CN115899964A - Multidimensional-based intelligent air conditioner monitoring method and system - Google Patents
Multidimensional-based intelligent air conditioner monitoring method and system Download PDFInfo
- Publication number
- CN115899964A CN115899964A CN202211657413.8A CN202211657413A CN115899964A CN 115899964 A CN115899964 A CN 115899964A CN 202211657413 A CN202211657413 A CN 202211657413A CN 115899964 A CN115899964 A CN 115899964A
- Authority
- CN
- China
- Prior art keywords
- model
- data
- fault
- air
- air conditioning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Abstract
The invention relates to an air conditioner intelligent monitoring method and system based on multiple dimensions. The method comprises the following steps: collecting air conditioner data through sensing equipment; uploading the collected air conditioner data to a model modeling tool, and establishing a system performance evaluation model, a system data driving model, a system fault diagnosis model and a knowledge base model; and obtaining a diagnosis evaluation result of the air conditioner performance according to the system performance evaluation model, the system data driving model, the system fault diagnosis model and the knowledge base model. The air conditioning system is monitored in a centralized manner based on an intelligent monitoring algorithm model and a big data analysis technology, the operation process of the air conditioning system participating in production is monitored, evaluated and analyzed in real time, equipment or system abnormity is found in time, effective information support can be provided for unattended field equipment, comprehensive, safe and reliable production task process of the air conditioning system is guaranteed, and auxiliary decision support is provided for production task organization and command.
Description
Technical Field
The invention relates to the technical field of big data and artificial intelligence, in particular to an air conditioner intelligent monitoring method and system based on multiple dimensions.
Background
In the production process, an enterprise carries out real-time monitoring, evaluation and analysis on the air conditioning system participating in the production guarantee, finds out equipment or system abnormality in time, provides effective information support for unattended field equipment, ensures the safety and reliability of the air conditioning system in the production task process, and provides auxiliary decision support for production task organization and command.
In practice, the inventors found that the above prior art has the following disadvantages: the information of the air conditioning system is not comprehensively and accurately controlled in the production process of an enterprise, so that the production task organization command decision is influenced, and certain hidden danger is caused to production safety.
Disclosure of Invention
In order to solve the technical problems, the invention provides an air conditioner intelligent monitoring method and system based on multiple dimensions.
The technical scheme adopted by the invention is as follows:
an air conditioner intelligent monitoring method based on multiple dimensions comprises the following steps:
collecting air conditioner data through sensing equipment;
uploading the collected air conditioner data to a model modeling tool, and establishing a system performance evaluation model, a system data driving model, a system fault diagnosis model and a knowledge base model;
and obtaining the diagnosis and evaluation result of the air conditioner performance according to the system performance evaluation model, the system data driving model, the system fault diagnosis model and the knowledge base model.
Further, the system performance evaluation model analyzes and evaluates the performance of the air conditioning system in real time according to an established evaluation index system, wherein evaluation indexes in the evaluation index system comprise:
1) Control accuracy range: the air conditioning unit can maintain the fluctuation range of the temperature and the relative humidity while directly providing the processing air to the closed space, the room or the area;
2) Refrigerating capacity: the sum of the heat removed by the air conditioning unit from the enclosed space, room or area per unit time;
3) Cooling power consumption: the total power consumed by the air conditioning unit under the specified refrigeration test working condition;
4) Coefficient of performance: the ratio of the refrigerating capacity Q and the refrigerating consumed power P of the air conditioning unit expressed by the same unit;
5) The total load IQ of the comprehensive performance;
6) The overall performance coefficient ICOP;
7) Precision curve: after the indoor state is balanced for 1h, testing the fluctuation of the temperature and the humidity of measuring points arranged in the room, and drawing a precision curve graph;
8) Refrigeration performance: the method comprises the steps of refrigeration temperature and humidity precision, refrigeration temperature and humidity stability and refrigeration temperature and humidity efficiency.
Further, the system performance evaluation model is an analysis evaluation model of the air conditioning system, which is established by adopting a grey correlation analysis method and comprehensively considering economy, energy conservation and environmental protection.
Furthermore, the system data driving model grasps the inherent relation between the variables and the parameters by means of a large amount of historical data of the air conditioning system under various operating conditions, including normal data and fault data, and further judges the fault condition in the new data and isolates the fault source through a mathematical model established in the machine learning process.
Furthermore, the system data driving model adopts Principal Component Analysis (PCA) method to diagnose faults of the process by utilizing the correlation among process multivariable, according to the historical data of the process variables, a multivariate projection method is utilized to decompose a multivariable sample space into a projection subspace with lower dimension formed by the principal component variables and a corresponding residual subspace, statistics capable of reflecting space change are respectively constructed in the two subspaces, then the original observation vectors are respectively projected to the two subspaces, and corresponding statistic indexes are calculated for process monitoring.
Furthermore, the system fault diagnosis model adopts the time sequence data of the guarantee parameters and the process data of the LSTM training air conditioning system to generate a time sequence prediction model of the guarantee parameters and the process data, selects a proper prediction step length according to the precision of the time sequence prediction model, and obtains predicted air conditioning system performance and risk data information by using the guarantee parameters and the process data information obtained through prediction to realize the early warning capability of the air conditioning system.
Further, the system fault diagnosis model predicts the safeguard parameters and the process data by adopting the following steps:
the method comprises the following steps: selecting historical data information with non-fault scenes and typical fault scenes, and using 80% of the historical data information as training of a prediction model and 20% as verification of the prediction model;
step two: carrying out standardization processing on historical data information;
step three: taking the guarantee parameters and the process data as the input and the output of a prediction model LSTM, inputting the time sequence data of the first 30min of selection, outputting the time sequence data of the last 15min of selection, setting the precision of the model to be 95%, and continuously training the model to ensure that the precision reaches the set requirement;
step four: selecting 20% of data quantity to verify the prediction model, wherein the prediction model meets the requirement when the verification precision reaches 95%, otherwise, returning to the third step to continue training the prediction model until the verification reaches the set requirement;
step five: and collecting real-time guarantee parameters and process data, and respectively predicting the guarantee parameters and the process data in the next 15 minutes by using the trained prediction model.
Further, the knowledge base model comprises an air conditioning system knowledge base, typical fault types are built in the knowledge base, and the knowledge base model comprises national and industrial specifications, a base operation instruction book and an emergency disposal plan; the method comprises the following steps of (1) inputting labels for a knowledge base, a typical fault base and a case base when the knowledge base, the typical fault base and the case base are created, wherein the labels comprise the knowledge base of a chilled water pump, the case base of a cooling tower and the fault base of the chilled water pump; in a typical fault library, the information of the created fault library can be checked, a fuzzy search mode of directly inputting characters is supported, and display according to label classification is also supported.
A multi-dimensional based air conditioner intelligent monitoring system comprises:
the system state monitoring module is used for realizing the situation perception of the state of the air-conditioning equipment and the system process and the real-time interpretation of the state of the air-conditioning equipment, key parameters and the system state through data acquisition of the air-conditioning equipment of each workshop;
the diagnosis evaluation analysis module is used for realizing the evaluation of the performance of the air conditioner by utilizing the system performance evaluation model, the system data driving model, the system fault diagnosis model and the knowledge base model, realizing the representation, query, retrieval and reasoning of knowledge through the management of the knowledge base and the fault base, providing fault detection and positioning capability and assisting in troubleshooting and tracking;
and the task auxiliary decision-making module is used for realizing the fault risk assessment, the technical state change influence assessment and the auxiliary judgment of the risk level of the air conditioning equipment according to the assessment and analysis result obtained by the diagnosis assessment and analysis module.
The invention has the following beneficial effects:
the air conditioning system is monitored in a centralized manner based on an intelligent monitoring algorithm model and a big data analysis technology, the operation process of the air conditioning system participating in production is monitored, evaluated and analyzed in real time, equipment or system abnormity is found in time, effective information support can be provided for unattended field equipment, comprehensive, safe and reliable production task process of the air conditioning system is guaranteed, and auxiliary decision support is provided for production task organization and command.
Drawings
FIG. 1 is a schematic diagram of the overall architecture of the present invention.
Fig. 2 is a flow chart of air conditioning system diagnostic evaluation.
Detailed Description
The present invention will be described in further detail below with reference to specific examples and the accompanying drawings.
In a first aspect, an embodiment of the present invention provides a multidimensional intelligent monitoring method for an air conditioner, where the method includes:
the collected air conditioner data are uploaded to a model modeling tool through the API through the sensing equipment, models such as a system performance evaluation model, a system data driving model, a system fault diagnosis model and a knowledge base model are built, and results output by the models are uploaded to an air conditioner monitoring system through the API to be evaluated and displayed in a database, as shown in figure 1.
1. System performance evaluation model
The field air conditioning system provides proper air for the factory building, so that the factory building environment (temperature, humidity, cleanliness and the like) can always meet the specified index requirements. Therefore, the performance of the air conditioning system needs to be analyzed and evaluated in real time, and an evaluation index system is established first.
1.1 Evaluation index of)
1.1.1 Control accuracy range)
Air conditioning assemblies maintain a range of temperature and relative humidity fluctuations while providing process air directly to an enclosed space, room or area.
1.1.2 Refrigerating capacity)
The sum of the heat removed by the unit from an enclosed space, room or area per unit time, unit: w is added.
Q＝G a (h a1 -h a2 )/[V a (1+W a )]
In the formula:
G a indoor air flow measurement in cubic meters per second (m) 3 /s)；
h a1 -enthalpy of air entering the room, in units of coke per kilogram (J/kg) relative to dry air;
h a2 -enthalpy of air leaving the indoor side, in units of coke per kilogram (J/kg) relative to dry air;
V a specific volume of air at the nozzle in cubic meters per kilogram (m) 3 /kg)；
W a Air at the nozzleThe moisture content is in kilograms per kilogram (kg/kg) relative to dry air.
1.1.3 Power consumption for refrigeration)
Under the specified refrigeration test working condition, the total power consumed by the unit is as follows: w.
1.1.4 Coefficient of performance (COP)
The ratio of the refrigerating capacity Q and the refrigerating consumed power P expressed by the same unit is taken as the coefficient of performance COP.
COP＝Q/P (0.1)
1.1.5 ) overall performance Total load (IQ)
The total load IQ of the comprehensive performance of the unit is calculated according to the following formula:
IQ＝|Q r0 |+|Q s0 |+∫ 0 |q re |dt+∫ 0 |q se |dt
in the formula:
Q r0 the indoor ICOP air handler increases or decreases the initial heat load to the room, so that the room state changes from the set working condition to the starting working condition, and the unit is coke (J);
Q s0 the indoor ICOP air handler increases or decreases the initial wet load to the room, so that the room state changes from the set working condition to the starting working condition, and the unit is coke (J);
q re -after starting up, the indoor ICOP air handler increases or decreases the heat load in watts (W) to the room per unit time;
q se -after power-on, the indoor ICOP air handler increases or decreases the moisture load to the room in watts (W) per unit time. Calculating the humidification quantity and the steam enthalpy value according to the humidity difference between the starting working condition and the set working condition: q. q.s se And = wh, wherein w is the mass flow of the added steam and h is the latent heat of vaporization of the water vapor at the target temperature.
τ -the time consumption of the unit from start to stop or the time each component runs separately, in seconds(s).
1.1.6 Coefficient of comprehensive Properties (ICOP)
The comprehensive performance coefficient of the unit is calculated according to the following formula:
ICOP＝IQ/IP
in the formula:
IQ-Total Performance Total load (including Heat load, wet load)
IP-Total energy consumption of comprehensive Properties.
In the formula:
i-number of running parts of unit, 1,2 \8230;, n
p i Power consumption of each functional unit in watt (W)
1.1.7 Precision curve
After the indoor state is balanced for 1h, the fluctuation of the temperature and the humidity of the measuring points arranged in the measuring room is tested, and a precision curve graph can be drawn.
1.1.8 Performance of refrigeration system)
The air conditioner refrigeration performance evaluation model is mainly used for ensuring that the indoor site meets the set requirements of constant temperature and constant humidity, and selecting the precision, stability and efficiency of the temperature and humidity so as to obtain the air conditioner refrigeration performance evaluation model. The specific evaluation index is defined as shown in table 1:
TABLE 1 air conditioning system safeguard parameters
| Serial number | Parameter name | Unit of | Parameter(s) | Guarantee value |
| 1 | Temperature in the field | ℃ | Temp | 26.00 |
| 2 | Humidity in the field | ％ | Humidity | 95.00 |
1.1.8.1 Precision of refrigeration temperature and humidity
Respectively calculating the guarantee precision of the temperature and the humidity every hour, every day, every week and every month, wherein the calculation formula is as follows:
A T ＝1-(AVG(T t1 ，T t2 ，...,T tn )-T)/T*100％
A H ＝1-(AVG(H t1 ，H t2 ，...,H tn )-H)/H*100％
wherein A represents the accuracy of the calculation, a T Indicating the accuracy of the temperature, A H Indicating the accuracy of the humidity, temperature T being the set guaranteed temperature, H being the set guaranteed humidity, T t1 Representing the acquisition temperature, H, at time t1 t1 Indicating the acquisition humidity at time T1, AVG (T) t1 ，T t2 ，...,T tn ) Represents the average value of the collected temperatures at the time points t1 to tn, AVG (H) t1 ，H t2 ，...,H tn ) Representing the average humidity values collected from time t1 to tn.
1.1.8.2 Temperature and humidity stability of refrigeration
Respectively calculating the temperature and humidity guarantee stability of each hour, each day, each week and each month, wherein the calculation formula is as follows:
S 2 T ＝[(T t1 -T) 2 +(T t2 -T) 2 +...+(T tn -T) 2 ]/n
S 2 H ＝[(H t1 -H) 2 +(H t2 -H) 2 +...+(H tn -H) 2 ]/n
wherein S 2 The variance of the temperature, i.e. the stability, is indicated. S 2 T Denotes the stability at temperature, S 2 H Indicating the stability of the humidity, temperature T being the set guaranteed temperature, H being the set guaranteed humidity, T t1 Representing the acquisition temperature, H, at time t1 t1 Represents the collection humidity at the time t1, and n is the number of collection points.
1.1.8.3 Temperature and humidity efficiency of refrigeration
Efficiency is understood as the rapid degree of attainment of the stated temperature and humidity, assuming a standard time of 1min for the temperature to reach 27 ℃ from 26 ℃, denoted as E TS The standard time for the humidity to reach 96% from 95% is 1min and is recorded as E Hs . The calculation formula of the guaranteed efficiency of the temperature and the humidity is as follows:
E T ＝(T T -E TS *step)/(E TS *step)*100％
E H ＝(T H -E HS *step)/(E HS *step)*100％
wherein E represents the efficiency, E T Efficiency in temperature, E H Efficiency in terms of humidity, step being the step size of the temperature or humidity change, T T Indicates the time, T, required for the temperature to actually reach the specified temperature H Indicating the time required for the humidity to actually reach the specified humidity.
1.2 Index evaluation model
On the basis of an index evaluation system, an index evaluation model is constructed, and the air conditioner performance change trend is mastered through data driving.
For comprehensive evaluation of a specific air conditioning system, various indexes such as system performance, reliability, efficiency and the like are involved, wherein the comparison of quantitative indexes and the description of qualitative indexes are typical multi-objective decision problems. Therefore, in order to establish the optimization of the air conditioning system solution on a scientific basis, it is necessary to use an objective and quantitative comprehensive evaluation method. The invention adopts a gray multilevel comprehensive evaluation method, comprehensively considers indexes of the scheme such as economy, energy conservation, environmental protection and the like, establishes an analysis evaluation model of the air-conditioning system, optimizes the scheme of the air-conditioning system and selects an optimal scheme. By applying the gray multilevel comprehensive evaluation method to the scheme evaluation of the air conditioning system, the one-sidedness and the blindness in the evaluation process can be reduced, and the evaluation result is more objective and comprehensive.
The grey multilevel comprehensive evaluation method is used for carrying out grey correlation analysis, the grey correlation analysis is a method for describing the strength, the size and the sequence of the relationship among the factors by using a grey correlation degree sequence, and the grey correlation analysis is a method for analyzing and determining the influence degree among the system factors or the contribution measure of the factors to the main behavior of the system by using the grey correlation degree. The grey correlation degree is a quantity used for describing the closeness degree of the relationship among the system factors, and is a measure of the change situation of the system, generally speaking, the change situation of the quantifiable system can be characterized by the change situation of the sequence, and the change situation of each sequence always changes according to a certain magnitude and trend (referring to a curve shape). Therefore, the closeness of the relationship between the system sequences is represented by the similarity of magnitude changes and the similarity of development trends between the two, which are two expression forms of gray correlation analysis, which are both different and mutually restricted. The magnitude change of magnitude can be measured by displacement difference (distance between points), and the development trend can be measured by first-order or second-order slope difference, so that the correlation can be characterized by the displacement difference and the slope difference (speed and acceleration).
The grey correlation analysis has no excessive requirements on the sample size, does not need a typical distribution rule, has small calculation amount, and the result is consistent with the result of qualitative analysis. The grey correlation analysis can deeply analyze and depict the essence and connotation of correlation between objects, because the situation consistency of any two objects in the development process is mainly reflected in the aspects of overall displacement difference, overall first-order slope difference, overall second-order slope difference and the like, and several existing correlation models are established on the basis of fully considering the displacement difference or slope difference among factors.
1.3 Application step of index evaluation model
Step 1, determining an evaluation index. Various attributes or performances of the air conditioning system to be evaluated, such as economy, energy conservation, environmental protection, reliability and the like, are selected, the quality of the evaluation object is comprehensively reflected on the basis of indexes, and a basis for evaluating the evaluation object is provided.
And 2, determining an optimal index set, selecting an optimal index from the same indexes of the multiple evaluation objects, and forming the optimal index set based on the optimal values of the multiple evaluation indexes.
And 3, carrying out data dimensionless processing, namely carrying out dimensionless processing on the original data (production data) based on a gray optimization process by combining a sequence formed by all factors (attributes and performances).
Step 4, establishing a difference sequence, and calculating corresponding absolute differences (differences between each factor sequence and original data) to form an absolute difference matrix;
step 5, determining an evaluation matrix, and providing data of an absolute difference matrix, namely matrix data = subtracting original data from each factor sequence;
step 6, determining a weight matrix of the evaluation index;
and 7, performing gray comprehensive evaluation by using the evaluation matrix in the step 5 and the weight matrix of the evaluation index in the step 6, and verifying the performance evaluation index system of the air conditioning system.
2. System data driven model
The data driving is based on a data processing method, a physical model is not required to be constructed, only a great amount of historical data under various operating conditions, including normal data and fault data, are used for mastering the inherent relation between variables and parameters, and a mathematical model established through the machine learning process can be used for judging the fault condition in new data and isolating a fault source. Data mining techniques are widely used to detect and diagnose sensor faults in air conditioning systems, and these applications typically extract correlations between variable data or hidden features of the system by collecting data, machine learning, inductive deduction, identifying faults or energy consumption patterns. Principal Component Analysis (PCA), among others, is a fault detection method that has been successfully used to build a purely mathematical model in many refrigeration air conditioning systems.
Principal Component Analysis (PCA) is one of multivariate statistical methods, and as a general dimension reduction tool, it adopts a method of projecting original high-dimensional data to a low-dimensional space through linear transformation, and retaining main information. The guiding idea of PCA is to find a projection method which can reflect the original data information most so as to ensure that the data after dimensionality reduction can not be distorted. In a data matrix formed by a plurality of multi-dimensional samples with a certain correlation relationship, a small number of new variables without correlation relationship are established, so that the new variables reflect main information represented in original high-dimensional data more intensively.
When a sensor fails, the correlation between the measured data changes, and the residual error may increase significantly, so that when the component of the sampled data fails, the residual error vector of the data deviates in the dimension (the deviated variable). This deviation will result in an increased contribution rate in the dimension. Therefore, by determining the dimension in which the maximum contribution rate is located, it can be determined that the sensor is located as a fault source. Wherein, the contribution rate refers to calculating multivariate statistical weight.
The fault diagnosis method based on multivariate statistical analysis is to diagnose the fault of the process by utilizing the correlation among process multivariable. According to the method, a multivariate sample space is decomposed into a projection subspace with lower dimensions and a corresponding residual error subspace formed by principal component variables by utilizing a multivariate projection method according to historical data of process variables, statistics which can reflect the space change are respectively constructed in the two subspaces, then original observation vectors are respectively projected to the two subspaces, and corresponding statistic indexes are calculated for process monitoring.
PCA decomposes a sample matrix or a sample variance-covariance matrix of the process variables, and selected principal component variables are uncorrelated and can be derived from the process variables in a linear combination. The projection subspace obtained by the PCA method reflects the main variations of the process variables, while the residual subspace mainly reflects the noise, interference, etc. of the process.
For data matrix
The number of sample points is n, and p is the number of variables. By x i ＝[x i1 x i2 …x ip ] T I =1,2, \ 8230, n denotes the ith sample point in the data table, each variable is denoted by v j ＝[v 1j v 2j …v nj ] T J =1,2, \8230;, p.
According to the fundamental principle of principal component analysis, principal component transformation is to project original data X in the principal component direction, and includes:
t j ＝Xp j
in statistical monitoring, the original component coefficient vector p j Called the load vector, principal component vector t j Referred to as a score vector, in matrix form:
T＝XP
wherein, P is called a load matrix, T is called a score matrix, and is composed of eigenvectors corresponding to m maximum eigenvalues in a variance covariance matrix S, and the column vector of P is a unit vector. Wherein X, P and T are used as intermediate variables in the calculation process.
3. System fault diagnosis model
The method aims at a technical area factory building air conditioner (including a refrigeration station), an FS area fairing air conditioner, a large closed air conditioner and a refrigeration station research diagnosis evaluation algorithm model. The method comprises the following specific steps:
3.1 PCA model for process monitoring
The PCA model for process monitoring is as follows:
T＝XP
obviously, since P consists of only the first m vectors in the principal component model, there must be a loss of information when the scoring matrix T is transformed back to the original space.
E is called the residual matrix.
Thus, X is decomposed into projection space
And E of residual space:
in the projection space, the basic principle of fault detection is to observe whether the original data obeys a certain statistical rule in the projection space, here, generally, the change situation of the distance between the data and the central point in the projection space is observed, so that the following statistics is constructed:
wherein, t h ＝P T x＝[t 1 t 2 …t m ] T Is the coordinate of the sample data vector x in projection space, or a score vector, λ h (h =1,2, \8230;, m) is the eigenvector corresponding to the first m largest eigenvalues of the data matrix X, i.e. the covariance matrix.
T 2 The statistical indicator visually reflects the statistical distance between the sample point and the origin in the projection space, and the weighted value is the variance of each main shaft. Obviously, T 2 The statistic is a multivariate index composed of m principal component scores, and T is monitored 2 The change of the statistic can realize the simultaneous monitoring of a plurality of main components, and further can judge whether the running state of the whole process is abnormal.
T 2 Critical value of statistic, T when process data obeys normal distribution 2 The statistics obey an F distribution, which is adjacent toThe cutoff value can be calculated as follows:
wherein, alpha is the significance level of F test, and is usually 0.01,0.05 and 0.1; m is the number of principal components, and n is the number of data matrix samples.
Similarly, in the residual space, a statistic Q reflecting the change of the residual, also called the sum of squared errors index SPE, is constructed, which is the distance of the measured value from the principal component model, and is defined as follows
Wherein, the first and the second end of the pipe are connected with each other,
is a sample vector reconstructed from the score vector, x j Represents a process variable at j, <' >>
Representing the reconstructed process variable at j.
The critical value of SPE is
Wherein the content of the first and second substances,
c α representing the set of alpha.
When the system is in a normal state, the PCA model established by the process data collected under the normal working condition can well describe the correlation among the current process variable measured values, and can obtain T under the normal state 2 And an SPE index. When an anomaly occurs in the system, the correlation between process variables may deviate from the normal correlation structure, resulting in T 2 And SPE index biasLarge, when exceeding the corresponding control limit or critical value, it can be found in time.
When the multivariate statistical index T 2 And when the SPE index exceeds the control line, the system can be represented only in the abnormal state, but the specific abnormal process variable cannot be pointed out. Fault identification, i.e. to determine which variable changes cause gross variation, requires that each process variable x be identified j (j =,2, \8230;, m) pair statistic T 2 The contribution of the change. From the above analysis, T 2 Numerically dependent score vector t i And t is i And is the observation vector x i Coordinates in projection space. According to the definition of principal component, sample (x) 1 ，x 2 ，…，x p ) The value t of the h-th principal component of (1) h Comprises the following steps:
wherein
Is a load vector in the h direction (the h-th principal component coefficient vector). Thus, the process variable x j For the h main component t h Is a single variable x j Multiplying by the corresponding principal component coefficient p jh In the principal component t h In a ratio of
Since for one sample, T 2 Is fixed, so that the principal component t h For T 2 Can be simply considered as
Thus, x j By a principal component t h For T 2 The contribution of (A) is as follows:
then, x j For T 2 Should be x j By all principal components t h (h =1,2, \8230;, m), for T 2 Sum of contributions of (c):
e.g., to calculate a process variable x 1 For T 2 Contribution rate of (2)
The contribution rate of the process variables in the residual error space is simple and intuitive, and the residual error contribution rate of each variable is equal to the residual error corresponding to each variable:
3.2 LSTM-based diagnostic evaluation model
And (3) selecting the time sequence data of the guarantee parameters (temperature and humidity) and the process data of the LSTM (long-short term memory network) training air-conditioning system to generate a time sequence prediction model of the guarantee parameters and the process data. And selecting a proper prediction step length according to the precision of the time sequence prediction model, wherein the precision of the model is assumed to reach 95%, and the prediction step length is 15min. And obtaining the predicted system performance and risk data information by using the predicted guarantee parameters and process data information and combining a system performance and risk data calculation method, thereby realizing the early warning capability of the air-conditioning system. The guarantee parameters refer to basic parameter standards for guaranteeing production tasks, and the process data refer to real measuring point data acquired in the production process.
The detailed design steps are shown in fig. 2, and specifically include:
the method comprises the following steps: selecting historical data information with non-fault scenes and typical fault scenes, and using 80% of the historical data information as training of a prediction model and 20% as verification of the prediction model.
Step two: and carrying out standardization processing on the historical data information.
Step three: and (3) taking the guarantee parameters, namely the temperature, the humidity and process data, as the input and the output of the LSTM, inputting the time sequence data 30min before selection, outputting the time sequence data 15min after selection, setting the model precision to 95%, and enabling the precision to meet the set requirement through continuous model training.
Step four: the data size of 20% is selected for verification of the prediction model. And when the verified accuracy reaches 95%, the prediction model meets the requirement. Otherwise, returning to the third step to continue training the prediction model until the verification reaches the set requirement.
Step five: and collecting real-time guarantee parameters and process data, and respectively predicting the guarantee parameters and the process data in the next 15 minutes by using the trained prediction model.
4. Knowledge base model
And establishing an air conditioning system knowledge base, and embedding typical fault types. In addition, national, industry codes, base work instructions, emergency treatment protocols, and the like should be included. When creating the knowledge base, the typical fault base and the case base, labels can be printed on the knowledge base, the case base and the case base, such as the knowledge base of the chilled water pump, the case base of the cooling tower, the fault base of the chilled water pump and the like. Taking a typical fault library as an example, the information of the created fault library can be checked, a fuzzy search mode of directly inputting characters is supported, and display according to label classification is also supported. Clicking on the "view details" of a particular failed library list may further view or edit the contents of the failed library.
In a second aspect, an embodiment of the present invention provides an air conditioner intelligent monitoring system based on multiple dimensions, which mainly includes three functional modules, namely a system state monitoring module, a diagnosis evaluation analysis module, and a task assistant decision module, for realizing the multi-dimensional intelligent monitoring of an air conditioner system.
A system state monitoring module: by collecting data of air-conditioning equipment of each factory building, situation perception of the state of the air-conditioning equipment and the system process is realized, and real-time interpretation of the state of the air-conditioning equipment, key parameters and the system state is realized; the data visualization interaction of the air conditioner data association analysis, the trend analysis and the video abnormity analysis is realized; and the intelligent alarm capacity of alarm analysis, alarm statistics and alarm configuration is realized.
A diagnostic evaluation analysis module: the intelligent air conditioner algorithm model (namely the system performance evaluation model, the system data driving model, the system fault diagnosis model and the knowledge base model) is used for realizing the performance evaluation of key equipment and system performance, and the representation, query, retrieval and reasoning of knowledge are realized through the management of the knowledge base/fault base, so that the fault detection and positioning capability is provided, and the troubleshooting and tracking are assisted.
A task assistant decision module: and tracking and analyzing the technical state change through an evaluation and analysis result obtained by the diagnosis evaluation and analysis module, and realizing equipment fault risk evaluation, technical state change influence evaluation and auxiliary judgment of risk level.
Parts of the invention not described in detail are well known to the person skilled in the art.
The particular embodiments of the present invention disclosed above are illustrative only and are not intended to be limiting, since various alternatives, modifications, and variations will be apparent to those skilled in the art without departing from the spirit and scope of the invention. The invention should not be limited to the disclosure of the embodiments in the present specification, but the scope of the invention is defined by the appended claims.
Claims (10)
1. An air conditioner intelligent monitoring method based on multiple dimensions is characterized by comprising the following steps:
collecting air conditioner data through sensing equipment;
uploading the collected air conditioner data to a model modeling tool, and establishing a system performance evaluation model, a system data driving model, a system fault diagnosis model and a knowledge base model;
and obtaining the diagnosis and evaluation result of the air conditioner performance according to the system performance evaluation model, the system data driving model, the system fault diagnosis model and the knowledge base model.
2. The method according to claim 1, wherein the system performance evaluation model analyzes and evaluates the performance of the air conditioning system in real time according to an established evaluation index system, and the evaluation index in the evaluation index system comprises:
1) Control accuracy range: the air conditioning unit can maintain the fluctuation range of the temperature and the relative humidity while directly providing the processing air to the closed space, the room or the area;
2) Refrigerating capacity: the sum of the heat removed by the air conditioning unit from the enclosed space, room or area per unit time;
3) Cooling power consumption: the total power consumed by the air conditioning unit under the specified refrigeration test working condition;
4) Coefficient of performance: the ratio of the refrigerating capacity Q and the refrigerating consumed power P of the air conditioning unit expressed by the same unit;
5) The total load IQ of the comprehensive performance;
6) The overall performance coefficient ICOP;
7) Precision curve: after the indoor state is balanced for 1h, testing the fluctuation of the temperature and the humidity of the measuring points arranged in the room, and drawing a precision curve graph;
8) Refrigeration performance: the method comprises the steps of refrigeration temperature and humidity precision, refrigeration temperature and humidity stability and refrigeration temperature and humidity efficiency.
3. The method of claim 2, wherein the aggregate performance total load IQ is calculated as:
IQ＝|Q r0 |+|Q s0 |+∫ 0 τ |q re |dt+∫ 0 τ |q se |dt
in the formula:
Q r0 -the initial heat load of the indoor ICOP air handler is increased or decreased to the room to change the room state from the set working condition to the starting working condition, unitIs coke;
Q s0 the indoor ICOP air processor increases or decreases the initial wet load to the room, so that the room state is changed from the set working condition to the starting working condition, and the unit is coke;
q re after the air conditioner is started, the indoor ICOP air processor increases or reduces the heat load to the room in watt in unit time;
q se after the air conditioner is started, the indoor ICOP air processor increases or reduces the moisture load to the room in units of watts;
the time consumed by the unit from start to stop or the time in seconds during which the components operate individually.
4. The method according to claim 1, wherein the system performance evaluation model is an analysis evaluation model of the air conditioning system established by comprehensively considering economy, energy conservation and environmental protection by using a grey correlation analysis method.
5. The method of claim 1, wherein the system data driving model learns the intrinsic relationship between variables and parameters by means of a large amount of historical data of the air conditioning system under various operating conditions, including normal data and fault data, and further identifies fault conditions in the new data and isolates fault sources by a mathematical model established through a machine learning process.
6. The method of claim 1, wherein the system data-driven model uses Principal Component Analysis (PCA) to diagnose faults in the process by using correlations between process variables, decomposes a multivariate sample space into a projection subspace with lower dimensions formed by the principal component variables and a corresponding residual subspace by using a multivariate projection method according to historical data of the process variables, and respectively constructs statistics capable of reflecting spatial changes in the two subspaces, then respectively projects the original observation vectors to the two subspaces, and calculates corresponding statistic indexes for process monitoring.
7. The method as claimed in claim 1, wherein the system fault diagnosis model adopts LSTM to train the time series data of the guarantee parameters and the process data of the air conditioning system, to generate a time series prediction model of the guarantee parameters and the process data, selects an appropriate prediction step length according to the accuracy of the time series prediction model, and obtains predicted performance and risk data information of the air conditioning system by using the guarantee parameters and the process data information obtained by prediction, thereby realizing the early warning capability of the air conditioning system.
8. The method of claim 7, wherein the system fault diagnosis model predicts the health parameters and process data using the steps of:
the method comprises the following steps: selecting historical data information with non-fault scenes and typical fault scenes, and using 80% of the historical data information as training of a prediction model and 20% as verification of the prediction model;
step two: carrying out standardization processing on historical data information;
step three: taking the guarantee parameters and the process data as the input and the output of a prediction model LSTM, inputting the time sequence data 30min before selection, outputting the time sequence data 15min after selection, setting the precision of the model to be 95%, and continuously training the model to ensure that the precision of the model meets the set requirement;
step four: selecting 20% of data volume to verify the prediction model, wherein the prediction model meets the requirement when the verification precision reaches 95%, otherwise, returning to the third step and continuing to train the prediction model until the verification reaches the set requirement;
step five: and collecting real-time guarantee parameters and process data, and respectively predicting the guarantee parameters and the process data in the next 15 minutes by using the trained prediction model.
9. The method of claim 1, wherein the knowledge base model comprises an air conditioning system knowledge base, embeds typical fault types, and includes country, industry code, and base work instructions, emergency treatment plans; the method comprises the following steps of (1) inputting labels for a knowledge base, a typical fault base and a case base when the knowledge base, the typical fault base and the case base are created, wherein the labels comprise the knowledge base of a chilled water pump, the case base of a cooling tower and the fault base of the chilled water pump; in a typical fault library, the information of the created fault library can be checked, a fuzzy search mode of directly inputting characters is supported, and display according to label classification is also supported.
10. The utility model provides an air conditioner intelligent monitoring system based on multidimension which characterized in that includes:
the system state monitoring module is used for realizing the situation perception of the state of the air-conditioning equipment and the system process and the real-time interpretation of the state of the air-conditioning equipment, key parameters and the system state through the data acquisition of the air-conditioning equipment of each factory building;
the diagnosis evaluation analysis module is used for realizing the evaluation of the performance of the air conditioner by utilizing the system performance evaluation model, the system data driving model, the system fault diagnosis model and the knowledge base model, realizing the representation, query, retrieval and reasoning of knowledge through the management of the knowledge base and the fault base, providing fault detection and positioning capability and assisting in troubleshooting and tracking;
and the task auxiliary decision-making module is used for realizing the fault risk assessment, the technical state change influence assessment and the auxiliary judgment of the risk level of the air conditioning equipment according to the assessment and analysis result obtained by the diagnosis assessment and analysis module.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211657413.8A CN115899964A (en) | 2022-12-22 | 2022-12-22 | Multidimensional-based intelligent air conditioner monitoring method and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211657413.8A CN115899964A (en) | 2022-12-22 | 2022-12-22 | Multidimensional-based intelligent air conditioner monitoring method and system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115899964A true CN115899964A (en) | 2023-04-04 |
Family
ID=86481756
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211657413.8A Pending CN115899964A (en) | 2022-12-22 | 2022-12-22 | Multidimensional-based intelligent air conditioner monitoring method and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115899964A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116857763A (en) * | 2023-09-04 | 2023-10-10 | 江苏瀚艺能源科技有限公司 | Intelligent adjusting system for commercial air conditioner refrigerating equipment |
| CN116975502A (en) * | 2023-09-21 | 2023-10-31 | 天津津轨汇海科技发展有限公司 | Energy efficiency monitoring management system for electromechanical equipment of subway train air conditioning system |
| CN117073129A (en) * | 2023-10-17 | 2023-11-17 | 山东意林建筑规划设计研究院有限公司 | Fault diagnosis method and system for heating ventilation air conditioning system |
| CN117145753A (en) * | 2023-10-31 | 2023-12-01 | 意朗智能科技(南通)有限公司 | Working quality testing method and device for air compressor |
| CN117677165A (en) * | 2024-01-31 | 2024-03-08 | 广东览讯科技开发有限公司 | Special air conditioning system of high-efficient computer lab and cooling tower |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110762768A (en) * | 2019-10-30 | 2020-02-07 | 南京亚派软件技术有限公司 | Energy efficiency ratio prediction method and device for refrigeration host of central air-conditioning system |
| CN111412579A (en) * | 2020-03-26 | 2020-07-14 | 上海建工四建集团有限公司 | Air conditioning unit fault type diagnosis method and system based on big data |
| CN112303810A (en) * | 2019-08-01 | 2021-02-02 | 山东朗进科技股份有限公司 | Air conditioner health prediction method based on machine learning |
| US20210080138A1 (en) * | 2017-11-16 | 2021-03-18 | Hitachi, Ltd. | Performance diagnosis device and performance diagnosis method for air conditioner |
| CN114353261A (en) * | 2021-12-13 | 2022-04-15 | 广东电网有限责任公司 | Air conditioning unit fault analysis method and device, terminal device and storage medium |
| CN115496325A (en) * | 2022-08-05 | 2022-12-20 | 广东机场白云信息科技有限公司 | Method, equipment and medium for evaluating energy efficiency of air conditioner cold source system |
-
2022
- 2022-12-22 CN CN202211657413.8A patent/CN115899964A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20210080138A1 (en) * | 2017-11-16 | 2021-03-18 | Hitachi, Ltd. | Performance diagnosis device and performance diagnosis method for air conditioner |
| CN112303810A (en) * | 2019-08-01 | 2021-02-02 | 山东朗进科技股份有限公司 | Air conditioner health prediction method based on machine learning |
| CN110762768A (en) * | 2019-10-30 | 2020-02-07 | 南京亚派软件技术有限公司 | Energy efficiency ratio prediction method and device for refrigeration host of central air-conditioning system |
| CN111412579A (en) * | 2020-03-26 | 2020-07-14 | 上海建工四建集团有限公司 | Air conditioning unit fault type diagnosis method and system based on big data |
| CN114353261A (en) * | 2021-12-13 | 2022-04-15 | 广东电网有限责任公司 | Air conditioning unit fault analysis method and device, terminal device and storage medium |
| CN115496325A (en) * | 2022-08-05 | 2022-12-20 | 广东机场白云信息科技有限公司 | Method, equipment and medium for evaluating energy efficiency of air conditioner cold source system |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116857763A (en) * | 2023-09-04 | 2023-10-10 | 江苏瀚艺能源科技有限公司 | Intelligent adjusting system for commercial air conditioner refrigerating equipment |
| CN116857763B (en) * | 2023-09-04 | 2023-12-26 | 江苏瀚艺能源科技有限公司 | Intelligent adjusting system for commercial air conditioner refrigerating equipment |
| CN116975502A (en) * | 2023-09-21 | 2023-10-31 | 天津津轨汇海科技发展有限公司 | Energy efficiency monitoring management system for electromechanical equipment of subway train air conditioning system |
| CN116975502B (en) * | 2023-09-21 | 2023-12-19 | 天津津轨汇海科技发展有限公司 | Energy efficiency monitoring management system for electromechanical equipment of subway train air conditioning system |
| CN117073129A (en) * | 2023-10-17 | 2023-11-17 | 山东意林建筑规划设计研究院有限公司 | Fault diagnosis method and system for heating ventilation air conditioning system |
| CN117073129B (en) * | 2023-10-17 | 2024-01-26 | 山东意林建筑规划设计研究院有限公司 | Fault diagnosis method and system for heating ventilation air conditioning system |
| CN117145753A (en) * | 2023-10-31 | 2023-12-01 | 意朗智能科技(南通)有限公司 | Working quality testing method and device for air compressor |
| CN117145753B (en) * | 2023-10-31 | 2023-12-22 | 意朗智能科技(南通)有限公司 | Working quality testing method and device for air compressor |
| CN117677165A (en) * | 2024-01-31 | 2024-03-08 | 广东览讯科技开发有限公司 | Special air conditioning system of high-efficient computer lab and cooling tower |
| CN117677165B (en) * | 2024-01-31 | 2024-04-16 | 广东览讯科技开发有限公司 | Special air conditioning system of high-efficient computer lab and cooling tower |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115899964A (en) | Multidimensional-based intelligent air conditioner monitoring method and system | |
| Zhang et al. | A comparison and evaluation of key performance indicator-based multivariate statistics process monitoring approaches | |
| Li et al. | Application of pattern matching method for detecting faults in air handling unit system | |
| Said et al. | Machine learning technique for data-driven fault detection of nonlinear processes | |
| Maurya et al. | Fault diagnosis by qualitative trend analysis of the principal components | |
| Jiang et al. | Weighted kernel principal component analysis based on probability density estimation and moving window and its application in nonlinear chemical process monitoring | |
| Lennox et al. | Application of multivariate statistical process control to batch operations | |
| Ma et al. | A novel hierarchical detection and isolation framework for quality-related multiple faults in large-scale processes | |
| Dobos et al. | On-line detection of homogeneous operation ranges by dynamic principal component analysis based time-series segmentation | |
| Yang et al. | A hybrid model-based fault detection strategy for air handling unit sensors | |
| Chen et al. | A whole building fault detection using weather based pattern matching and feature based PCA method | |
| Li et al. | Application of Big Data Technology in Evaluation of Operating State of High-Pressure Hydrogenation Heat Exchanger | |
| CN110942258B (en) | Performance-driven industrial process anomaly monitoring method | |
| CN117390536A (en) | Operation and maintenance management method and system based on artificial intelligence | |
| Yang et al. | Similarity-based information fusion grey model for remaining useful life prediction of aircraft engines | |
| CN110751217A (en) | Equipment energy consumption ratio early warning analysis method based on principal component analysis | |
| Chen et al. | Whole building system fault detection based on weather pattern matching and PCA method | |
| JP5155740B2 (en) | MONITORING CONDITION DETERMINING DEVICE, MONITORING CONDITION DETERMINING METHOD, AND MONITORING DIAGNOSIS DEVICE | |
| Huang et al. | Static and dynamic joint analysis for operation condition division of industrial process with incremental learning | |
| CN116011982A (en) | Online monitoring method and system for breakage of grinding roller of coal mill | |
| CN116204825A (en) | Production line equipment fault detection method based on data driving | |
| Humberstone et al. | Differentiating between expanded and fault conditions using principal component analysis | |
| Li et al. | Combining canonical variate analysis, probability approach and support vector regression for failure time prediction | |
| Geng et al. | Fault detection of large-scale process control system with higher-order statistical and interpretative structural model | |
| Cempel | Decomposition of the symptom observation matrix and grey forecasting in vibration condition monitoring of machines |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |