Heating and ventilation equipment fault probability prediction method suitable for multiple working conditions

Abstract

The invention belongs to the field of heating and ventilation equipment fault prediction, and provides a heating and ventilation equipment fault probability prediction method suitable for multiple working conditions. Comprising the steps of defining equipment types and equipment parameters, collecting recorded data, constructing a data-fault mapping relation, constructing an equipment fault probability formula, reconstructing data, cleaning data, acquiring equipment operation and maintenance parameters, adjusting the equipment operation and maintenance parameters, predicting the equipment operation and maintenance parameters and classifying predicted operation and maintenance parameters. Through real-time operation and maintenance data prediction and fault judgment, the defect that an existing method depends on a single monitoring point, upper and lower limit alarm and manual tracking is overcome, and prediction and early warning of equipment faults are achieved; reliable prediction of different working conditions is realized by constructing an environment parameter matrix and a working condition set; by adjusting the operation and maintenance parameters of the equipment, the influence of manual operation on the operation and maintenance parameters of the equipment is reduced, and the reliability and accuracy of a final prediction result are improved.

Description

Technical Field

[0001] The invention relates to the field of HVAC equipment failure prediction, and in particular to a HVAC equipment failure probability prediction method adapting to multiple working conditions. Background Art

[0002] With the development of contemporary science and technology, a general trend of modern equipment development is to develop in the direction of complexity, intelligence and automation. The safety and reliability of equipment operation have an important impact on the national economy and people's livelihood, social stability, national resources and environment. The urgency of ensuring the safe and reliable operation of equipment is becoming increasingly prominent, and the safety and maintenance of equipment service have attracted more and more attention. As an essential component of a building, the stability and reliability of HVAC equipment have a great impact on the comfort of building personnel. The main significance of establishing the equipment failure probability prediction method is to assist equipment operation and maintenance personnel to achieve stable operation and on-demand maintenance of HVAC equipment, and prevent economic losses caused by equipment failure. HVAC equipment failures are mainly divided into two categories: one is hard failure and the other is soft failure. Hard failure refers to a complete failure of equipment and devices during operation, such as sudden shutdown of the fan, broken belt, no output signal from the sensor, etc. This type of failure is relatively easy to monitor. Soft failure refers to various failures in which the performance of equipment and devices is reduced or partially failed, such as the gradual blockage of the fan coil due to scaling. This type of failure is generally progressive, and the signs before it occurs are not obvious, and it is often difficult to monitor through sensors. In fact, soft faults are caused by the gradual deterioration of system parameters and are more harmful than hard faults.

[0003] At present, there is no mature prediction method to support the fault prediction of HVAC equipment. Traditional solutions rely on the establishment of physical models and the setting of thresholds. Such solutions have certain defects: they cannot adapt to all equipment models; they cannot adapt to more diverse working conditions; they rely too much on thresholds in fault prediction and are not accurate enough. In-service equipment is often in harsh working conditions, diverse, high-power, heavy-loaded and continuously running. Serious accidents caused by early fault development often occur. In order to eliminate its hidden faults and avoid safety accidents, modern industry urgently needs to adopt relevant monitoring technologies and inherent analysis technologies to ensure the safe service and operation of in-service equipment. Summary of the invention

[0004] In order to overcome the shortcomings of the prior art, the purpose of the present invention is to provide a method for predicting the probability of failure of HVAC equipment that is adaptable to multiple working conditions. By real-time prediction of operation and maintenance data and fault judgment, the defects of the existing methods that rely on a single monitoring point, upper and lower limit alarms, and manual tracking are solved, so as to realize the prediction and early warning of equipment failures; by constructing an environmental parameter matrix and an operating condition set, reliable prediction of different operating conditions is realized; by adjusting the equipment operation and maintenance parameters, the impact of manual operation on the equipment operation and maintenance parameters is reduced, thereby improving the reliability and accuracy of the final prediction results.

[0005] To achieve the above object, the present invention provides the following solutions:

[0006] A method for predicting the probability of failure of HVAC equipment suitable for multiple working conditions, comprising:

[0007] Define device type and device parameters;

[0008] Collecting recorded data for all of the described equipment types;

[0009] Building a data-fault mapping relationship based on the device type, the device parameters and the recorded data;

[0010] Constructing a device failure probability formula according to the data-fault mapping relationship and the recorded data;

[0011] Reconstructing the recorded data in chronological order to obtain reconstructed data;

[0012] Performing data cleaning on the reconstructed data to obtain historical data;

[0013] Constructing an environmental parameter matrix, using the environmental parameter matrix to obtain a set of operating conditions, and querying equipment operation and maintenance parameters based on the set of operating conditions;

[0014] Adjusting the equipment operation and maintenance parameters based on the regression model to obtain adjusted operation and maintenance data;

[0015] Predicting the equipment operation and maintenance parameters in the next time period according to the adjusted operation and maintenance data to obtain predicted operation and maintenance parameters;

[0016] The predicted operation and maintenance parameters are classified using the equipment failure probability formula to obtain equipment failure probability data.

[0017] Preferably, the equipment failure probability formula is:

[0018]

[0019] Among them, Pf t is the equipment failure probability data at prediction time t; S fis the fault state; s is the equipment state; N is the total number of equipment states; A is the equipment state transition probability; B is the occurrence probability of operation and maintenance parameters; S f is a fault state; Pf t-1 (s) is the probability that the device is in state s at time t-1.

[0020] Preferably, the environmental parameter matrix is:

[0021]

[0022] Among them, P x is the xth environmental parameter value; T i is the time node of the i-th row of data; P ij is the parameter value of the jth environment at the i-th time node.

[0023] Preferably, the calculation formula of the working condition set is:

[0024]

[0025] Wherein, K is the working condition set; K k is the set of environmental parameters for the kth working condition; is the e-th environmental parameter in the k-th operating condition; n k is the total number of environmental parameters under the kth working condition.

[0026] Preferably, the calculation formula for adjusting the operation and maintenance data is:

[0027] O ′ t =∑ p∈L [(p(t+δ)-p(t-δ))-(β0+β1I(t)+∈(t))];

[0028] Among them, O ′ t is the adjustment operation and maintenance data; p is the equipment operation and maintenance parameter; L is the total number of equipment operation and maintenance parameters; t is the prediction time; δ is the time window before and after the intervention; β0 is the intercept term; β1 is the regression coefficient; I(t) is the manual operation indicator function; ∈(t) is the random error term.

[0029] Preferably, the calculation formula for predicting operation and maintenance parameters is:

[0030]

[0031] Among them, O t+1 (O ′ t) is the predicted operation and maintenance parameter; T is the time step of the equipment operation and maintenance parameter; Softmax is a fully connected operation; σ is an activation function; For the i t The weight matrix under the time step index; For the i t The bias under the time step index; m(j t ) is the jth t The attention score at the time step index.

[0032] Preferably, the data cleaning includes: deleting duplicate values, deleting null values ​​and deleting outliers.

[0033] Preferably, the calculation formula for the device state transition probability is: The calculation formula for the occurrence probability of the operation and maintenance parameters is: Among them, N ij N is the number of times the i-th device state is transferred to the j-th device state; i is the number of occurrences of the i-th device state.

[0034] The present invention discloses the following technical effects:

[0035] The present invention provides a method for predicting the failure probability of HVAC equipment that is adaptable to multiple working conditions. By predicting operation and maintenance data and fault judgment in real time, the defects of the existing method that it relies on a single monitoring point, upper and lower limit alarms, and manual tracking are solved, and the prediction and early warning of equipment failures are realized; by constructing an environmental parameter matrix and an operating condition set, the problem of different degrees of equipment parameter changes under different operating conditions is solved, and reliable prediction of different operating conditions is realized; by adjusting the equipment operation and maintenance parameters, the influence of manual operation on the equipment operation and maintenance parameters is solved, and the prediction of the failure probability of HVAC equipment that takes human factors into consideration is realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0037] Figure 1 A schematic diagram of a HVAC equipment failure probability prediction process that is adaptable to multiple working conditions provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0039] The purpose of the present invention is to provide a method for predicting the probability of failure of HVAC equipment that is adaptable to multiple working conditions. By real-time prediction of operation and maintenance data and fault judgment, the defects of existing methods that rely on a single monitoring point, upper and lower limit alarms, and manual tracking are solved, so as to realize the prediction and early warning of equipment failures; by constructing an environmental parameter matrix and an operating condition set, reliable prediction of different operating conditions is realized; by adjusting the equipment operation and maintenance parameters, the influence of manual operation on the equipment operation and maintenance parameters is reduced, and the reliability and accuracy of the final prediction results are improved.

[0040] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments.

[0041] Figure 1 A schematic diagram of a HVAC equipment failure probability prediction process for multiple working conditions provided by an embodiment of the present invention is shown in FIG. Figure 1 As shown, the present invention provides a method for predicting the probability of failure of HVAC equipment that is adaptable to multiple working conditions, including:

[0042] Step 100: define device type and device parameters;

[0043] Step 200: Collecting recorded data of all device types;

[0044] Step 300: constructing a data-fault mapping relationship based on device type, device parameters and recorded data;

[0045] Step 400: constructing a device failure probability formula according to the data-fault mapping relationship and the recorded data;

[0046] Step 500: reconstructing the recorded data in chronological order to obtain reconstructed data;

[0047] Step 600: Clean the reconstructed data to obtain historical data;

[0048] Step 700: construct an environmental parameter matrix, use the environmental parameter matrix to obtain a set of working conditions, and query equipment operation and maintenance parameters according to the set of working conditions;

[0049] Step 800: adjusting the equipment operation and maintenance parameters based on the regression model to obtain adjusted operation and maintenance data;

[0050] Step 900: predicting the equipment operation and maintenance parameters for the next time period according to the adjusted operation and maintenance data to obtain predicted operation and maintenance parameters;

[0051] Step 1000: Classify the predicted operation and maintenance parameters using the equipment failure probability formula to obtain equipment failure probability data.

[0052] Specifically, the equipment failure probability formula is:

[0053]

[0054] Among them, Pf t is the equipment failure probability data at prediction time t; S f is the fault state; s is the equipment state; N is the total number of equipment states; A is the equipment state transition probability; B is the occurrence probability of operation and maintenance parameters; S f is a fault state; Pf t-1 (s) is the probability that the device is in state s at time t-1.

[0055] Furthermore, the environmental parameter matrix is:

[0056]

[0057] Among them, P x is the xth environmental parameter value; T i is the time node of the i-th row of data; P ij is the parameter value of the jth environment at the i-th time node.

[0058] Specifically, the calculation formula of the working condition set is:

[0059]

[0060] Among them, K is the working condition set; K k is the set of environmental parameters for the kth working condition; is the e-th environmental parameter in the k-th operating condition; n k is the total number of environmental parameters under the kth working condition.

[0061] Furthermore, the calculation formula for adjusting the operation and maintenance data is:

[0062]

[0063] Among them, O ′ t is the adjustment operation and maintenance data; p is the equipment operation and maintenance parameter;

[0064] L is the total number of equipment operation and maintenance parameters; t is the prediction time; δ is the time window before and after the intervention; β0 is the intercept term; β1 is the regression coefficient; I(t) is the manual operation indicator function; ∈(t) is the random error term.

[0065] Specifically, the calculation formula for predicting operation and maintenance parameters is:

[0066]

[0067] Among them, O t+1 (O ′ t ) is the predicted operation and maintenance parameter; T is the time step of the equipment operation and maintenance parameter; Softmax is a fully connected operation; σ is an activation operation; For the i t The weight matrix under the time step index; For the i t The bias under the time step index; m(j t ) is the jth t The attention score at the time step index.

[0068] Furthermore, data cleaning includes: deleting duplicate values, deleting null values, and deleting outliers.

[0069] Specifically, the calculation formula for the device state transition probability is: The calculation formula for the occurrence probability of operation and maintenance parameters is: Among them, N ij N is the number of times the i-th device state is transferred to the j-th device state; i is the number of occurrences of the i-th device state.

[0070] Specifically, the detailed process of HVAC equipment failure probability prediction includes:

[0071] S1, define the type of monitored equipment and the corresponding parameters. Equipment types include: chillers, fresh air units, air conditioning units, cooling water pumps, refrigeration water pumps, cooling towers. Environmental parameters include: temperature, relative humidity, heat load, number of building activities, indoor / outdoor air quality, and environmental parameters are represented by the symbol P.

[0072] S2, collects historical operation and maintenance data O, environmental parameters P, equipment failure data F (operation and maintenance data when equipment fails), equipment rated parameters, equipment factory operation curves, and manual operation records I (such as equipment maintenance, spare parts replacement, etc.) of various types of equipment.

[0073] S3, based on equipment failure data, equipment rated parameters, equipment factory operation curves and historical operation and maintenance data, classify the equipment operation status of each type of equipment and build a mapping relationship between "data-status". The relationship includes: health status Sh , the corresponding data are the equipment rated parameters, equipment factory operation curve, fault status S f , which can include multiple fault states such as minor faults and serious faults, corresponding to equipment fault data. The data corresponding to the above states are arranged in chronological order to form the equipment state set S s .

[0074] S4, reconstruct the historical operation and maintenance data O, environmental parameters P, manual operation records I, and equipment failure data F based on the timestamp, and arrange all the reconstructed data of the same model equipment at the same time in positive chronological order. Table 1 is the data structure of the chiller after reconstruction.

[0075] Table 1

[0076]

[0077] In Table 1, the value 0 corresponding to the manual operation record indicates no operation, and 1 indicates operation. i ∈S s .

[0078] S5, clean the reconstructed data in S4, remove duplicate values, null values, and abnormal values, and use them as historical data T.

[0079] S6, constructing a working condition classification function D, the execution steps of which are as follows:

[0080] S6.1, construct the environmental parameter matrix P x , whose expression is:

[0081]

[0082] S6.2, according to the matrix P x It is classified into k working conditions, and the set is represented by K. The value of k is manually set by the user. The larger the k value, the richer the working condition division. The classification method is shown in the following formula:

[0083]

[0084] S6.3, query the equipment operation and maintenance parameters O of the corresponding working condition according to the working condition set K t .

[0085] S7, build equipment operation and maintenance simulation model based on perceptron network, add pre-perception module to improve simulation accuracy, and simulate equipment operation and maintenance data for the next time period. t+1 The model construction process is as follows.

[0086] S7.1, construct a pre-perception module based on the regression model to make the model, which can automatically identify and adjust the changes in equipment operation and maintenance parameters caused by human intervention. The equipment operation and maintenance data screened by the working condition classification function of the perception module is the data set O t Make adjustments, and the adjustment logic is shown in the following formula:

[0087]

[0088] Among them, I(t) = 1 when there is manual operation at time t, otherwise it is 0. β0 is the intercept term of the regression model, which represents the baseline change when there is no manual operation. β1 is the coefficient of the regression model, which represents the degree of influence of manual operation on the parameter change. ∈(t) represents the random error term when no model is built.

[0089] S7.2, the equipment operation and maintenance parameters obtained by the pre-sensing module are used to simulate the equipment operation and maintenance parameters in the future time period. The mathematical formula of the simulation logic is as follows:

[0090]

[0091] i t Indicates the index of the time step, that is, when the model simulates the equipment operation and maintenance parameters at the next moment (time step t+1), i corresponds to the current time step t. The model processes the operation and maintenance data at time step t and outputs the simulation value. t Represents the index of the activation function in the model, corresponding to the output of different layers or nodes in the perceptron network. It is used to represent the contribution of the perceptron network to calculate the operation data in multiple dimensions at each time step t. m represents a perceptron network, which is used to evaluate the attention score of the processed time step. m(i t ) refers to the perceptron network in the operation and maintenance simulation model for time step i t The attention score of m(j t ) is similar to m(i t ), represents the attention scores at different time steps or different working conditions, and aggregates the states of multiple time steps through weighted summation to simulate the equipment operation and maintenance state at the next moment.

[0092] S8, constructing a device failure probability diagnosis model to infer the predicted device operation and maintenance data O t+1 The probability of causing failure in the construction process is as follows.

[0093] S8.1, according to the device state set S s The equipment state transition probability A and the equipment operation and maintenance parameter occurrence probability B under specified working conditions are obtained by combining with historical data T. The expression formulas are as follows:

[0094]

[0095] S8.2, determine the probability Pf of equipment failure at time t t , the formula is:

[0096]

[0097] When t=1, the probability of the device being in the state P ′ It is expressed by the following formula:

[0098]

[0099] In the above formula, N represents the total number of times all states of the device occur.

[0100] S9, execute the working condition classification function to retrieve the historical data T, input the output value as a data set into the equipment operation and maintenance simulation model for training respectively, and obtain the equipment operation and maintenance simulation model set M.

[0101] S10, obtaining the environmental parameters at the prediction time t, searching the equipment operation and maintenance simulation model set M, and obtaining the simulation model M that best matches the environmental parameters ′ .

[0102] S11, using simulation model M ′ Simulate the equipment operation and maintenance parameters T for the next time period at the prediction time t t+1 .

[0103] S12, using equipment failure probability diagnosis model to process T t+1 , and obtain the probability of equipment failure in this time period.

[0104] The beneficial effects of the present invention are as follows:

[0105] The present invention solves the defects of existing methods that rely on a single monitoring point, upper and lower limit alarms and manual tracking by predicting operation and maintenance data and fault judgment in real time, thereby realizing the prediction and early warning of equipment failures; by constructing an environmental parameter matrix and an operating condition set, reliable prediction of different operating conditions is realized; by adjusting the equipment operation and maintenance parameters, the influence of manual operation on the equipment operation and maintenance parameters is reduced, thereby improving the reliability and accuracy of the final prediction results.

[0106] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0107] The principles and implementation methods of the present invention are described in this article using specific examples. The description of the above embodiments is only used to help understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, according to the idea of ​​the present invention, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as limiting the present invention.

Claims

1. A method for predicting the probability of failure of HVAC equipment that is adaptable to multiple working conditions, characterized in that: include: Define device type and device parameters; Collecting recorded data for all of the described equipment types; Building a data-fault mapping relationship based on the device type, the device parameters and the recorded data; Constructing a device failure probability formula according to the data-fault mapping relationship and the recorded data; Reconstructing the recorded data in chronological order to obtain reconstructed data; Performing data cleaning on the reconstructed data to obtain historical data; Constructing an environmental parameter matrix, using the environmental parameter matrix to obtain a set of operating conditions, and querying equipment operation and maintenance parameters based on the set of operating conditions; Adjusting the equipment operation and maintenance parameters based on the regression model to obtain adjusted operation and maintenance data; Predicting the equipment operation and maintenance parameters in the next time period according to the adjusted operation and maintenance data to obtain predicted operation and maintenance parameters; The predicted operation and maintenance parameters are classified using the equipment failure probability formula to obtain equipment failure probability data.

2. A method for predicting the probability of failure of HVAC equipment adapting to multiple working conditions according to claim 1, characterized in that: The equipment failure probability formula is: Pf t =∑ s∈Sf Pf t-1 (s)AB; Among them, Pf t is the equipment failure probability data at prediction time t; S f is the fault state; s is the equipment state; N is the total number of equipment states; A is the equipment state transition probability; B is the occurrence probability of operation and maintenance parameters; S f is a fault state; Pf t-1 (s) is the probability that the device is in state s at time t-1.

3. A method for predicting the probability of failure of HVAC equipment adapting to multiple working conditions according to claim 1, characterized in that: The environmental parameter matrix is: Among them, P x is the xth environmental parameter value; T i is the time node of the i-th row of data; P ij is the parameter value of the jth environment at the i-th time node.

4. A method for predicting the probability of failure of HVAC equipment adapting to multiple working conditions according to claim 1, characterized in that: The calculation formula of the working condition set is: Wherein, K is the working condition set; K k is the set of environmental parameters for the kth working condition; is the e-th environmental parameter in the k-th operating condition; n k is the total number of environmental parameters under the kth working condition.

5. The method for predicting the probability of failure of HVAC equipment adapting to multiple working conditions according to claim 1 is characterized in that: The calculation formula for adjusting the operation and maintenance data is: O ′ t =∑ p∈L [(p(t+δ)-p(t-δ))-(β0+β1I(t)+∈(t))]; Among them, O ′ t is the adjustment operation and maintenance data; p is the equipment operation and maintenance parameter; L is the total number of equipment operation and maintenance parameters; t is the prediction time; δ is the time window before and after the intervention; β0 is the intercept term; β1 is the regression coefficient; I(t) is the manual operation indicator function; ∈(t) is the random error term.

6. A method for predicting the probability of failure of HVAC equipment adapting to multiple working conditions according to claim 1, characterized in that: The calculation formula for the predicted operation and maintenance parameters is: Among them, O t+1 (O ′ t ) is the predicted operation and maintenance parameter; T is the time step of the equipment operation and maintenance parameter; Softmax is a fully connected operation; σ is an activation function; For the i t The weight matrix under the time step index; For the i t The bias under the time step index; m(j t ) is the jth t The attention score at the time step index.

7. The method for predicting the failure probability of HVAC equipment adapting to multiple working conditions according to claim 1 is characterized in that: The data cleaning includes: deleting duplicate values, deleting empty values ​​and deleting outliers.

8. The method for predicting the probability of failure of HVAC equipment adapting to multiple working conditions according to claim 2 is characterized in that: The calculation formula of the device state transition probability is: The calculation formula for the occurrence probability of the operation and maintenance parameters is: Among them, N ij N is the number of times the i-th device state is transferred to the j-th device state; i is the number of occurrences of the i-th device state.

Images