Heating and ventilation equipment fault prediction system and method

Abstract

The invention relates to the technical field of fault prediction, in particular to a heating and ventilation equipment fault prediction system and method, and the system comprises a data collection module which collects the temperature, pressure, humidity and operation duration of heating and ventilation equipment and generates serialized input data, and a model initialization module which builds a fault prediction model according to the data and sets initial probability distribution, model parameters are formed, the online learning module adjusts and updates the model parameters in real time, and equipment behavior changes are dynamically matched. According to the method, fault prediction is more accurate and instant through real-time monitoring and integration of continuous data streams, the change of equipment behaviors can be dynamically adapted by establishing a fault prediction model based on the data and initializing and continuously updating model parameters, the prediction accuracy of future fault points is improved, and the fault prediction efficiency is improved. By dynamically adjusting the learning rate and the model parameters, the system can adapt to environmental changes, prediction failure caused by outdated models is avoided, the stability of equipment operation is improved, and the maintenance cost is reduced.

Description

Technical Field

[0001] The present invention relates to the technical field of fault prediction, and in particular to a HVAC equipment fault prediction system and method. Background Art

[0002] The field of fault prediction technology involves the use of algorithms and data analysis tools to predict possible equipment failures so that timely maintenance or replacement can be performed, thereby improving the reliability and efficiency of the equipment. This technology usually includes steps such as data collection, feature extraction, model training, and prediction output. Using real-time data collected by various sensors, machine learning or statistical models are used to analyze the data to predict the future state of the equipment and possible failure points. Fault prediction technology is widely used in many fields such as industrial production, energy management, and transportation. This technology is of great value, especially for equipment with high maintenance costs or serious consequences of failure.

[0003] Among them, the HVAC equipment failure prediction system refers to a failure prediction solution designed specifically for HVAC systems. Its purpose is to predict possible failures or performance degradation by continuously monitoring the operating status and performance of the equipment and analyzing data in real time. The system uses a prediction model to warn of potential problems by comparing historical data with real-time data, allowing facility managers to take measures in advance to avoid large-scale failures and expensive emergency repairs, thereby ensuring the efficient and stable operation of the HVAC system.

[0004] Existing technologies usually rely on preset fixed models and parameters, which often lead to inaccurate or delayed fault prediction when the equipment status or environment changes. Fixed models are difficult to respond to real-time changes in equipment and environment, which may lead to untimely fault diagnosis and increase the difficulty and cost of equipment maintenance. For example, if they fail to adapt to rapidly changing temperatures or pressures, existing systems may miss early warnings of key components, causing unexpected system downtime and affecting overall operational efficiency. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a HVAC equipment fault prediction system and method.

[0006] In order to achieve the above object, the present invention adopts the following technical solution: A HVAC equipment fault prediction system includes: The data acquisition module collects the operating data of HVAC equipment, including temperature, pressure, humidity and operating time, monitors data quality in real time, eliminates abnormal values ​​and interference in the data, integrates them into a data stream in a unified format, and generates serialized input data; The model initialization module receives the serialized input data, establishes a fault prediction model, and performs initial parameter settings, including defining a target state transition matrix and emission probability for each device state, setting an initial probability distribution, and forming initial model parameters; The online learning module continuously receives the serialized input data, and uses the collected data to adjust and update the initialization model parameters in real time, adjust the state transfer matrix and the emission probability, so that the fault prediction model dynamically matches the changes in the device behavior, and generates optimized model parameters; The fault detection module uses the optimized model parameters to analyze the current data of the equipment, detect the key points of deviation from the fault prediction model, evaluate the probability of occurrence of multiple faults in real time, determine the fault type according to the probability of occurrence of the fault and the model output, and generate a fault detection result; The system adaptive module dynamically adjusts the learning rate and parameters of the fault prediction model, including the state transfer matrix, according to the fault detection result, optimizes the model to match the current device state and environmental changes, and generates an adaptive adjustment result.

[0007] The serialized input data includes temperature data, pressure data, humidity data, and operating time data; the initialized model parameters include a state transfer matrix, a transmission probability, and an initial probability distribution; the optimized model parameters include an updated state transfer matrix and an updated transmission probability; the fault detection results include a fault occurrence probability and a determined fault type; and the adaptive adjustment results specifically include an updated learning rate and an adjusted model parameter.

[0008] As a further solution of the present invention, the step of obtaining the serialized input data is specifically as follows: Collect real-time data on HVAC equipment operation, including temperature, pressure, humidity, and operating hours, update data points every minute, and generate HVAC equipment raw data sets; Perform outlier detection and interference data removal on the original data set of the HVAC equipment, determine the abnormal threshold using the Z score method, remove data points that exceed three times the standard deviation, and obtain a cleaned data set; The cleaned data set is subjected to data serialization processing using the formula: ; Generate serialized input data ; in, A measurement value representing a single data point, used to represent the actual reading monitored from the device. Represents the average value of the measurement, which is used to balance the impact of outliers. Represents the total number of data points, providing the size of the dataset, is the weight of data point i, which is used to weight the criticality of each point in the overall data analysis.

[0009] As a further solution of the present invention, the step of obtaining the initialization model parameters is specifically as follows: Receiving the serialized input data, identifying differentiated equipment operation modes and fault characteristics through operation data analysis, defining the states of multiple equipment, and generating an equipment state definition set; Using the device state definition set, the state transition probability and emission probability are calculated, a state transition matrix and an emission probability matrix are constructed, and the probability of the device transitioning from one state to another and the probability of generating observation data under multiple states are combined to obtain a state transition and emission probability matrix set; Starting from the state transition and emission probability matrix set, the formula is adopted: ; Calculate the initial probability of each device state to form the initialization model parameters; in, represents the initial probability of state i, Represents the distance from state i to state j, which is used to adjust the probability basis of state transition. is the state transfer weight, emphasizing the importance of target state transfer. Indicates the total number of states.

[0010] As a further solution of the present invention, the step of obtaining the optimized model parameters is specifically as follows: Receive the serialized input data, monitor device behavior in real time, identify any deviations in device status or current failure modes, determine device status changes through data sampling and feature extraction, and generate device status change analysis results; Based on the equipment state change analysis results, refine the adjustment of the existing state transfer matrix and emission probability, re-estimate the probability parameters and inter-state probabilities through the Bayesian update method, and combine the time series analysis to generate the adjusted state transfer matrix and emission probability; Combining the adjusted state transfer matrix and emission probability, the formula is adopted: ; Generate optimized model parameters; in, represents the newly optimized model parameters, are the previous model parameters, Adjusted according to real-time data, the weights reflect the actual observed criticality of the state transitions, Ensure computing stability, Used to adjust how quickly the model responds to new data.

[0011] As a further solution of the present invention, the step of obtaining the fault detection result is specifically: Utilize the optimized model parameters to analyze the current operation data of the equipment, extract and identify the key points of deviation from the fault prediction model through fine-grained data features, monitor data fluctuations and abnormal indicators, and generate deviation key point analysis results; Based on the analysis results of the key points of the deviation, the probability of occurrence of various faults is calculated by applying the probability statistics method, and the influencing factors and probability of each fault mode are evaluated by combining the data pattern recognition technology, and the probability value of each fault type is output to generate the fault probability calculation result; The failure probability calculation results and model output are processed by decision analysis, using the formula: ; Obtain fault type assessment and generate fault detection results; in, Indicates the comprehensive evaluation results. Represents the calculated probability of failure, which is used to measure the predictability of failure occurrence. is the model’s response strength to faults, Set the baseline for fault determination. and Adjust the speed and accuracy of the model's response to new data.

[0012] As a further solution of the present invention, the step of obtaining the adaptive adjustment result is specifically: Analyze the fault detection results, identify model parameters that need to be adjusted, including learning rate and state transfer matrix, perform parameter analysis based on device state and environmental changes, and generate parameter adjustment requirements; Apply the parameter adjustment requirements, recalculate the values ​​of the learning rate and the state transfer matrix, adjust the parameters using the optimization algorithm, update the parameters, and generate the model adjustment results; Integrate the model adjustment results and update the fault prediction model using the formula: ; Optimize parameters based on the gradient descent method, match the working status of the current equipment, and generate adaptive adjustment results; in, represents the current model parameters, is the original parameter, is the adjusted learning rate, It is the parameter change calculated based on real-time data, which guides the direction and magnitude of model adjustment.

[0013] A method for predicting a fault of a heating and ventilation equipment is provided, wherein the method is performed based on the above-mentioned system for predicting a fault of a heating and ventilation equipment, and comprises the following steps: S1: Based on operation data collection, collect details of temperature, pressure, humidity and operation time, remove outliers and interference factors from the data, unify the data into a standard format, perform data quality monitoring, and generate serialized input data; S2: Based on the serialized input data, set the target state transfer matrix and emission probability of each device state, initialize the probability distribution, perform parameter setting, and establish the initialization model parameters; S3: Based on the initialized model parameters, continuously receive current serialized input data, monitor changes in device behavior in real time, adjust and update the state transfer matrix and emission probability in real time to dynamically match changes in device behavior, and generate optimized model parameters through multiple parameter adjustment actions; S4: Based on the optimized model parameters, analyze the current equipment data, monitor the key points of deviation from the model, assess the probability of fault occurrence, determine the fault type according to the probability of fault occurrence, and generate a fault detection result.

[0014] Compared with the prior art, the advantages and positive effects of the present invention are: In the present invention, through the integration of real-time monitoring and continuous data flow, fault prediction is made more accurate and immediate. By establishing a fault prediction model based on these data, and initializing and continuously updating the model parameters, it is possible to dynamically adapt to changes in equipment behavior and improve the accuracy of prediction of future fault points. By dynamically adjusting the learning rate and model parameters, the system can adapt to environmental changes and avoid prediction failures caused by outdated models, thereby improving the stability of equipment operation and reducing maintenance costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a system flow chart of the present invention; Figure 2 A flow chart of the steps for obtaining serialized input data of the present invention; Figure 3 A flowchart of the steps for obtaining the initialization model parameters of the present invention; Figure 4 This is a flow chart of the steps for obtaining the optimized model parameters of the present invention; Figure 5 A flowchart of the steps for obtaining the fault detection result of the present invention; Figure 6 This is a flow chart of the steps for obtaining the adaptive adjustment results of the present invention. DETAILED DESCRIPTION

[0016] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0017] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0018] Example 1: Please refer to Figure 1 The present invention provides a technical solution: a HVAC equipment fault prediction system comprising: The data acquisition module collects the operating data of HVAC equipment, including temperature, pressure, humidity and operating time, monitors data quality in real time, eliminates abnormal values ​​and interference in the data, integrates them into a data stream in a unified format, and generates serialized input data; The model initialization module receives serialized input data, establishes a fault prediction model, and performs initial parameter settings, including defining the target state transition matrix and emission probability for each device state, setting the initial probability distribution, and forming the initial model parameters; The online learning module continuously receives serialized input data, uses the collected data to adjust and update the initialization model parameters in real time, adjusts the state transition matrix and emission probability, makes the fault prediction model dynamically match the changes in equipment behavior, and generates optimized model parameters; The fault detection module uses the optimized model parameters to analyze the current data of the equipment, detect the key points of deviation from the fault prediction model, evaluate the probability of occurrence of multiple faults in real time, determine the fault type based on the probability of occurrence of the fault and the model output, and generate the fault detection results; The system adaptive module dynamically adjusts the learning rate and parameters of the fault prediction model, including the state transfer matrix, based on the fault detection results, optimizes the model to match the current device state and environmental changes, and generates adaptive adjustment results.

[0019] The serialized input data includes temperature data, pressure data, humidity data, and running time data. The initialized model parameters include the state transfer matrix, emission probability, and initial probability distribution. The optimized model parameters include the updated state transfer matrix and the updated emission probability. The fault detection results include the fault occurrence probability and the determined fault type. The adaptive adjustment results are specifically the updated learning rate and the adjusted model parameters.

[0020] See also Figure 2 , the specific steps for obtaining serialized input data are: Collect real-time data on HVAC equipment operation, including temperature, pressure, humidity, and operating hours, update data points every minute, and generate HVAC equipment raw data sets; In the process of collecting the operating data of HVAC equipment, the monitoring sensors are first set to the key parts of the equipment to facilitate the real-time collection of temperature, pressure, humidity and operating time. These data points are updated every minute to ensure the freshness and accuracy of the data. The key to this step is to correctly place the sensors and calibrate them to ensure the accuracy of data collection. The selection of monitoring points is set according to the key performance parameters of the equipment operation, including setting temperature and pressure monitoring points at the inlet and outlet of the heat exchanger, and setting humidity monitoring points near the air handling unit. The operating time is directly read through the operating system of the equipment. Each data point is marked with a timestamp to facilitate time series analysis in subsequent data processing, thereby generating the original data set of the HVAC equipment.

[0021] Perform outlier detection and interference data removal on the original data set of HVAC equipment, use the Z-score method to determine the abnormal threshold, remove data points that exceed three times the standard deviation, and obtain the cleaned data set; During the data cleaning process, the data set is first automatically screened for outliers through an algorithm. The key to this step is to use the Z-score method to determine whether the data exceeds the normal range. Data points that exceed three times the standard deviation will be considered abnormal and eliminated. This method is based on statistical principles and can effectively identify and eliminate data deviations caused by equipment failure or external interference. To this end, the Z-score calculation of each data point is based on the mean and standard deviation of the entire data set. The key in this process is to maintain consistency and accuracy in data processing to ensure that each data set after data cleaning can reflect the actual operating status of the equipment.

[0022] Perform data serialization processing on the cleaned data set using the formula: ; Generate serialized input data ; in, A measurement value representing a single data point, used to represent the actual reading monitored from the device. Represents the average value of the measurement, which is used to balance the impact of outliers. Represents the total number of data points, providing the size of the dataset, is the weight of data point i, which is used to weight the criticality of each point in the overall data analysis.

[0023] formula: ; The benefit of the formula is that by introducing the weight coefficient The square root operation enhances the sensitivity to abnormal data, and through weight adjustment, more attention can be paid to the changes of key data points, effectively improving the accuracy and flexibility of data serialization processing.

[0024] Detailed explanation of the formula and the process of formula calculation and derivation: Consider a specific data set, assuming that there are the following data points with temperature values ,average value , the total number of data points , and assuming that the weight of each data point is equal, that is, First, calculate the value: ; Then apply the square root and multiply by the weight : ; Adding these values ​​together gives: ; Finally, divide by Get serialized input data .

[0025] This result shows that the serialized input data It represents the average value after considering the weight and sensitivity of abnormal data, and provides an accurate and weighted measure of the current operating status of the equipment. Through such calculation, the health status of the equipment can be monitored more accurately and necessary adjustments or maintenance can be made.

[0026] See also Figure 3 , the specific steps to obtain the initialization model parameters are: Receive serialized input data, identify differentiated equipment operation modes and fault characteristics through operation data analysis, define the status of multiple equipment, and generate equipment status definition sets; The received serialized input data includes the operating information of multiple devices. These data are sorted by analyzing the operating modes and fault characteristics of the devices. The data are classified through a clustering algorithm, and a set of states is defined for each device state. These states reflect the performance of the device under differentiated operating conditions. The characteristics of multiple states are determined through statistical analysis, and the state set of the device is defined accordingly. The process involves the technical steps of data preprocessing, feature extraction and pattern recognition, which ensures the accuracy of the data set and the breadth of application, and forms a set of device state definitions.

[0027] Using the device state definition set, calculate the state transition probability and emission probability, construct the state transition matrix and emission probability matrix, and obtain the state transition and emission probability matrix set by combining the probability of the device transitioning from one state to another and the generation probability of observation data under multiple states; Using the device state definition set, combined with the state identification algorithm and probabilistic statistical methods, the transition probability from one state to another is calculated, including the probability evaluation of state transition events and the establishment of probability distribution. The state transition matrix M and the emission probability matrix E are established through a data-driven approach. These two matrices not only describe the transition probability of the device state, but also include the predictability of data observations under the target state. By analyzing batch data and optimizing the probabilistic model, the predictive ability of the model and the practicality of the application are enhanced.

[0028] Starting from the state transition and emission probability matrix set, the formula is adopted: ; Calculate the initial probability of each device state to form the initialization model parameters; in, represents the initial probability of state i, Represents the distance from state i to state j, which is used to adjust the probability basis of state transition. is the state transfer weight, emphasizing the importance of target state transfer. Indicates the total number of states.

[0029] formula: ; The benefit of the formula is that by introducing the distance and weight between states, the predictability of each state transition is adjusted, thereby optimizing the accuracy of state prediction, especially in complex equipment systems, and being able to better match the actual conditions of multiple state transitions.

[0030] Detailed explanation of the formula and the process of formula calculation and derivation: First, set the distance from state i to state j The distance is calculated by the change of the performance index of the device, and the difference between the two states is measured by the index of current fluctuation and temperature change. Then, the weight parameter is set. These weights are determined by the comprehensive factors of the working environment and state data of the equipment, reflecting the criticality of the transition from state i to state j; by calculating the distance of each pair of state transitions Perform logarithmic transformation to obtain the adjusted transition probability, and obtain the initial probability of state i through weighted average .

[0031] Set the distance from state 1 to state 2 , the distance from state 1 to state 3 , and the corresponding weights are and , the calculation process is as follows: ; The calculation results are: ; The results show that the initial probability of state 1 is about 0.406, which reflects the criticality of state 1 in the equipment system and provides a probabilistic basis for the subsequent state transition prediction.

[0032] See also Figure 4 , the steps to obtain the optimized model parameters are as follows: Receive serialized input data, monitor device behavior in real time, identify any deviations in device status or current failure modes, determine device status changes through data sampling and feature extraction, and generate device status change analysis results; Receive serialized input data, monitor device behavior in real time, and identify any deviations in device status or current fault modes through regular data sampling frequency and feature extraction technology. After preliminary screening and processing, these data determine the details of device status changes, and use statistical analysis methods to analyze data deviation trends and abnormal patterns. These analysis results will provide a basis for real-time updates of the fault prediction model, match the behavior changes of the equipment, and generate equipment status change analysis results.

[0033] Based on the results of the equipment state change analysis, the adjustment of the existing state transfer matrix and emission probability is refined, and the probability parameters and inter-state probabilities are re-estimated through the Bayesian update method. Combined with time series analysis, the adjusted state transfer matrix and emission probability are generated; Based on the results of the device state change analysis, refine the adjustment of the existing state transfer matrix and emission probability. This step involves using the Bayesian update method to re-estimate the state transition probability and emission probability. The focus is on adjusting these probability parameters in real time to match the current device behavior data, and calculating the current transition probability through a probabilistic statistical model. At the same time, combined with the time series data of device behavior, optimize the transition probability between each state. This requires not only calculating the probability value under the current state, but also predicting future state changes to generate an adjusted state transfer matrix and emission probability.

[0034] Combined with the adjusted state transfer matrix and emission probability, the formula is adopted: ; Generate optimized model parameters; in, represents the newly optimized model parameters, are the previous model parameters, Adjusted according to real-time data, the weights reflect the actual observed criticality of the state transitions, Ensure computing stability, Used to adjust how quickly the model responds to new data.

[0035] formula: ; The benefit of the formula is that it provides a method to dynamically adjust model parameters, allowing the model to match changes in device behavior, and by introducing the concepts of weight and distance, it enhances the model's sensitivity and responsiveness to new data.

[0036] Detailed explanation of the formula and the process of formula calculation and derivation: Set the original model parameters is 0.6, the smoothing coefficient is 0.5, the total number of states is 3, the weight between states The distance between states is set to 0.2, 0.3, and 0.5 respectively. Set to 1, 2, 3, small constant is 0.01, and the calculation process is as follows: ; The results show that the current model parameters It is 0.5565, which shows that the model adjusts its parameters after receiving new data to make it more suitable for the state changes of the current equipment and improve the accuracy of the prediction.

[0037] See also Figure 5 , the specific steps for obtaining the fault detection results are: Utilize the optimized model parameters to analyze the current operation data of the equipment, extract and identify the key deviation points from the fault prediction model through fine-grained data features, monitor data fluctuations and abnormal indicators, and generate deviation key point analysis results; The optimized model parameters are used to analyze the current operating data of the equipment. Through feature extraction and pattern recognition, the key points of deviation from the fault prediction model are identified, data fluctuations and abnormal indicators are monitored, and in-depth data analysis is performed, including time series analysis and anomaly point detection, in order to accurately monitor the equipment status and respond to potential failures. This method relies on continuous data flow and real-time processing algorithms to ensure the high sensitivity and responsiveness of the system, generate deviation key point analysis results, and provide the maintenance team with the necessary information to take preventive or repair measures.

[0038] Based on the analysis results of the key points of the deviation, the probability of occurrence of various faults is calculated using the probability statistics method, and the influencing factors and probability of each fault mode are evaluated by combining the data pattern recognition technology, and the probability value of each fault type is output to generate the fault probability calculation results; Based on the results of the deviation key point analysis, the probability of failure is calculated using statistical and probability theory methods, and the influencing factors of each failure mode are evaluated. The probability of occurrence of multiple failures is calculated by using Bayesian networks and decision trees. These calculations rely on the data collected from the equipment and its performance indicators, combined with the operating conditions and environmental factors of the equipment, to output the probability value of each failure type and generate failure probability calculation results, providing a scientific basis for subsequent failure type evaluation and failure response decisions.

[0039] Decision analysis is used to process the failure probability calculation results and model output, using the formula: ; Obtain fault type assessment and generate fault detection results; in, Indicates the comprehensive evaluation results. Represents the calculated probability of failure, which is used to measure the predictability of failure occurrence. is the model’s response strength to faults, Set the baseline for fault determination. and Adjust the speed and accuracy of the model's response to new data.

[0040] formula: ; The formula is useful in that by incorporating the probability of failure And the model output , the weighted logarithmic transformation enhances the model's sensitivity to small-probability failures, thereby more accurately identifying and predicting rare events. Adjusting parameters and Additional flexibility is provided to adjust fault detection thresholds and responses based on differentiated operating environments.

[0041] Detailed explanation of the formula and the process of formula calculation and derivation: Set the number of fault types , As an overall scaling factor, To avoid the situation where the denominator is zero, set the probability of each failure , , , model output , , , fault threshold , , . Calculate the contribution of multiple fault terms: ; ; ; Integrate multiple: ; The result shows that based on the current input parameters and model response, the overall fault risk assessment value is 0.0119, which is the unified output result after the model calculates each fault item. This value, as the final evaluation indicator of the step calculation, reflects the weighted impact intensity of each type of fault in the current state. This value can be used to further compare with the preset fault judgment threshold to determine whether to trigger the alarm mechanism or take subsequent processing measures, thereby completing the transition from quantitative calculation to actual judgment steps.

[0042] See also Figure 6 , the specific steps for obtaining the adaptive adjustment results are: Analyze fault detection results, identify model parameters that need to be adjusted, including learning rate and state transition matrix, perform parameter analysis based on device status and environmental changes, and generate parameter adjustment requirements; According to the fault detection results, the model parameters need to be adjusted, including the learning rate and state transfer matrix. The learning rate is obtained through experiments, and the convergence speed and stability of the model training under the differentiated learning rate need to be evaluated. The state transfer matrix depends on the state of the equipment and environmental changes. The transition probability between differentiated states needs to be set according to the equipment's fault and environmental monitoring data. For the adjustment of the learning rate, it is first necessary to calculate the preliminary learning rate value by collecting previous data, including factors such as equipment fault duration, fault frequency, and environmental changes, and determine the optimal value through experimental tests. The setting of the state transfer matrix needs to be based on the transition of the equipment under differentiated working conditions, combined with the state transfer model, and modeled through the Markov process to obtain the transition probability between states, and gradually adjusted and optimized in a data-driven manner to improve the accuracy of the prediction model.

[0043] Apply parameter adjustment requirements, recalculate the values ​​of learning rate and state transfer matrix, use optimization algorithm to adjust parameters, update parameters, and generate model adjustment results; When the application parameter adjustment needs to be met, the learning rate and state transfer matrix must first be recalculated based on the previous fault detection results. The recalculation of the learning rate depends on the experimental results or the feedback of the optimization algorithm, including the common gradient descent method. By adjusting the learning rate, the model parameters are optimally updated to ensure the convergence speed and stability during the training process. The adjustment process of the state transfer matrix is ​​more complicated, involving the optimization of the transition probability between equipment states through simulation and field test data. In this process, the optimization algorithms such as genetic algorithms and particle swarm optimization can effectively adjust the parameters of the state transfer matrix, thereby improving the model's matching ability and prediction accuracy for equipment state transitions in a dynamically changing environment. Through the cyclic optimization and feedback mechanism, the current adjusted parameters are obtained, so that the model can still operate stably and provide state prediction under changing environmental conditions.

[0044] Integrate the model adjustment results and update the fault prediction model using the formula: ; Optimize parameters based on the gradient descent method, match the working status of the current equipment, and generate adaptive adjustment results; in, represents the current model parameters, is the original parameter, is the adjusted learning rate, It is the parameter change calculated based on real-time data, which guides the direction and magnitude of model adjustment.

[0045] formula: ; The benefit of the formula is that it improves the model's prediction accuracy for faults by dynamically adjusting the learning rate and parameter updates to match changes in device status.

[0046] Detailed explanation of the formula and the process of formula calculation and derivation: Assume that the initial state transfer matrix parameters are , the adjusted learning rate is , parameter adjustment obtained through monitoring data analysis , substitute into the formula to calculate the current state transfer matrix parameters: ; The results show that by fine-tuning the learning rate, the model's adaptability to environmental changes is enhanced, which helps improve the sensitivity and accuracy of predictions.

[0047] A method for predicting a fault of a heating and ventilation equipment is provided. The method for predicting a fault of a heating and ventilation equipment is performed based on the above-mentioned system for predicting a fault of a heating and ventilation equipment, and comprises the following steps: S1: Based on operation data collection, collect details of temperature, pressure, humidity and operation time, remove outliers and interference factors from the data, unify the data into a standard format, perform data quality monitoring, and generate serialized input data; S2: Based on the serialized input data, set the target state transfer matrix and emission probability of each device state, initialize the probability distribution, perform parameter setting, and establish the initialization model parameters; S3: Based on the initialized model parameters, the current serialized input data is continuously received, the changes in device behavior are monitored in real time, and the state transition matrix and emission probability are adjusted and updated in real time to dynamically match the changes in device behavior. Through multiple parameter adjustment actions, the optimized model parameters are generated; S4: Based on the optimized model parameters, analyze the current equipment data, monitor the key points of deviation from the model, assess the probability of fault occurrence, determine the fault type based on the probability of fault occurrence, and generate fault detection results.

[0048] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. A HVAC equipment fault prediction system, characterized in that: The system comprises: The data acquisition module collects the operating data of HVAC equipment, including temperature, pressure, humidity and operating time, monitors data quality in real time, eliminates abnormal values ​​and interference in the data, integrates them into a data stream in a unified format, and generates serialized input data; The model initialization module receives the serialized input data, establishes a fault prediction model, and performs initial parameter settings, including defining a target state transition matrix and emission probability for each device state, setting an initial probability distribution, and forming initial model parameters; The online learning module continuously receives the serialized input data, and uses the collected data to adjust and update the initialization model parameters in real time, adjust the state transfer matrix and the emission probability, so that the fault prediction model dynamically matches the changes in the device behavior, and generates optimized model parameters; The fault detection module uses the optimized model parameters to analyze the current data of the equipment, detect the key points of deviation from the fault prediction model, evaluate the probability of occurrence of multiple faults in real time, determine the fault type according to the probability of occurrence of the fault and the model output, and generate a fault detection result; The system adaptive module dynamically adjusts the learning rate and parameters of the fault prediction model, including the state transfer matrix, according to the fault detection result, optimizes the model to match the current device state and environmental changes, and generates an adaptive adjustment result.

2. The HVAC equipment failure prediction system according to claim 1, characterized in that: The serialized input data includes temperature data, pressure data, humidity data, and operating time data; the initialized model parameters include a state transfer matrix, a transmission probability, and an initial probability distribution; the optimized model parameters include an updated state transfer matrix and an updated transmission probability; the fault detection results include a fault occurrence probability and a determined fault type; and the adaptive adjustment results specifically include an updated learning rate and an adjusted model parameter.

3. The HVAC equipment failure prediction system according to claim 2, characterized in that: The steps for obtaining the serialized input data are specifically as follows: Collect real-time data on HVAC equipment operation, including temperature, pressure, humidity, and operating hours, update data points every minute, and generate HVAC equipment raw data sets; Perform outlier detection and interference data removal on the original data set of the HVAC equipment, determine the abnormal threshold using the Z score method, remove data points that exceed three times the standard deviation, and obtain a cleaned data set; The cleaned data set is subjected to data serialization processing using the formula: ； Generate serialized input data ; in, A measurement value representing a single data point, used to represent the actual reading monitored from the device. Represents the average value of the measurement, which is used to balance the impact of outliers. Represents the total number of data points, providing the size of the dataset, is the weight of data point i, which is used to weight the criticality of each point in the overall data analysis.

4. The HVAC equipment failure prediction system according to claim 3, characterized in that: The steps for obtaining the initialization model parameters are specifically as follows: Receiving the serialized input data, identifying differentiated equipment operation modes and fault characteristics through operation data analysis, defining the states of multiple equipment, and generating an equipment state definition set; Using the device state definition set, the state transition probability and the emission probability are calculated, a state transition matrix and an emission probability matrix are constructed, and the probability of the device transitioning from one state to another and the generation probability of observation data under multiple states are combined to obtain a state transition and emission probability matrix set; Starting from the state transition and emission probability matrix set, the formula is adopted: ； Calculate the initial probability of each device state to form the initialization model parameters; in, represents the initial probability of state i, Represents the distance from state i to state j, which is used to adjust the probability basis of state transition. is the state transfer weight, emphasizing the importance of target state transfer. Indicates the total number of states.

5. The HVAC equipment failure prediction system according to claim 4, characterized in that: The steps for obtaining the optimized model parameters are specifically as follows: Receive the serialized input data, monitor device behavior in real time, identify any deviations in device status or current failure modes, determine device status changes through data sampling and feature extraction, and generate device status change analysis results; Based on the equipment state change analysis results, refine the adjustment of the existing state transfer matrix and emission probability, re-estimate the probability parameters and inter-state probabilities through the Bayesian update method, and combine the time series analysis to generate the adjusted state transfer matrix and emission probability; Combining the adjusted state transfer matrix and emission probability, the formula is adopted: ； Generate optimized model parameters; in, represents the newly optimized model parameters, are the previous model parameters, Adjusted according to real-time data, the weights reflect the actual observed criticality of the state transitions, Ensure computing stability, Used to adjust how quickly the model responds to new data.

6. The HVAC equipment failure prediction system according to claim 5, characterized in that: The steps for obtaining the fault detection result are specifically as follows: Utilize the optimized model parameters to analyze the current operation data of the equipment, extract and identify the key points of deviation from the fault prediction model through fine-grained data features, monitor data fluctuations and abnormal indicators, and generate deviation key point analysis results; Based on the analysis results of the key points of the deviation, the probability of occurrence of various faults is calculated by applying the probability statistics method, and the influencing factors and probability of each fault mode are evaluated by combining the data pattern recognition technology, and the probability value of each fault type is output to generate the fault probability calculation result; The failure probability calculation results and model output are processed by decision analysis, using the formula: ； Obtain fault type assessment and generate fault detection results; in, Indicates the comprehensive evaluation results. Represents the calculated probability of failure, which is used to measure the predictability of failure occurrence. is the model’s response strength to faults, Set the baseline for fault determination. and Adjust the speed and accuracy of the model's response to new data.

7. The HVAC equipment failure prediction system according to claim 6, characterized in that: The steps for obtaining the adaptive adjustment result are specifically as follows: Analyze the fault detection results, identify model parameters that need to be adjusted, including learning rate and state transfer matrix, perform parameter analysis based on device state and environmental changes, and generate parameter adjustment requirements; Apply the parameter adjustment requirements, recalculate the values ​​of the learning rate and the state transfer matrix, adjust the parameters using the optimization algorithm, update the parameters, and generate the model adjustment results; Integrate the model adjustment results and update the fault prediction model using the formula: ； Optimize parameters based on the gradient descent method, match the working status of the current equipment, and generate adaptive adjustment results; in, represents the current model parameters, is the original parameter, is the adjusted learning rate, It is the parameter change calculated based on real-time data, which guides the direction and magnitude of model adjustment.

8. A method for predicting failure of HVAC equipment, characterized in that: The HVAC equipment fault prediction system according to any one of claims 1 to 7 comprises the following steps: Based on operation data collection, collect details of temperature, pressure, humidity and operation time, remove outliers and interference factors from the data, unify the data into a standard format, perform data quality monitoring, and generate serialized input data; Based on the serialized input data, the target state transfer matrix and emission probability of each device state are set, the probability distribution is initialized, the parameters are set, and the initialization model parameters are established; Based on the initialized model parameters, the current serialized input data is continuously received, the changes in the device behavior are monitored in real time, the state transfer matrix and the emission probability are adjusted and updated in real time to dynamically match the changes in the device behavior, and the optimized model parameters are generated through multiple parameter adjustment actions; Based on the optimized model parameters, the current equipment data is analyzed, the key points of deviation from the model are monitored, the probability of fault occurrence is assessed, the fault type is determined based on the probability of fault occurrence, and the fault detection result is generated.

Images