the covariance deviation is greater than the set covariance deviation threshold or the correlation coefficient between the current and the vibration is greater than the corresponding vibration coefficient threshold, the equipment is judged to be in an abnormal state. Determine whether an environmental anomaly has occurred based on the current workshop ambient temperature and humidity. The workshop ambient temperature and humidity within a set period are input into the isolated forest algorithm to output unsupervised anomaly scores. The air conditioning anomaly classification is determined based on whether equipment status abnormalities have occurred, whether environmental status abnormalities have occurred, and unsupervised anomaly scores.

2. The method for monitoring anomalies in a constant temperature and humidity air conditioner based on fusion state covariance analysis according to claim 1, characterized in that: The air conditioning equipment is divided into different operating conditions based on its current. A benchmark covariance matrix and vibration coefficient threshold are established for each operating condition based on the historical state variables of the air conditioner's normal operation. Specifically: The ratio of current to rated current is used as the load factor. Different ranges are set for each operating condition. The operating condition corresponding to the range of load factor is the current operating condition. The average value of each historical state variable under each operating condition is obtained. The conditional standard deviation and original residual of each average value are obtained by fitting the GARCH model. The historical dynamic covariance matrix is ​​established based on the conditional standard deviation and original residual as the benchmark covariance matrix for the corresponding operating condition. Based on the historical state variables of the air conditioner's normal operation under each operating condition, establish the benchmark covariance matrix and vibration coefficient threshold for different operating conditions; Obtain all historical currents and historical vibrations for each operating condition of the air conditioner, solve for the correlation coefficients between all corresponding historical currents and vibrations, and calculate the average and standard deviation of all historical current and vibration correlation coefficients. The average minus three times the standard deviation is the vibration coefficient threshold.

3. The method for monitoring anomalies in a constant temperature and humidity air conditioner based on fusion state covariance analysis according to claim 2, characterized in that: The establishment of the dynamic covariant matrix based on the conditional standard deviation and the original residuals is specifically as follows: in, It is a dynamic covariant matrix; for t A diagonal matrix composed of the conditional standard deviations of the state variables at each time step; for t The time-varying correlation matrix at time t is a matrix composed of the correlation coefficients of every two state variables; , All are attenuation coefficients. To extract the diagonal elements of the matrix, create a diagonal matrix. for t The unstandardized dynamic covariance matrix at time t. for t The unstandardized dynamic covariance matrix at time -1; is the unconditional covariance matrix, and is the covariance matrix of the standardized residual vectors of the historical data; for t The standardized residual vector at time -1 is a vector composed of the standardized residuals of each state variable. The standardized residual is the original residual divided by the corresponding conditional standard deviation. This is the transpose symbol.

4. The method for monitoring anomalies in a constant temperature and humidity air conditioner based on fusion state covariance analysis according to claim 3, characterized in that: Through recursive calculation t The unstandardized dynamic covariance matrix at time t and t The unstandardized dynamic covariance matrix at time -1 ,when t When it is 0 ， Set as unconditional covariance matrix .

5. The method for monitoring anomalies in a constant temperature and humidity air conditioner based on fusion state covariance analysis according to claim 4, characterized in that: The calculation of the covariance deviation based on the dynamic covariance matrix and the reference covariance matrix corresponding to the current operating condition is specifically as follows: in, This is a vectorized operation, meaning that the columns of the matrix are stacked together to form a column vector; It is a dynamic covariant matrix; This is the baseline covariance matrix corresponding to the current operating condition; for The inverse covariance matrix.

6. The method for monitoring anomalies in a constant temperature and humidity air conditioner based on fusion state covariance analysis according to claim 2, characterized in that: The determination of whether an environmental anomaly has occurred based on the current workshop ambient temperature and humidity is as follows: Calculate the rate of change of the current workshop ambient temperature and humidity, as well as the difference between the temperature and humidity and the historical temperature and humidity corresponding to the normal operation of the air conditioner under the current working conditions. If the current workshop ambient temperature is greater than the set temperature threshold, or the rate of temperature change is greater than the set temperature change rate threshold, or the difference between the current temperature and the historical temperature is greater than the set temperature difference threshold; or the humidity is greater than the set humidity threshold, or the rate of humidity change is greater than the set humidity change rate threshold, or the difference between the current humidity and the historical humidity is greater than the set humidity difference threshold; then an abnormal environmental condition is determined to have occurred.

7. The method for monitoring anomalies in a constant temperature and humidity air conditioner based on fusion state covariance analysis according to claim 1, characterized in that: The process of inputting the workshop ambient temperature and humidity within a set period into the isolated forest algorithm to output unsupervised anomaly scores is as follows: Based on the workshop ambient temperature and humidity within a set period, calculate multiple feature data. Traverse each isolated tree in the isolated forest algorithm, calculate the path length of all feature data from the root node to the leaf node for each isolated tree, calculate the average path length of all isolated trees, and substitute it into the following formula to obtain the anomaly score. : in, For the first period within the set period i A collection of feature data for a sliding window; Indicates the path length of an isolated tree; This is the average path length of all isolated trees; For normalization term, This represents the number of training samples used when training the Isolation Forest algorithm. For the first The harmonic number of the term.

8. The method for monitoring anomalies in a constant temperature and humidity air conditioner based on fusion state covariance analysis according to claim 7, characterized in that: The feature data specifically includes: The set period is divided into multiple sliding windows according to the set sliding window length and step size. For each sliding window, the feature data includes the average temperature, average humidity, temperature variance, humidity variance, range, and correlation coefficient between temperature and humidity.

9. The method for monitoring anomalies in a constant temperature and humidity air conditioner based on fusion state covariance analysis according to claim 1, characterized in that: The air conditioning anomaly classification is determined based on whether equipment status anomalies, environmental status anomalies, and unsupervised anomaly scores. Specifically: If the unsupervised anomaly score is greater than the set first anomaly threshold and less than or equal to the set second threshold, then a minor unsupervised anomaly is determined to have occurred; if the unsupervised anomaly score is greater than the set second anomaly threshold, then a severe unsupervised anomaly is determined to have occurred. If the abnormal equipment status, abnormal environmental status, minor unsupervised abnormality, and severe unsupervised abnormality are identified, and none of the above abnormalities occur, then the air conditioner is considered to be normal. If it is determined that only one of the following abnormalities has occurred: abnormal equipment status, abnormal environmental status, or minor unsupervised abnormality, then it is considered a minor air conditioning abnormality. If it is determined that both equipment status abnormality and unsupervised minor abnormality occur simultaneously, or both environmental status abnormality and unsupervised minor abnormality occur simultaneously, then it is considered a common air conditioning abnormality. If it is determined that both equipment status abnormality and environmental status abnormality occur simultaneously, or if an unsupervised serious abnormality occurs, then it is considered a serious air conditioning abnormality.

10. A constant temperature and humidity air conditioning anomaly monitoring system based on the fusion state covariance analysis of the method described in any one of claims 1-9, comprising a benchmark setting module, a dynamic covariance matrix calculation module, an equipment state anomaly judgment module, an environmental state anomaly judgment module, an unsupervised anomaly score judgment module, and an anomaly classification module, characterized in that: The benchmark setting module is used to divide the operation of the air conditioning equipment into different working conditions based on the current of the air conditioning equipment, and to establish the benchmark covariance matrix and vibration coefficient threshold for each working condition based on the historical state variables of the normal operation of the air conditioning under each working condition. State variables include current, voltage, and vibration. Data acquisition module: used to collect real-time ambient temperature and humidity in the workshop, as well as various status variables of the air conditioning equipment; Dynamic covariance matrix calculation module: used to calculate the correlation coefficient between current and vibration, and to fit all state variables through the GARCH model to obtain the conditional standard deviation and original residual of each state variable, and to establish the dynamic covariance matrix based on the conditional standard deviation and original residual; Equipment status anomaly judgment module: It is used to judge the current working condition by the current in the status variable, calculate the covariance deviation degree according to the dynamic covariance matrix and the reference covariance matrix corresponding to the current working condition, and judge the equipment status anomaly if the covariance deviation degree is greater than the set covariance deviation degree threshold or the correlation coefficient between the current and the vibration is greater than the corresponding vibration coefficient threshold. Environmental anomaly judgment module: used to determine whether an environmental anomaly has occurred based on the current workshop ambient temperature and humidity; Unsupervised anomaly score judgment module: Used to input the workshop ambient temperature and humidity within a set period into the isolated forest algorithm to output unsupervised anomaly scores; Anomaly Classification Module: Based on whether equipment status anomalies have occurred, whether environmental status anomalies have occurred, and unsupervised anomaly scores, the module classifies air conditioning anomalies.

Images