In modern control systems, due to the increasing complexity of the system and the ever-expanding scale, the system often faces unpredictable changes. In order to design a reliable fault-tolerant control system, or to maintain the performance of a complex system at a normal level during operation of the equipment, it is necessary to properly detect and diagnose these changes and take appropriate measures to reconfigure the system. Therefore, in order to ensure the safety and reliability of the control system, fault diagnosis and fault-tolerant control technology based on system model has become a hot research direction in the field of automatic control. In the system operation, with the accumulation of experience, many quantitative process knowledge, such as steady state gain between various process variables, will be obtained. However, traditional qualitative model-based diagnostic systems do not have the ability to utilize these quantitative process knowledge. . To this end, the combination of quantitative and qualitative knowledge in fault diagnosis has been raised. Previous studies include: directed graph + fault tree, order of magnitude analysis, fuzzy set modeling, interval number analysis, and so on. The diagnostic algorithm proposed here is a combination of qualitative and quantitative diagnosis methods based on constraint propagation. It has certain self-learning ability for fault diagnosis, and initially solves the problem of combined explosion and diagnostic results redundancy when using qualitative knowledge diagnosis. . 1QSIM fault diagnosis technology based on qualitative simulation theory Based on qualitative simulation theory (QSIM) fault diagnosis technology, the theoretical basis is the qualitative simulation theory based on qualitative differential equation (QDE) proposed by Kuipers in 1986. In this theory, the system model consists of parameters that represent system variables and constraints that describe the relationship between parameters. The qualitative state of the system is a combination of qualitative states of all variables of the system, and the qualitative behavior of the system is the sequence of qualitative states of the system. Qualitative simulation starts from the initial state of the system, generates all possible successor states according to the preset state transition table, checks the state consistency with constraints, filters the inconsistent state, reconstructs a new current state set, and selects one from it. The new current state, repeat the above process, and finally get all the behavior of the system. In 1987, Kuipers proposed a QSIM-based diagnostic technology, using a fault-based diagnostic strategy, using the QSIM method to build a system fault model, and simulating the system behavior to obtain a predicted system state. The observed fault behavior is then compared to these predicted behaviors. If the predicted behavior is consistent with the observed fault behavior, the system has failed. The fault model with the predicted behavior and the fault behavior becomes the model of the current system, and the type and cause of the fault are further diagnosed according to the prior knowledge when the model is established. This gives full play to QSIM's powerful features based on deep knowledge modeling and reasoning. This process is called a hypothetical D modeling D simulation D matching loop. The above method is suitable for diagnosing all systems with clear faults. If there is an unknown fault, the fault diagnosis cannot be accurately performed due to the lack of a corresponding fault model, which is the limitation of the method itself. Moreover, since the qualitative description is not accurate enough with respect to the quantitative description, it may occur that different failure causes correspond to the same failure behavior or one failure corresponds to multiple diagnostic results containing redundancy. Therefore, the introduction of quantitative information in qualitative fault diagnosis has become a research focus of fault diagnosis. 2 The establishment of a combination of qualitative and quantitative models 2.1 The difference limit of the difference detection setting parameters is ε. Calculate the value of the parameter at a certain continuous time, and compare the two adjacent time points T i and T i+1 if the ratio of the difference between the parameters V i and V i+1 and V i is ε ≥ + i V 1 (1) determines that this parameter V is in a new state at time T i+1 . The normal range of the variables is (a, b), where a : the condensing pressure is 1519 kgf/cm2 absolute, when the measured value is 21 kgf/cm2 absolute. The normal variable range δ that is exceeded is 19 15 21 () 2 0.235 0.25 ((19 15) 2δ+= < +=) The threshold is set to be normal. 2.2 Transfer of quantitative constraints In the fault diagnosis, the measurement result of a certain parameter is usually expressed by a numerical interval, where the constraint relationship between the parameters is transmitted by the arithmetic operation rule of the interval algebra. 2.3 Diagnostic process In the fault diagnosis system combining qualitative and quantitative, Reiter's fault diagnosis theory is applied. Under the condition that the system is offline, the system knows the fault model according to the structural knowledge and experience knowledge of the system. After the fault model is established. Real-time observation and diagnosis of the diagnostic object using the fault diagnosis system as shown. The process is as follows: (1) first put the initial state into the current state; (2) predict the state of the system through a hybrid system model of qualitative and quantitative. (3) Calculate the difference range between the observed value and the predicted value of each variable in the current state, that is, the fitting degree of the predictive variable and the system variable, and distinguish the system from the normal fluctuation range and the occurrence of the system fault according to the corresponding defined threshold. (4) Compare the predicted state and actual state of the system. If âˆÎ¼x=1, the output is normal "system state is normal"; if âˆÎ¼x=0, then go to (5). (5) Matching the fault variable set V={Xã„μx=0} with the fault set M={F xã„X i1∧X i2...∧X ik...∧X in =1=>F i}. The IF fault F i exists, F i ∈ M, Then outputs "Fault F i" Else to (6). (6) According to the fault variable set in (5), combined with the system's hybrid diagnosis, according to the system constraint relationship, derive the new fault model, and join the original fault model, and jump to (5). 3 Diagnostic Modeling and Simulation of Refrigeration System 3.1 System Description Based on Kuipers' QSIM algorithm for fault diagnosis, a diagnostic method combining qualitative and quantitative is proposed and applied to fault diagnosis in refrigeration system. In order to simplify the system, we use the system consisting of several main components of the refrigeration system: compressor, condenser, expansion valve and evaporator to carry out simulation diagnosis. The system structure is as shown. 3.2 System parameters The refrigeration system is mainly affected by several factors: the heat that the refrigeration system takes away from the refrigerator; the heat loss of the room; the outdoor air temperature, that is, the system condensation temperature; the working state of the condensation fan. The definition of each parameter is as shown. The black body part is a measurable variable, which is the basis for diagnosis and reasoning of the system. 3.3 Constraint relations According to the laws of thermodynamics, the constraints of variables can be obtained as follows: (1) the condensation temperature is proportional to the condensation pressure M + (P c, T c); (2) the evaporation temperature is proportional to the evaporation pressure M + ( P e, T e); (3) the superheat degree T g is proportional to the opening degree α of the valve M + (α, T g); (4) the temperature increase of the compressor and the pressure increase constraint relationship M + (P z , T z); According to the system structure and the function of the components to establish the constraint relationship is as follows: (1) compressor pressure increase constraint ADD (P j, P z, P w); (2) compressor temperature increase constraint relationship ADD (T j, T z, T w); (3) expansion valve pressure reduction constraint relationship ADD (P e, P s, P c); (4) superheat degree reflects the difference between the evaporator suction temperature and the evaporation temperature ADD (T e , T g, T j); (5) relationship between outdoor air temperature T a and condensation temperature T c MULT (σ, T c, T a), where σ reflects the working condition of the condenser; (6) room temperature T r and evaporation The relationship between the T e of the device reflects the working condition of the blower MULT(β, T e, T r); (7) the relationship between the opening and the pressure drop of the valve M + 3.4 System Prior Failure Model For the illustrated refrigeration system, three system failure models were established based on the structural knowledge of the system. The values ​​in the table are the predicted values ​​of each variable based on the system model. Among them: F 1 is the compressor failure, the pressure increase Pz is “-†or decrease, F 2 is the pipeline blockage of the refrigeration system; F 3 is the shortage of refrigerant in the system, and the high pressure and low pressure are reduced. 4 Diagnostic simulation Based on the above-mentioned qualitative and quantitative diagnostic methods, taking the actual state of an air-conditioning refrigeration system as an example, the comprehensive diagnosis theory is applied to verify the simulation. 4.1 System status For R22 system, the evaporation pressure is 57kgf/cm2, the evaporation temperature is 310°C, the temperature is 35°C, the condensing pressure is 1519kgf/cm2 absolute pressure, the condensation temperature is 4050°C, and the temperature is 515°C. . The normal indoor temperature, the refrigeration system works under such conditions, can get normal cooling performance. For an air conditioning refrigeration system, the fault is as follows: the compressor does not stop working continuously, the cooling effect is not good, and the compressor inlet is frosted. At time 04, the observations of the air conditioning system are as shown. 4.2 Quantitative information processing The data of the parameters in time 04 is averaged, and the initial state of the system is sorted according to the method of difference detection, and the initial state of the system is set to the current state. The table data processing determines the initial state of the system according to equation (1) as shown. 4.3 State Prediction and Diagnosis According to the mixed diagnosis model, the system behavior is inferred and simulated. According to the operation rules of interval algebra, the system parameters of the quantitative information are transmitted, and the measurement information of each parameter is propagated according to the qualitative constraints of the system. The variable, in its corresponding value field, calculates the numerical range of the variable according to the rules, and applies the difference detection to calculate the qualitative state of the parameter. According to the relevant constraints, check the constraint satisfaction between the variable and the assigned variable, that is, check whether the parameter value is "default". If it is satisfied, continue to the next variable. Otherwise, select other values ​​in the value field of the variable. If it finds that there are no other values ​​in its value field, then the dead end occurs, and the prediction state is predicted by the system. c =<15.2,16.4>kgf/cm 2 P e =<2.0,2.9>kgf/cm 2 according to the constraint: M +(P c, P w), ADD (P e, P s, P c), ADD( P j, P z, P w) and M + (P e, P j), the applied constraint transfer rules are available: P s =<12.3, 14.4>, P z=<12.3, 14.4>. Condenser and evaporation The pressure drop P s between the devices is obviously higher than the normal value, and although the pressure increase of the compressor increases, but the normal operation is within a certain range, the actual fault state of the system is as shown. Comparing the predicted state with the known system's a priori fault model and not matching the known prior fault models F 1, F 2, F 3, the new fault diagnosis model is derived based on the system's fault symptoms: F 4, that is, the low pressure section of the evaporator of the refrigeration system is low, the evaporation temperature is also low, the temperature T j at the inlet of the compressor is low and the pressure of the condensation section is basically normal. The failure of the system is poor heat dissipation of the evaporator, and the room temperature T r is high, then evaporation The heat exchange of the device is not sufficient. The boot check found that the filter screen at the air supply outlet was blocked, and the fault was eliminated after cleaning. The fault model F 4 is added to the prior diagnostic system and used as a knowledge base for the diagnostic system for system diagnostics and detection. 4.4 Simulation evaluation For the above system symptoms, the results of manual disassembling diagnosis are consistent with the results of simulation diagnosis, verifying the correctness of the concept of fault diagnosis combined with qualitative and quantitative information, and further expanding the simulation for fault diagnosis and detection. Application range. 5 Conclusions The combination of qualitative and quantitative diagnostic methods not only preserves the accuracy of quantitative simulation and the flexibility of qualitative simulation, but also overcomes the problem of quantitative information modeling and qualitative simulation of redundant information, so that it is not only very Strong system description ability, and greatly reduce the uncertainty of qualitative fault diagnosis, can quickly and accurately reflect the possible faulty components in the system, is a fault diagnosis theory with good development prospects. Barudan Machinery Fittings,Embroidery Machine Part,High Speed Embroidery Machine Knitting Machine,Textile Machinery Parts Co., Ltd. , http://www.computer-embroidery-machine.com
Diagnostic Measures of QSIM Problem Based on Qualitative Simulation Theory