WO2011006344A1 - 砂尘环境试验系统的温度调节装置及智能温度控制方法 - Google Patents

砂尘环境试验系统的温度调节装置及智能温度控制方法 Download PDF

Info

Publication number
WO2011006344A1
WO2011006344A1 PCT/CN2010/000720 CN2010000720W WO2011006344A1 WO 2011006344 A1 WO2011006344 A1 WO 2011006344A1 CN 2010000720 W CN2010000720 W CN 2010000720W WO 2011006344 A1 WO2011006344 A1 WO 2011006344A1
Authority
WO
WIPO (PCT)
Prior art keywords
control
temperature
layer
neural network
input
Prior art date
Application number
PCT/CN2010/000720
Other languages
English (en)
French (fr)
Inventor
王浚
张华�
李运泽
明章鹏
朱克勇
马志宏
张利珍
赵晨
Original Assignee
北京航空航天大学
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Priority claimed from CN2009100892898A external-priority patent/CN101639397B/zh
Priority claimed from CN201010034360.5A external-priority patent/CN102129259B/zh
Application filed by 北京航空航天大学 filed Critical 北京航空航天大学
Priority to US13/318,327 priority Critical patent/US20120048953A1/en
Publication of WO2011006344A1 publication Critical patent/WO2011006344A1/zh

Links

Classifications

    • GPHYSICS
    • G01MEASURING; TESTING
    • G01MTESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M9/00Aerodynamic testing; Arrangements in or on wind tunnels
    • G01M9/02Wind tunnels
    • G01M9/04Details
    • GPHYSICS
    • G01MEASURING; TESTING
    • G01MTESTING STATIC OR DYNAMIC BALANCE OF MACHINES OR STRUCTURES; TESTING OF STRUCTURES OR APPARATUS, NOT OTHERWISE PROVIDED FOR
    • G01M99/00Subject matter not provided for in other groups of this subclass
    • G01M99/002Thermal testing

Definitions

  • the invention relates to a temperature adjusting device and an intelligent temperature control method for a sand dust environment testing system, and belongs to the technical field of sand dust environment testing.
  • the dust environment is an important environmental factor that causes many engineering or military weapons and equipment to fail.
  • the main types of damage are: erosion, wear, corrosion and penetration.
  • Blowing and sand blowing tests are an important means of testing the environmental adaptability and environmental reliability of vehicles, aircraft, electrical equipment, military equipment in deserts, dry areas and windy weather conditions.
  • the existing sand dust environment test system uses two large and one small air conditioners to complete the temperature control.
  • the Chinese invention patent: ZL200510000070. 8 discloses a system, wherein the system includes: a circulation air duct 1 , large air conditioner 2, small air conditioner 3, air cooler finned tube 4, finned electric heater 5, variable frequency speed control fan 6, large air conditioning inlet passage 7, small air conditioning inlet passage 8, air conditioning return passage 9,
  • the chiller water source 10 the large air-conditioning compressed air blow-off pipe 11, and the small air-conditioning compressed air blow-off pipe 12.
  • the system has the following drawbacks:
  • the circulation air duct 1 adopts the external thermal insulation structure, and the heat capacity is very large.
  • the heat transfer area of the air cooler 4 is limited, it takes a long time to cool, which seriously affects The utilization efficiency of the test device system.
  • the temperature control of large-scale turbulent sand dust environment test wind tunnels in China is mainly equipped with two different temperature control devices: air cooler and electric heater. And because of the low-speed dust blowing and high-speed dust blowing test, the two different temperature control measures, heating and cooling, are adopted to increase the temperature difference range and increase the control difficulty.
  • the fan speed changes, the auxiliary air flow rate and temperature change, and the change of the ambient temperature will cause temperature disturbance in the air duct.
  • the temperature disturbance can be controlled, but
  • the air conditioning unit requires a special air conditioning duct and heat exchange surface, and brings additional operating energy consumption.
  • the presence of the air conditioning duct also affects the flow quality of the airflow in the circulating duct.
  • Another object of the present invention is to provide an intelligent temperature control method for a sand dust environment test system to overcome the deficiencies of the prior art in terms of high precision, high stability, and coordinated control.
  • the invention relates to a temperature adjusting device for a sand dust environment testing system
  • the sand dust environment testing system mainly comprises: a circulating air channel, which is an airtight channel with irregular structure, and is used for providing a place for blowing sand and dust blowing environment;
  • the air duct includes a second contraction section, a test section and a separation section, and the like.
  • a main fan for driving the flow of the airflow in the circulation duct
  • a U-type separator disposed in the separation section for separating and recovering the sand dust after the test; and
  • the guide vanes are arranged in four groups at four corners of the circulation duct;
  • the utility model is characterized in that: the temperature regulating device comprises:
  • the four sets of guide vanes are respectively provided with heat exchange working medium through holes in the three sets of guide vanes; and another set of guide vanes are provided with heating means;
  • the temperature adjustment device further comprises: a circulating cooling water source;
  • the circulating cooling water source, the chiller and the electric boiler are respectively connected to the guide vane and the u-type separator having the heat exchange function through the pipeline; and the valves for adjusting the flow rate of the hot and cold working fluid are arranged on each pipeline.
  • the heating device in the guide vane is an electric heater, and the electric heater is sequentially coated with an insulating layer and a wear-resistant protective layer.
  • the invention relates to an intelligent temperature control method for a sand dust environment test system, which is realized by a neural network based PI intelligent temperature control system, the intelligent temperature control system comprising:
  • Wind speed sensor installed in the circulation air duct
  • the neural network controller derives a coordinated control factor based on the wind speed measured from the wind speed sensor to determine which device is the primary control device and which device is the auxiliary control device to fully achieve the purpose of coordinated control.
  • the neural network controller specifically includes an input layer, an implicit layer and an output layer;
  • the PI controller is configured to receive a coordinated control factor from the neural network controller and a temperature value measured from the temperature sensor, and transmit the processed control amount to the temperature control device of the dust environment test system.
  • the PI controller comprises:
  • the intelligent temperature control system further includes: a first limiter for limiting and optimizing the first coarse control amount, thereby generating an optimized fine control quantity for the electric boiler return water control valve,
  • the invention relates to an intelligent temperature control method for a sand dust environment test system, which comprises the following steps:
  • Step 1 Establish a neural network system structure
  • the invention adopts a three-layer feedforward network with a single hidden layer; the transformation function of the hidden layer unit adopts positive and negative symmetry
  • the three-layer feedforward network Includes:
  • Output layer The output of the output layer corresponds to the coordinated control factor Z cv , which determines which device is the primary control device and which is the auxiliary control device.
  • Step 2 Mixed learning training of neural network parameters
  • Step 3 PI control
  • the coordinated control factor is input to the PI controller, and the PI controller further performs the operation and combination of the coordinated control factor and its parameters, and finally obtains the control variable for the controlled institution. Take control.
  • Step four limiting processing
  • the S-type function is applied to limit the processing to optimize the control variable.
  • the hybrid learning training of the neural network parameters is performed, and the determining module is performed in advance, and when the temperature difference is better than the predetermined threshold value, the offline dynamic temperature model learning training based on the BP learning algorithm is performed, and the weight is implemented. Offline global optimization optimizes the error and shortens the time required for online learning. Conversely, when within the defined value range, the genetic algorithm is used for online adjustment to optimize the weight.
  • a coordinated control factor is obtained; and one of the at least two temperature control mechanisms is determined as the main control mechanism according to the coordinated control factor, and at least two temperature control are performed.
  • the remaining control mechanisms in the mechanism are determined as auxiliary control mechanisms by which the coarse control amount is obtained based on the temperature value and the coordinated control factor.
  • the present invention discards two large air conditioning tanks, the influence of the air conditioning box ventilation circuit on the airflow quality in the circulating air duct is eliminated. Moreover, the structure is simpler and the space is smaller; 2) Since the invention does not have the air cooler and the electric heater in the air-conditioning box, the inconvenience caused by cleaning dust and the influence of dust accumulation on heat transfer are eliminated;
  • the invention adds a circulating hot water system on the basis of the conventional cooling system, so that the conventional cooling system becomes a hot and cold multi-purpose temperature control system, which improves the utilization rate of the equipment; 4 >
  • the invention is in the contraction section and contraction An electric heating sheet is respectively added to a set of guide vanes before the segment, which can further accelerate the heating speed of the airflow in the circulating air passage and improve the working efficiency of the system; 5) the present invention lays a heat exchange tube in the U-shaped separator, and Reconstructing the remaining three sets of guide vanes, providing through holes in
  • Figure 1 is a schematic view showing the connection of the sand dust environment test system in the prior art.
  • Figure 2 is a schematic view showing the aerodynamic layout structure of the circulating air passage of the present invention.
  • Figure 3 is a cross-sectional view of the circulation duct of the present invention.
  • Figure 4A is a view of the U-shaped separator shaft with heat exchange function of the present invention
  • Figure 4B is a cross-sectional view of a U-shaped separator having a heat exchange function of the present invention.
  • Figure 5A is an axial view of a guide vane having heat exchange function of the present invention
  • Figure 5B is a cross-sectional view of a guide vane having a heat exchange function of the present invention.
  • Figure 6 is a partial cross-sectional view of the circulation duct of the present invention.
  • Figure 7 is a cross-sectional view of a guide vane having a heating function of the present invention.
  • Figure 8 is a schematic view of the temperature regulating device of the dust environment test system of the present invention.
  • Figure 9 is a schematic illustration of a conventional temperature control system of the present invention.
  • Figure 10 is a general flow chart of the sand dust environment test wind tunnel of the present invention.
  • Figure 11 is a flow chart showing the control of the wind tunnel temperature intelligent control of the sand dust environment test of the present invention.
  • Fig. 12 is a schematic structural diagram of the wind tunnel temperature intelligent control based on the neural network PI control of the sand dust environment test of the present invention.
  • Figure 13 is a structural diagram of a neural network controller for intelligent control of wind tunnel temperature in a sand dust environment test according to the present invention.
  • Fig. 14 is a structural diagram of a neural network PI control based on a hybrid learning training algorithm for intelligent control of wind tunnel temperature in sand dust environment test of the present invention.
  • Fig. 15 is a general flow chart of the learning and training algorithm based on the neural network PI intelligent temperature control system of the sand dust environment test wind tunnel of the present invention.
  • FIG. 16 is a flow chart of an improved BP learning algorithm for intelligent control of wind tunnel temperature in a sand dust environment test according to the present invention.
  • Circulation duct 101 Test section 102. Diffusion section 103. Separation section 104. First corner section 105. First contraction section 106. Second corner section. 107. Square transition 108. Power section 109. Round transition 110. Variable section 111. Third corner section 112. Fourth corner section 113. Stabilization section 114. Second contraction section 1141. Electric heater 115. First guide vane 116. Second guide vane
  • Electric heater 119 Heat exchange working medium through hole 2. Large air conditioner 3. Small air conditioner 4. Surface cooler finned tube 5. Fin electric heater 6. Frequency control air conditioner ' 7. Large air conditioning inlet channel 8. Small air conditioning intake passage
  • Air-conditioning return air passage 10 Water source of chiller 11. Large air-conditioning dust blowing pipe 12. Small air-conditioning dust blowing pipe 13. Main fan 14. Cooling fan 15. Rectifier grille 16. Circulating cooling water source
  • Heat exchange tube 192 U-shaped piece 193. Angle steel retainer
  • Temperature sensor 102 First PI controller
  • Wind speed sensor 107 Neural network controller
  • sand blowing and dust blowing environment testing equipment generally including a circulating air duct, a compressed air source, a sand dust system, a temperature control system, a humidity control system, and a duct pressure control system, wherein the circulating air duct is used to provide the test piece In the place of blowing sand and dust blowing environment; compressed air source, sand dust system (not shown) is used to achieve the concentration of the experimental environment; temperature control system, humidity control system, air duct pressure control system constitute air conditioning system and Through the pipeline and the circulation air duct, a plurality of valves are arranged on each pipeline. The valve is arranged to effectively adjust the flow of sand dust in the air passage, the magnitude of the pneumatic force, the adjustment of the pressure, the adjustment of the humidity, the adjustment of the temperature, etc. .
  • FIG. 2 it is a schematic diagram of the aerodynamic layout structure of the circulating air duct 1 of the present invention.
  • the circulating air duct 1 is divided into a test section 101, a diffusing section 102, a separating section 103, and a first corner section 104 according to different structural sections thereof.
  • the first guide vane 115, the second guide vane 116, the third guide vane 117, and the fourth guide vane 118 are respectively disposed at the four corner sections 104, 106, 111, 112;
  • a main fan 13 is disposed in the power section 108, and a cooling fan 14 for cooling the main fan is disposed outside the main fan; and a honeycomb structure rectifying grid 15 is disposed in the stabilizing section 113.
  • the multi-functional U-shaped separator 19 in the separation section 103 of the present invention is an axial view and a cross-sectional view.
  • the multi-functional U-shaped separator is composed of a U-shaped member 192, an angle steel holder 193, and a heat exchange tube 191.
  • the U-shaped member 192 is welded to the angle steel 193 by spot welding, and the heat exchange tube 191 is welded to the bottom portion of the U-shaped member 192 by brazing, thereby realizing a separator having a heat exchange function.
  • the heat exchange tube is made of seamless steel pipe or welded steel pipe.
  • the axial view and the cross-sectional view of the guide vanes 115, 116, and 117 having the heat exchange function at the corners 104, 106, and 111 of the circulating air passage of the present invention are illustrated by taking 115 as an example.
  • the heat transfer medium through hole 119 is reserved in the guide vane of the streamlined curved surface, and the heat exchange working medium is injected into the heat exchange through hole 119, thereby realizing Guide vanes for heat exchange function.
  • an electric heater 1183 is embedded in the fourth guide vane 118 located at the corner portion 112 of the circulation duct, and a temperature sensor connected to the electric heater realizes rapid, efficient, and sensitive control of the temperature.
  • FIG. 7 a cross-sectional view of the fourth guide vane 118 of the present invention, the electric heater 1183 inside the guide vane 118 is sequentially coated with an electric heater insulating layer 1181 and a wear-resistant protective layer 1182, the fourth guide.
  • the flow vane 118 is connected to the temperature sensor by embedding the electric heater 1183 therein to realize uniform and rapid heating of the circulating air to the internal airflow on the basis of the airflow guiding function;
  • the electric heater insulating layer 1181 is made of an insulating material thermal conductive grease having excellent thermal conductivity.
  • an electric heater 1141 is disposed in the constricted section 114 of the circulation duct, and is connected to a temperature sensor connected thereto to achieve rapid, efficient, and sensitive temperature control;
  • the temperature control device of the present invention provides heat exchange working fluids of different temperatures from three devices of circulating cooling water source 16, chiller 17, and electric boiler 18, and heat exchange working sources of these three different temperatures.
  • Parallel connection between the pipeline and the manual valve, and through the pipeline and the U-shaped separator 19 with heat exchange function and the guide vanes 115-117 with heat exchange function in the circulation duct constitute a conventional temperature control system, and through the respective pipelines
  • the regulating valve realizes the control of the flow rate, so that the temperature of the airflow in the circulating air duct can be controlled; an electric heater is arranged inside the contraction section 114, and the contraction section 114 is located upstream of the air passage of the test section 101, and the temperature influence on the test section 101 is affected.
  • the electric heater 20 is also embedded in the guide vane 118 at the corner section 112.
  • the two sets of electric heaters constitute a rapid heating system, so that the two most recent places before the test section 101 It realizes rapid heating of the airflow in the circulating air duct, which improves the heating speed and heat utilization efficiency.
  • the temperature regulating system working fluid is provided by the circulating cooling water 16, the chiller 17, and the electric boiler 18.
  • the cooling and heating multi-purpose temperature control mode is adopted, and the circulating cooling water passes through the special pipeline. It is connected to the through holes of the heat pipe and the guide vanes 115-117 of the U-shaped separator 19, and the circulating cooling water back through the circulating cooling water outlet valve 163 and the circulating cooling water return pipe 162 provided on the circulating cooling water outlet pipe 161.
  • the water valve 164 controls the flow rate of the cooling water during the test; the chilled water from the chiller 17 passes through the chiller water outlet control valve 173 on the chiller water outlet pipe 171 and the chiller return water control valve 174 on the chiller return water pipe 172 To control the flow rate of the chilled water during the test; the circulating hot water heated in the electric boiler 18 passes through the electric boiler water outlet control valve 183 on the electric boiler water outlet pipe 181 and the electric boiler return water control valve 184 on the electric boiler return water pipe 182. Control the flow of hot water during the test.
  • the cooling medium of the chiller 17 is provided by the circulating cooling water source 16, and the electric boiler is selected according to the actual situation. Appropriate power.
  • the principle of the invention is as follows: a circulating cooling water source, a chiller, an electric boiler thermal circulating water and respective pipes form a refrigerant and a heat medium uniform temperature distribution network, and the distribution network and the heat exchange medium through holes in the three sets of guide vanes And the heat exchange tube in the u-type separator is connected, and the working fluid in the distribution network flows through the heat exchange tube in the u-type separator and the air flow in the circulation air passage to perform heat exchange, thereby realizing temperature adjustment, and the working fluid is still flowing.
  • the heat exchange between the heat exchange medium through holes in the three sets of guide vanes and the air flow in the circulation air duct realizes the adjustment of the air flow temperature in the circulation air passage.
  • the refrigerant and heat medium uniform temperature distribution network controls the flow rate of the working fluid in the heat exchange tube of the u-type separator and the heat exchange medium of the guide vane by adjusting a plurality of valves on the passage, thereby realizing automatic temperature control.
  • An electric heater is installed in the constricted section, and an electric heater is also added to the guide vane at the corner before the constricted section, and the electric heater is closedly controlled by the temperature signal set in the test section to quickly change the electric heater.
  • the output heat, the two sets of electric heaters constitute a rapid heating system for circulating airflow.
  • the temperature regulating device of the entire circulation duct is composed of two methods: heat exchange of the working medium and heat transfer of the electric heater.
  • the test air circulates in the circulation duct 1 and forms a test wind speed in the test section 101 of the circulation duct 1 to meet the experimental requirements, blowing 18 to 30 tn. /s, low-speed blowing dust 1. 5m / s, high-speed blowing dust 8. 9m / s, the speed of the wind speed can be closed-loop control by adjusting the speed of the main fan 13, the aerodynamic layout of the circulating air duct 1 is shown in Figure 2.
  • the section 114 and the stabilizing section 113 are provided with a rectifying grid 16 in the stabilizing section 113 to reduce the turbulence of the test section 101; in order to prevent the large-size sand dust particles from injuring the main fan 13 blade, the outlet of the test section 101 is provided.
  • the splitting block 102 and the separating section 103 are provided with four rows of U-shaped separators 19 arranged in the wind in the separating section 103.
  • the diffusing section 102 is disposed between the test section 101 and the separating section 103 to reduce the inlet section of the separating section 103. Wind speed, thus improving Solid separation efficiency of the action of the segments 103 away.
  • the conventional temperature control in the circulation duct is through a U-shaped separator 19 having a heat exchange function installed in the separation section 103, and three sets of guide vanes having heat exchange functions disposed at the circulation duct corners 104, 106, 111. 115-117 to control, the system is adjusted to the temperature of test section 101.
  • the effluent temperature of the circulating cooling water is 35 ° C, which is used to balance the system heat load during the high temperature sand blowing test in which the temperature requirement of the test section 101 is 70 ° C.
  • the effluent temperature of the chiller 17 is
  • TC is used to balance the system heat load during normal temperature blowing (23 ° C) and high temperature blowing (70 ° C).
  • the outlet temperature of the electric boiler is 90 °C, which is used to rapidly heat the temperature of the airflow in the circulation duct 1, so that the temperature of the test section 101 quickly reaches a preset state; in the test process, it is first detected whether the temperature of the test section 101 is satisfied.
  • Test expected requirements if temperature If the degree is lower than the expected temperature, the control valves 163, 164 on the outlet and return passages of the circulating cooling water and the control valves 173, 174 on the outlet and return passages of the chiller 17 are closed, and the electric boiler is manually adjusted by a long distance.
  • the regulating valve 183 on the water outlet passage and the regulating valve 184 of the return water passage are used to adjust the hot water flow of the U-shaped separator heat exchange tube 191 and the guide vane heat exchange medium through hole 119 in the circulation duct 1 to adjust the circulation.
  • the regulating valve 164 adjusts the cooling water flow of the U-shaped separator heat exchange tube 191 and the guide vane heat exchange medium through hole 119 in the circulation duct 1, and can also manually adjust the outlet passage regulating valve of the chiller 17 by adjusting the distance.
  • the 173 and the return water passage regulating valve 174 adjust the chilled water flow rate of the U-type separator heat exchange tube 191 and the guide vane heat exchange medium through hole 119 in the circulation duct 1.
  • the rapid temperature control system in the circulation duct 1 is controlled by an electric heater 1141 installed on the inner wall of the constricted section 114 and an electric heater 1183 in the fourth guide vane 118 having a heat exchange function inside the circulation duct corner 112.
  • the system is adjusted to the temperature of the test section 101.
  • the closed loop automatically adjusts the power of the electric heater so that the temperature of the test section 101 meets the test requirements.
  • an intelligent temperature control method for a sand dust environment test system is implemented by a PI intelligent temperature control system based on a neural network, the intelligent temperature control system comprising:
  • the wind speed sensor 106 is installed in the circulation duct 1
  • a temperature sensor 101 is installed in the circulation duct 1
  • the neural network controller 107 derives a coordinated control factor based on the wind speed measured from the wind speed sensor 106, and has determined which device is the primary control device and which device is the auxiliary control device.
  • the neural network controller specifically includes an input layer, an implicit layer and an output layer;
  • the PI controller is configured to receive a coordinated control factor from the neural network controller and a temperature value measured from the temperature sensor, and transmit the processed control amount to the temperature control device of the dust environment test system.
  • the PI controller comprises:
  • a first PI controller 102 configured to generate a first coarse control amount according to the temperature value and the coordinated control factor
  • a second PI controller 109 configured to generate a second coarse control according to the temperature value and the coordinated control factor, wherein the intelligent temperature control system further includes:
  • the second limiter 110 is configured to limit and optimize the second coarse control amount to generate an optimized fine control amount for circulating the cooling water outlet valve 163.
  • the invention relates to an intelligent temperature control method for a dust environment test system, which comprises the following steps: Step 1: Establish a neural network system structure
  • a neural network in accordance with an embodiment of the present invention is constructed using a three-layer feedforward network of a single hidden layer.
  • the transformation function of the hidden layer unit adopts the positive and negative symmetry 1 ⁇ ' ⁇ function, which is a locally distributed non-negative nonlinear function for the radial symmetric attenuation of the center point.
  • the mapping of the neural network from the input layer space to the hidden layer space is nonlinear, and the hidden light from the hidden layer space to the output layer space is linear.
  • Such a neural network realizes the characteristics of speeding up learning and avoids Certain oscillation and local minimum values. As shown in Figure 13.
  • the input and output functions of the input layer neurons are:
  • the output of the hidden layer neurons is:
  • Tanh(x) eXp( ) - eXp( - )
  • the output of the output layer neurons is:
  • the output of the neurons in the output layer corresponds to the coordinated control factor Zcv, which determines the second step of the hybrid learning training of neural network parameters.
  • the improved BP algorithm makes the network constantly approach the real model by continuously correcting the weights and thresholds between the network neurons.
  • the online genetic learning training algorithm is to overcome some limitations of BP offline learning training. Because the genetic learning training algorithm applies high-dimensional feasible solution space in the optimization process to generate multiple starting points randomly and start searching at the same time, the fitness function is obtained. To guide the search direction, the search area is wide and the search efficiency is high. Moreover, in real-time control, before each online adjustment of the weight parameter, it is necessary to determine whether the temperature difference is too large. If the specified limit is exceeded, the offline model learning training must be performed until the difference is within the allowable range. After the internal training, online learning. This reduces oscillations and reduces control time, as shown in Figure 15.
  • BP Back Propagation
  • the dynamic model is initialized, it is processed from the input layer through the hidden layer unit layer to the output layer, which is a function of the input and the weight.
  • the output of each unit of the hidden layer and the output layer is obtained, and the expectation is obtained.
  • the deviation between the output and the actual output judge the deviation value, if there is a deviation between the actual output and the expected output, enter the back propagation, reverse back according to the original forward propagation path, calculate the hidden layer unit error, and press the error function
  • the negative gradient direction is solved, and then the weight coefficients of the neurons in each layer are corrected, and finally the expected error function tends to be minimized.
  • the momentum factor is a weighting factor that determines the past learning effect, usually 0 ⁇ 1 , and the corresponding other variation coefficient 77 ranges from 0 ⁇ 77 ⁇ () ⁇ 5 , and the effect is ideal. After each iteration, these two factors will be continuously adjusted and corrected. Considering the coupling between variables, the performance index function is:
  • the genetic algorithm is used to correct the weight parameters of the obtained network online in real time.
  • the search process is based on the original parameters, and the original parameters are smaller. Conducted in the field.
  • Genetic algorithm is a search algorithm based on natural selection and population genetic mechanism. It simulates the breeding, mating and variation phenomena in natural selection and natural genetic transmission. It treats each possible solution as an individual in the group, and encodes each individual into a form of a string, evaluating each individual according to a predetermined objective function, giving an fitness value.
  • the fitness values of these individuals use genetic operators to genetically manipulate these individuals, retain superior individuals, and eliminate poor individuals, so that the final weights develop toward an excellent state.
  • the actual value of the decision variable is used, and the length of the individual's code is equal to the number of decision variables.
  • the selection algorithm in genetic manipulation uses a standardized geometric ordering approach. The sorting method sorts the individuals by the sum of the values, and assigns a probability according to the position of the individual. Standardized geometric ordering defines the individual's choice probability formula as -
  • is the selection probability of the best individual
  • A is the serial number of the individual
  • the cross algorithm uses a combination of mathematical crossover and heuristic crossover. The two crossover methods are combined. The detection capability of the algorithm can be enhanced. In order to maintain the diversity of the population and prevent precocity, a random perturbation of the genes in the original population is required.
  • the mutation operation in this system uses the multiNonUnJfMutafion strategy to generate the mutated gene to form a new population, that is, after the non-uniform variation of the independent variable in its solution space, a random combination is taken as the variation result.
  • the principle is:
  • the training steps for optimizing the neural network weight coefficient by the genetic algorithm are as follows:
  • the forward propagation algorithm is used to obtain N network outputs corresponding to the N group network weights.
  • step 9 returns to step 4 until a performance requirement is met, resulting in a set of optimized weight coefficients.
  • the coordinated control factor is input to the PI controller, and the PI controller further coordinates and combines the coordinated control factor with its parameters to obtain a control variable, thereby performing coordinated and effective control on the controlled mechanism.
  • the governing equation for circulating cooling water flow is:
  • G - the mass flow of the cooling water, the temperature difference, the proportional coefficient of the circulating cooling water flow controller
  • K cd differential coefficient of circulating cooling water flow controller
  • the S-function formula is:
  • PI controller is widely used in process control and motion control of electromechanical, metallurgical, mechanical, chemical and other industries due to its simple design, easy implementation, high reliability, etc., especially suitable for establishing Deterministic control system for accurate mathematical models.
  • PI controller is widely used in process control and motion control of electromechanical, metallurgical, mechanical, chemical and other industries due to its simple design, easy implementation, high reliability, etc., especially suitable for establishing Deterministic control system for accurate mathematical models.
  • the neural network has the characteristics of nonlinear mapping, self-learning ability, distributed storage capacity and processing information. Therefore, the neural network is combined with the PI controller, but it is different from the traditional combination.
  • the combination is based on the combination of coordinated control. Due to the uncertainty and complexity of coordinated control, we use the nonlinear, self-learning ability of the neural network to coordinate control before the input signal enters the PI controller. A pre-judgment, a coordination control factor is determined, which control mechanism is the main control mechanism and which is the auxiliary control mechanism among all the control mechanisms of the coordinated control, so that the control system not only has the processing inaccuracy, the uncertainty The ability to coordinate control stability.
  • the neural network connection weight is continuously corrected, and the coordination control factor is adjusted, so that the factor is the optimal coordinate value of the optimal coordinated control, in order to meet the requirements of the system performance index.
  • the coordinated control factor is sent to the PI controller to perform the corresponding control operation. After the PI controller output signal is output, the corresponding limiting processing is also performed, and the control variable is further refined to finally reach the sand dust environment test.
  • the coordination and reliability of the temperature control of the wind tunnel It reduces the fluctuation range of the dust temperature adjustment process.
  • the neural network structure of the temperature intelligent control system based on the neural network PI is composed of a three-layer network with a single hidden layer.
  • the transformation function of the hidden layer element uses a function in the radial basis function, which is a locally distributed non-negative nonlinear function of the radial symmetric attenuation of the center point.
  • the mapping of the neural network from the input space to the hidden layer space is nonlinear, and the hidden light from the hidden layer space to the output layer space is linear.
  • Such a neural network realizes the characteristics of speeding up learning and avoids certain Oscillation and local minimum values.
  • the neural network after learning and training can adjust the parameters well, and the nonlinear function of the input and output signals of the temperature intelligent control is highly accurate, and has a strong generalization ability.
  • the process of learning and training is to enter the judgment module to determine whether the temperature difference exceeds a limit value. If the limit value is exceeded, offline dynamic temperature model learning is performed, so that the error amount can be reduced before online learning , shorten the time for online learning.
  • the offline model learning adjustment parameter is essentially an offline global optimization of weights using the improved BP algorithm. After that, the judgment is continued until the difference is within the limit value range, and then the online learning training phase is performed.
  • the online learning adjustment weight parameter is mainly generated by a genetic algorithm, and a plurality of starting points are randomly generated in the high-dimensional feasible solution space and simultaneously The initial search, the fitness function guides the search direction, and finally obtains the optimal weight parameter.
  • a dynamic model of the expert system is employed -
  • the invention can realize the temperature control of the dust environment test device system, meets the temperature conditions required for the environmental test, is not only simple in structure, realizes integration, and can quickly and efficiently control the temperature, and can be used for structure merging, convenient cleaning, frequent The occasion of the test.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Fluid Mechanics (AREA)
  • Testing Resistance To Weather, Investigating Materials By Mechanical Methods (AREA)

Description

砂尘环境试验系统的温度调节装置及智能温度控制方法 技术领域
本发明涉及一种砂尘环境试验系统的温度调节装置及智能温度控制方法,属于砂尘环境 试验技术领域。
背景技术
砂尘环境是引起许多工程或军用武器装备失效的一个重要环境因素, 其主要损坏类型 有: 冲蚀、 磨损、 腐蚀及渗透等。 吹尘和吹砂试验是检验车辆、 飞行器、 电器设备、 军用装 备在沙漠、 千旱地区和风沙天气条件下环境适应性和环境可靠性的重要手段。
现有的砂尘环境试验系统采用一大一小两个空调箱完成对温度的控制,如图 1所示中国 发明专利: ZL200510000070. 8公布了一种系统, 其中该系统包括: 循环风道 1, 大空调 2, 小 空调 3, 表冷器翅片管 4, 翅片电加热器 5, 变频调速风机 6, 大空调进气通道 7, 小空调进气 通道 8, 空调回气通道 9, 冷水机组水源 10, 大空调压缩空气吹除管道 11, 小空调压缩空气吹 除管道 12。 该系统存在如下缺陷:
1、 由于使用了大空调 2与小空调 3两个空调来控制风洞温度, 使得该系统的结构复杂, 并且占用了大量的空间, 大空调进气通道 7、 小空调进气通道 8及空调回气通道 9都与循环风 道直接相通, 这些通道 7、 8、 9影响了风道 1内气流的品质;
2、为实现对温度的控制, 它们分别使用循环冷却水和来自冷水机组水源 10的冷冻水作 为冷却工质, 并且小空调箱中还装有电加热器 5和变频调速风机 6, 它们共同相配合完成不同 工况和不同热负荷下的温度调节任务, 冷却工质是通过流经表冷器翅片管 4以实现冷却, 相 应的通过电加热器 5实现加热, 由于气流带砂尘, 通过表冷器翅片管 4和翅片式电加热管 5时 会积尘, 严重影响传热, 需要额外设置压缩空气吹除管道 11、 12, 清理极不方便;
3、 循环风道 1采用外保温结构, 热容很大, 当做完一次高温试验后再去做低温试验时, 由于表冷器 4传热面积有限, 需要很长的时间来冷却, 严重影响了试验装置系统的利用效率。
另外, 由于在砂尘环境试验中, 试验风速及风洞内热负荷的变化范围很大以及突然性, 这就越来越凸显出了砂尘环境试验温度控制的重要性。目前国内在大型洄流式砂尘环境试验 风洞的温度控制主要是釆用空调旁路中装备表冷器和电加热器两种截然不同的温控设备。并 且由于在进行低速吹尘和高速吹尘试验时,所采取的是加热和制冷两种完全不同的温度控制 措施, 使温差变化范围增大, 控制难度增大。 同时由于在砂尘环境试验中的温度控制具有纯 滞性, 非线性以及参数的一些不确定性等特点, 很容易产生较大的振荡或误差性, 因此传统 的控制策略往往很难满足性能的指标要求,同时也是砂尘环境试验风洞温度控制不同于普通 温度控制对象的特点与难点。
循环风道内, 风机转速变化, 辅助气流流量与温度变化, 环境温度的变化均会引起风 道内的温度扰动, 而通过调节电加热器加热功率或冷却水流量, 则能使温度扰动得到控制, 但风道内的热负荷差别很大, 需要采用不同的协调控制措施。如何消除这些干扰源引起的扰 动, 并克服各种非线性的因素对温度控制系统带来的影响, 即温度的高精度和高协调性控制 方法是整个温度控制领域的关键技术。
目前, 解决砂尘环境试验风洞下的温度控制有两种思路, 一种是在硬件结构上采用温 控设备, 让温度可在正常的范围内进行调控。 这种方案虽然可以很好的提高温度的控制, 但 是却增加了硬件的复杂性,提高了成本,并且在可靠性上有很大的限制;二是软件控制方式。 该方式通过采用建模和控制算法, 即通过控制策略和学习能力逼近任意的非线性映射的能 力, 对温度进行实时的控制, 并且对由于非线性因素引起的误差进行有效的补偿, 影响其补 偿的精度, 保证其精确性和协调性。 发明内容
本发明的目的在于提供一种砂尘环境试验系统的温度调节装置及智能温度控制方法,以 克服现有技术中存在的如下缺陷:
1 ) 空调装置需要专门的空调风道和换热面、 并带来额外的运行能耗, 此外此空调 风道的存在也影响循环风道内气流的流动品质。
2 ) 由于循环风道内气流带砂尘, 通过表冷器时和电加热管时会积尘, 严重影响传 热, 并且清理极不方便。
3) 利用循环风道内部已有的分离器、 导流叶片及收缩段, 实现对循环风道内气流 温度的控制。 提供一种结构紧凑、 清理方便、 设备利用效率高的温度调节系统。 本发明的另一目的在于提供一种砂尘环境试验系统的智能温度控制方法,以克服现有技 术在高精度, 高稳定性和协调控制方面存在的不足。
本发明一种砂尘环境试验系统的温度调节装置, 该砂尘环境试验系统主要包括: 循环风道, 为不规则结构的密闭风道, 用于提供吹砂和吹尘环境的场所; 该循环风道包 括有第二收缩段、 试验段及分离段等,
主风机, 用于驱动所述循环风道内气流的流动;
U型分离器, 设置于分离段, 用于对试验后的砂尘进行分离回收; 及 导流叶片, 分四组分别设置在循环风道的四个拐角处;
其特征在于: 该温度调节装置包括:
在所述的 U型分离器内部铺设热交换管;
所述四组导流叶片, 在其中的三组导流叶片中分别设置有热交换工质通孔; 另外一组导 流叶片中设置有加热装置;
在所述的第二收缩段设置有电加热器。 其中, 所述的温度调节装置进一步包括- 循环冷却水水源;
冷水机组; 及
电锅炉;
所述的循环冷却水水源、冷水机组及电锅炉分别通过管道与具有热交换功能的导流叶片 及 u型分离器相连; 在各管道上设置有用于调节冷热工质流量的阀门。 其中, 所述的导流叶片中的加热装置为一电加热器, 该电加热器外依次包覆有绝缘层及 耐磨保护层。 本发明一种砂尘环境试验系统的智能温度控制方法, 该方法是通过基于神经网络的 PI 智能温度控制系统实现, 该智能温度控制系统包括:
风速感应器, 安装于循环风道内,
温度感应器, 安装于循环风道内,
神经网络控制器, 根据来自风速感应器所测得的风速值得出一个协调控制因子, 以决定 哪种设备是主要控制设备, 哪种设备是辅助控制设备, 充分达到协调控制的目的。 该神经网 络控制器具体包括一输入层, 一隐含层和一输出层;
PI控制器,用于接收来自神经网络控制器的协调控制因子以及来自温度感应器所测得的 温度值, 并将经过处理的控制量传递给沙尘环境试验系统的温度控制装置。
其中, 所述的 PI控制器包括:
第一 PI控制器, 用于根据所述温度值和所述协调控制因子, 产生第一粗略控制量, 第二 PI控制器, 用于根据所述温度值和所述协调控制因子, 产生第二粗略控制量。 其中, 所述的智能温度控制系统, 进一步包括: 第一限幅器, 用于对所述第一粗略控制量进行限幅和优化处理, 从而产生用于电 锅炉回水控制阀的优化精细控制量,
第二限幅器, 用于对所述第二粗略控制量进行限幅和优化处理, 从而产生用于循 环冷却水出水阀的优化精细控制量。 本发明一种砂尘环境试验系统的智能温度控制方法, 该方法包括如下步骤:
步骤一、 建立神经网络系统结构
本发明采用单隐层的三层前馈网络构成; 隐含层单元的变换函数采用正负对称的
S gm 'i/函数, 它是一种局部分布的对中心点径向对称衰减的非负非线性函数, 此神经网络 由输入空间到隐含层空间的映射是非线性的, 而从隐含层空间到输出层空间的隐射是线性 的, 这样的神经网络实现了加快学习速度的特点, 以及避免了一定的振荡性和局部极小值问 题, 如图 13所示, 该三层前馈网络包括:
1 ) 输入层
输入层采用了特殊的 3个输入, 分别对应输入 V (风速值), 通过安装在循环风道内的风 速感应器获得;误差 ee (温度的误差),通过安装在循环风道内的温度感应器获得温度值之后, 进行减法运算获得; 常量 1, 常量在这里起一个干扰的作用, 则输入模式向量为 x = [v,ee,l], 比起 2个的输入向量的结构 jc = [v,ej更符合实际的工作环境;
2 ) 隐含层 ;
3 )输出层 : 输出层的输出对应着协调控制因子 Zcv , 决定哪种设备是主要控制设备, 哪种是辅助控制设备。 步骤二、 神经网络参数的混合学习训练
在所述步骤一得到的神经网络模型的基础上,采用在线训练和离线训练相结合的方式进 行, 经过预先判断之后, 决定是进行 BP离线训练, 还是进行基于遗传算法的在线训练; 步骤三、 PI控制
由所述步骤一和二最终得到协调控制因子之后, 将协调控制因子输入给 PI控制器, PI 控制器将协调控制因子与其参数进行进一步的运算与结合, 最终得出控制变量, 对被控机构 进行控制。
步骤四、 限幅处理
通过步骤三得到控制变量之后, 应用 S型函数进行限幅处理, 使控制变量最优化。 其中, 所述步骤二中, 进行神经网络参数的混合学习训练, 要预先进行判断模块, 当 温度差值好过规定界限值值时, 进行基于 BP学习算法的离线动态温度模型学习训练, 实现 权值的离线全局寻优,使误差量缩小, 而缩短在线学习所需的时间; 相反, 当在界定值范围 以内时, 利用遗传算法进行在线调整, 使权值最优化。
其中, 利用所述神经网络控制器, 根据所述风速值, 得出一个协调控制因子; 根据该协 调控制因子把至少两个温度控制机构中的一个确定为主要控制机构,把至少两个温度控制机 构中其余的控制机构确定为辅助控制机构, 利用所述 PI控制器, 根据所述温度值和所述协 调控制因子而得到所述粗略控制量。
本发明的优点: 1 ) 与传统的吹砂吹尘环境试验装置系统相比, 由于本发明舍弃了一大 一下两个空调箱, 故消除了空调箱通气回路对循环风道内气流品质的影响, 并且使结构更加 简单, 占据空间更小; 2 ) 由于本发明没有了空调箱中的表冷器和电加热器, 故消除了清理 积尘带来的不便和积尘对传热的影响; 3 ) 本发明在传统的冷却系统的基础之上加入循环热 水系统, 故使传统的冷却系统变成了冷热多用温度控制系统, 提高了设备的利用率; 4 >本 发明在收缩段和收缩段之前的一组导流叶片上分别加入了电加热片,可以进一步加快对循环 风道内气流的加热速度, 提高系统的工作效率; 5 ) 本发明在 U型分离器内铺设热交换管, 并且对剩下的三组导流叶片进行改造, 在每个导流叶片内部设置通孔, 最后分别将热交换管 和通孔与冷热温度控制系统相连, 故充分利用了己有设备, 提高了设备利用效率, 更重要的 是增加了热交换面积, 使传热更加快速。
附图说明
图 1 为现有技术中砂尘环境试验系统的连接示意图。
图 2 本发明循环风道气动布局结构示意图。
图 3 是本发明循环风道剖视图。
图 4A 是本发明具有热交换功能的 U型分离器轴视图
图 4B是本发明具有热交换功能的 U型分离器的剖视图。
图 5A是本发明具有热交换功能的导流叶片的轴视图
图 5B是本发明具有热交换功能的导流叶片的剖视图。
图 6 是本发明循环风道局部剖视图。 图 7 是本发明具有加热功能的导流叶片剖视图。
图 8 是本发明沙尘环境试验系统的温度调节装置示意图。
图 9 是本发明常规温度控制系统示意图。
图 10为本发明的砂尘环境试验风洞的总体流程图。
图 11为本发明的砂尘环境试验风洞温度智能控制的控制流程图。
图 12为本发明的砂尘环境试验风洞温度智能控制基于神经网络 PI控制的原理结构图。 图 13为本发明的砂尘环境试验风洞温度智能控制的神经网络控制器的结构图。
图 14 为本发明的砂尘环境试验风洞温度智能控制的基于混合学习训练算法的神经网络 PI控制结构图。
图 15为本发明的砂尘环境试验风洞基于神经网络 PI智能温度控制系统学习训练算法总 体流程图。
图 16为本发明的砂尘环境试验风洞温度智能控制的改进 BP学习算法的流程图。 图中: 1.循环风道 101.试验段 102.扩压段 103.分离段 104.第一拐角段 105.第一收缩段 106.第二拐角段. 107.方圆过渡 108.动力段 109.圆方过渡 110.变截面段 111.第三拐角段 112.第四拐角段 113.稳定段 114.第二收缩段 1141.电加热器 115.第一导流叶片 116.第二导流叶片
117.第三导流叶片 118.第四导流叶片 1181.绝缘层 1182.耐磨保护层
1183.电加热器 119.热交换工质通孔 2.大空调 3.小空调 4.表冷器翅片管 5. 翅片电加热器 6.变频调速空调 ' 7.大空调进气通道 8.小空调进气通道
9.空调回气通道 10.冷水机组水源 11.大空调吹尘管道 12.小空调吹尘管道 13.主风机 14.冷风机 15.整流格栅 16.循环冷却水水源
161.循环冷却水出水管道 162.循环冷却水回水管道 163. 循环冷却水出水阀
164.循环冷却水回水阀 17.冷水机组 171.冷水机组出水管道 172.冷水机组回水管道 173.冷水机组出水控制阀 174.冷水机组回水控制阀 18.电锅炉 181.电锅炉出水管道 182.电锅炉回水管道
183.电锅炉出水控制阀 184.电锅炉回水控制阀 19. U型分离器
191.热交换管 192. U型件 193.角钢固定器
e (t) : 温度偏差的精确量 ec: 误差变化率的精确量
r (t) : 温度输入值 y (t) : 被控温度量输出值 W: 每层之间的连接权值 Zc„: 协调控制因子
V: 风速值
101: 温度感应器 102: 第一 PI控制器
106: 风速感应器 107: 神经网络控制器
108: 第一限幅器 109: 第二 PI控制器
110: 第二限幅器 具体实施方式
下面将结合附图对本发明的技术方案作进一步的详细说明。
对于吹砂、 吹尘环境试验设备中, 一般包括循环风道, 压缩空气源、 砂尘料系统、 温度 控制系统、 湿度控制系统、 风道压力控制系统, 其中, 循环风道用于提供试件在吹砂和吹尘 环境的场所; 压缩空气源、 砂尘料系统(图中未示)用于实现实验环境的浓度; 温度控制系 统、 湿度控制系统、 风道压力控制系统组成空气调节系统并通过管道与循环风道实现连通, 各个管道上设有多个阀门, 阀门的设置是为有效的调节风道中的砂尘的流量、 气力的大小、 压力的调节、 湿度的调节、 温度的调节等。
如图 2所示, 为本发明循环风道 1的气动布局结构示意图, 循环风道 1按其不同结构段 依次分为试验段 101、 扩压段 102、 分离段 103、 第一拐角段 104、 第一收缩段 105、 第二拐 角段 106、 方圆过渡段 107、 动力段 108、 圆方过渡段 109、 变截面段 110、 第三拐角段 111、 第四拐角段 112、 稳定段 113、 第二收缩段 114。
如图 3所示, 在四个拐角段 104、 106、 111、 112分别设有第一导流叶片 115、 第二导 流叶片 116、 第三导流叶片 117、 第四导流叶片 118; 在动力段 108内设有主风机 13, 主风 机外部设有用于对其进行降温的冷却风机 14; 在稳定段 113内设有蜂窝结构整流格栅 15。
如图 4A及图 4B所示, 为本发明位于分离段 103的多功能 U型分离器 19轴视图和剖视 图, 多功能 U型分离器由 U型件 192、 角钢固定器 193、 热交换管 191构成, U型件 192通 过点焊焊接在角钢 193上, 热交换管 191通过钎焊焊接在 U型件 192内侧底部, 实现了具有 热交换功能分离器。
其中热交换管采用无缝钢管, 或者焊接钢管。
如图 5A及图 5B所示, 为本发明位于循环风道拐角 104、 106、 111处的具有热交换功能 的导流叶片 115、 116、 117轴视图及剖视图, 以 115为例说明, 通过在呈流线型曲面的导流 叶片内部预留热交换工质通孔 119, 于改热交换通孔 119中注入热交换工质, 便实现了具有 热交换功能的导流叶片。
如图 6所示, 在位于循环风道拐角段 112处的第四导流叶片 118内嵌入电加热器 1183, 通过与该电热器相连的温度传感器, 实现温度的快速, 高效, 灵敏的控制。
如图 7所示,为本发明第四导流叶片 118的剖视图,导流叶片 118内部的电加热器 1183, 依次包覆有电加热器绝缘层 1181、 耐磨保护层 1182, 该第四导流叶片 118通过在其内部嵌 入电加热器 1183与温度传感器连接实现在对气流导向功能的基础之上再实现对循环风到内 气流的均匀快速加热;
其中, 电加热器绝缘层 1181采用导热性能优异的绝缘材料导热硅脂。
再参阅图 6所示, 在循环风道的收缩段 114内铺设有电加热器 1141, 通过与其相连的 温度传感器接, 实现温度的快速, 高效, 灵敏的控制;
上述第四导流叶片 118内的电加热器 1183与收缩段 114的电加热器 1141共同完成对循 环风道内气流的快速均匀加热。
如图 8所示, 本发明的温度控制装置由循环冷却水水源 16、 冷水机组 17、 电锅炉 18三 种装置提供不同温度的热交换工质,将这三种不同温度的热交换工质源通过管道和手动阀门 并联, 再通过管道和循环风道内的具有热交换功能的 U型分离器 19及具有热交换功能的导 流叶片 115- 117相通组成了常规温度控制系统,并且通过各自管路上的调节阀门实现对流量 的控制, 这样就可以对循环风道内气流温度实施控制; 在收缩段 114内部设置电加热器, 收 缩段 114处在试验段 101风道上游, 对试验段 101的温度影响最直接, 加热效果最明显, 在 拐角段 112处的导流叶片 118里面也嵌入了电加热器 20, 两组电加热器组成了快速加热系 统, 这样就能在试验段 101之前最近的两处实现对循环风道内气流的快速加热, 提高了加热 速度和热利用效率。
如图 9 所示, 为本发明的常规温度控制系统示意图, 温度调节系统工质由循环冷却水 16、 冷水机组 17、 电锅炉 18提供, 采用冷热多用温度控制模式, 循环冷却水通过专用管道 和 U型分离器 19热管及导流叶片 115-117的通孔相连, 通过设在循环冷却水出水管道 161 上的循环冷却水出水阀 163和循环冷却水回水管道 162上的循环冷却水回水阀 164来控制试 验时冷却水的流量; 从冷水机组 17出来的冷冻水通过冷水机组出水管道 171上的冷水机组 出水控制阀 173和冷水机组回水管道 172上的冷水机组回水控制阀 174来控制试验时冷冻水 的流量;在电锅炉 18中加热的循环热水通过电锅炉出水管道 181上的电锅炉出水控制阀 183 和电锅炉回水管道 182上的电锅炉回水控制阀 184来控制试验时热水的流量。
其中, 冷水机组 17的冷却工质由循环冷却水水源 16提供, 电锅炉根据实际情况选择合 适的功率。
本发明的原理为: 循环冷却水水源、 冷水机组、 电锅炉热循环水以及各自管道组成一个 冷媒、热媒统一温度调配网络, 该调配网络与三组导流叶片内的热交换工质通孔及位于 u型 分离器内的热交换管连接,调配网络内的工质流经 u型分离器内的热交换管和循环风道内的 气流进行热交换, 进而实现温度的调节, 工质还流经三组导流叶片内的热交换工质通孔和循 环风道内气流进行热交换, 从而实现对循环风道内气流温度的调节。 冷媒、 热媒统一温度调 配网络通过调节通道上的多个阀门来控改变 u 型分离器热交换管及导流叶片热交换工质通 孔内工质的流量, 从而实现对温度的自动控制。 在收缩段安装有电加热器, 并且在收缩段之 前拐角处的导流叶片内也加入了电加热器,再通过设在试验段的温度信号对电加热器做闭环 控制来快速改变电加热器的输出热量,这两组电加热器组成了对循环风洞气流的快速加热系 统。 整个循环风道的温度调节装置由工质的热交换和电加器的热输两种方式共同组成。
下面将介绍本发明设备的工作过程:
请参见附图 8所示, 在主风机 13的驱动下, 试验用空气在循环风道 1中循环流动并在 循环风道 1的试验段 101形成满足实验要求的试验风速,吹砂 18〜30tn/s,低速吹尘 1. 5m/s, 高速吹尘 8. 9m/s, 风速的大小可以通过调整主风机 13转速的方式加以闭环控制, 循环风道 1的气动布局如图 2所示, 为了减小循环风道 1的阻力损失, 在循环风道 1的四个拐角设有 四组导流叶片 115-118; 为了改善试验段 101的气流品质, 在试验段 101上游设有第二收缩 段 114和稳定段 113, 稳定段 113内设有整流格栅 16以减少试验段 101的紊流度; 为了避 免大粒径砂尘颗粒对主风机 13叶片造成伤害, 在试验段 101出口设有扩压断 102与分离段 103, 在分离段 103内设有迎风交错布置四排 U型分离器 19, 扩压段 102设在试验段 101与 分离段 103之间以降低分离段 103入口截面的风速,从而起到提高分离段 103的气固分离效 率的作用。
循环风道内的常规温度控制是通过安装在分离段 103 的具有热交换功能的 U型分离器 19, 以及设置在循环风道拐角 104、 106、 111处的三组具有热交换功能的导流叶片 115- 117 来控制的, 系统的被调量为试验段 101温度。 循环冷却水的的出水温度为 35°C, 用于平衡 试验段 101温度要求为 70°C的高温吹砂试验时的系统热负荷。 冷水机组 17 的出水温度为
TC , 用于平衡常温吹砂 (23°C )、 高温吹尘 (70°C ) 试验时的系统热负荷。 电锅炉的出 水口温度为 90°C, 用来快速加热循环风道 1 内气流的温度, 使试验段 101温度快速达到预 先设定的状态; 在试验过程中首先检测试验段 101的温度是否满足试验预期的要求, 如果温 度低于预期温度, 则关闭循环冷却水的出水和回水通道上的控制阀 163、 164与冷水机组 17 的出水和回水通道上的控制阀 173、 174, 并通过远距离手工调节电锅炉出水通道上的调节 阀 183与回水通道的调节阀 184, 用来调整循环风道 1内 U型分离器热交换管 191及导流叶 片热交换工质通孔 119的热水流量来调整循环风道内的气流温度;如果试验段的温度高于预 期温度, 则关闭电锅炉出水和回水通道的控制阔 183、 184, 并通过远距离手工调节循环冷 却水出水通道调节阀 163和回水通道调节阀 164来调整循环风道 1 内 U型分离器热交换管 191及导流叶片热交换工质通孔 119的冷却水流量, 还可以通过调节远距离手工调节冷水机 组 17的出水通道调节阀 173和回水通道调节阀 174来调整循环风道 1内 U型分离器热交换 管 191及导流叶片热交换工质通孔 119的冷冻水流量。
循环风道 1内快速温度控制系统是通过安装在收缩段 114内壁的电加热器 1141,以及循 环风道拐角 112内部的具有热交换功能的第四导流叶片 118内的电加热器 1183来控制的, 系统的被调量为试验段 101温度,根据试验段 101的温度信号反馈闭环自动调节电加热器的 功率使试验段 101温度满足试验要求。
如图 10、 11所示, 一种砂尘环境试验系统的智能温度控制方法, 该方法是通过基于神 经网络的 PI智能温度控制系统实现, 该智能温度控制系统包括:
风速感应器 106, 安装于循环风道 1内,
温度感应器 101 , 安装于循环风道 1内,
神经网络控制器 107,根据来自风速感应器 106所测得的风速值得出一个协调控制因子, 已决定哪种设备是主要控制设备, 哪种设备是辅助控制设备。 该神经网络控制器具体包括一 输入层, 一个隐含层和一输出层;
PI控制器,用于接收来自神经网络控制器的协调控制因子以及来自温度感应器所测得的 温度值, 并将经过处理的控制量传递给沙尘环境试验系统的温度控制装置。
其中, 所述的 PI控制器包括:
第一 PI控制器 102, 用于根据所述温度值和所述协调控制因子, 产生第一粗略控 制量,
第二 PI控制器 109, 用于根据所述温度值和所述协调控制因子, 产生第二粗略控 其中, 所述的智能温度控制系统, 进一步包括:
第一限幅器 108, 用于对所述第一粗略控制量进行限幅和优化处理, 从而产生用 于电锅炉回水控制阀 184的优化精细控制量, 第二限幅器 110, 用于对所述第二粗略控制量进行限幅和优化处理, 从而产生用 于循环冷却水出水阀 163的优化精细控制量。
本发明一种沙尘环境试验系统的智能温度控制方法, 该方法包括如下步骤: 步骤一、 建立神经网络系统结构
根据本发明的一个具体实施例的神经网络采用单隐层的三层前馈网络构成。隐含层单元 的变换函数采用正负对称的1 ^'^^^函数, 它是一种局部分布的对中心点径向对称衰减的非 负非线性函数。此神经网络由输入层空间到隐含层空间的映射是非线性的, 而从隐含层空间 到输出层空间的隐射是线性的, 这样的神经网络实现了加快学习速度的特点, 以及避免了一 定的振荡性和局部极小值问题。 如图 13所示。
1 ) 输入层
输入层采用了特殊的 3个输入, 分别对应输入 V (风速值), 误差 ec (温度的误差), 和常 量 1, 常量在这里起一个干扰的作用, 则输入模式向量为 X = [V, ,1], 比起 2个的输入向量 的结构 x = [v, ]更符合实际的工作环境。
输入层神经元的输入输出函数为:
0, = x(i) 式中, 为输入层神经元的个数, = 123
2 ) 隐含层 隐含层的神经元的输入为: netj (k) = j wjiOi
'=' (2) 式中, W '为输入层到隐含层的权值
隐含层神经元的输出为:
= (3) 式中, 隐含层神经元的个数, 为隐含层的激活函数 隐含层的激活函数取正负对称的1 ^g^^ 函数: tanh(x) = eXp( )-eXp(- )
exp( ) + exp(- ) (4)
3) 输出层 输出层的神经元的输入为:
7=1 (5) 式中, 为隐含层到输出层的权值
输出层神经元的输出为:
Ol {k) = g{netl {k)) (6)
式中,
exp(x)
g(x) = -(l + tanh(x))
2 exp(x) + exp(-x) (7)
输出层的神经元的输出对应着协调控制因子 Zcv, 决定 步骤二 神经网络参数的混合学习训练
在上述神经网络模型的基础上, 采用在线训练和离线训练相结合的方式进行, 经过预先 判断之后, 决定是进行基于 BP算法的离线训练, 还是进行基于遗传算法的在线训练。
在混合学习算法中, 改进之后的 BP算法是通过不断离线修正网络神经元之间的权值和 阀值, 使网络不断逼近真实的模型。 在线遗传学习训练算法, 是为了克服 BP离线学习训练 的一些局限性, 由于遗传学习训练算法是在寻优过程中应用高维可行解空间随机产生多个起 始点并同时开始搜索, 由适应度函数来指导搜索方向, 因而搜索区域广,搜索效率高。并且, 在实时控制时, 在每次进行在线调整权值参数之前, 要先判断温度差值是否过大, 如果超过 规定限额, 则还是要先进行离线的模型学习训练, 直到差值在允许范围以内之后, 再进行在 线的学习训练。 这样可以减少振荡, 以及缩短控制时间, 如图 15所示。
由于误差反向传播(Back Propagation, 简称 BP)算法具有简单、 易学、 收敛速度较快 等优点, 因而被广泛地用于网络权值的调整中。 但 BP算法在应用于实时控制时, 其学习速 度慢, 容易陷入局部极小的缺点, 有时还可能得不到全局最优。 这是由于当网络由一个训练 样本换成另一个训练样本时。 由于较大的初值误差容易引起权系数的过调从而加长调整时 间。 因此, 根据本发明的一个实施例, 引入了一个动量因子 α, 以减少过调量, 更有利于离 线的加速学习。 如图 16所示, 动态模型初始化之后, 从输入层经过隐层单元层层处理, 直 至输出层, 是输入和权值的函数, 求出隐含层和输出层各单元的输出, 求出期望输出与实际 输出的偏差, 对偏差值进行判断, 若实际输出与期望输出之间存在偏差则进入反向传播, 按 原正向传播途径反向回传, 计算隐层单元误差, 并按误差函数的负梯度方向进行求解, 之后 对各层神经元的权系数进行修正, 最终使期望的误差函数趋向最小。 动量因子"是一个确定 过去学习效果的加权因子, 通常0〈^^ 1, 所对应的另一变化系数77的范围为 0 < 77 < ()·5, 此 时的效果较为理想。 同时, 在每次进行迭代之后, 这两个因子都会不断调整, 修正。 考虑各 变量之间的耦合作用, 取性能指标函数为:
J= E (k) = (ZCT - ZJ + -X [r(^) - y(k)]
(8) 最后得权值的修正计算公式为:
(9)
W(,— 1)
通过上述离线学习调整完权值之后, 难免与实际参数有一定的差距, 为此, 随后利用遗 传算法在线实时修正所得到网络的权值参数, 搜索过程以原参数为基础, 在原参数的较小领 域内进行。
遗传算法是一种基于自然选择和群体遗传机制的搜索算法,它模拟了自然选择和自然遗 传过程中的繁殖, 交配和变异现象。 它将每个可能的解看作是群体中的一个个体, 并将每个 个体编码成字符串的形式, 根据预定的目标函数对每个个体进行评价, 给出一个适应度值。 这些个体的适应度值利用遗传算子对这些个体进行遗传操作, 保留优个体, 淘汰差个体, 使 最终的权值朝一个优良的状态发展。 在此, 我们应用传统的二进制编码方式, 当神经网络的 规模稍大时, 染色体个体的长度就会很大, 从而影响遗传算法的效率, 为此本智能系统采用 浮点数编码, 浮点数编码方法使用的是决策变量的真实值,个体的编码长度等于其决策变量 的个数。本系统中神经网络的权值变量共有 24个, 选用误差平方和的倒数作为适应度函数。 对于遗传操作来说, 遗传操作中的选择算法采用标准化几何排序的方法。该排序方法按拟和 值对个体进行排序, 根据个体的位置分配选择概率。标准化几何排序定义个体的选择概率公 式为-
Pt = ^ ^ ~~ (1-^Γ1
! -d - ^ (11) 其中: ^为最佳个体的选择概率, A"为个体的序列号, "为种群大小 针对浮点编码, 交叉算法采用数学交叉和启发式交叉两种相结合的方式, 两种交叉方法 结合使用可以增强算法的探测能力。为保持种群的多样性和防止早熟现象, 需要对原种群中 的基因加一随机扰动。 本系统中的变异操作采用多元非均匀变异 (multiNonUnJfMutafion) 策略产生变异基因来构成新的种群, 即分别对自变量在其解空间进行非均匀变异后, 再随机 取一组合作为变异结果。 原理式为:
X^ X^ iX - ^fig) ^ r,≥0.5 ( 12) 其中-
Figure imgf000016_0001
\ 为 [OJ]之间均匀随即变量, A , 6'分别是变量上、 下限, g为当前优化代 数, 为最大优化代数, 为衡量扰动程度的系统参数.
基于上述描述, 得出由遗传算法进行在线的神经网络权系数优化的训练步骤如下:
①给定神经网络的输入、 输出集.
②确定网络权系数的编码方式, 选定遗传操作, 设置遗传参数
③以设定的种群规模 N, 随即产生初始种群
④译码种群中每一个体位串, 求得 N组网络权系数, 得到具有相同结构的 N个网络
⑤由输入样本集经前向传播算法, 求得 N组网络权值对应的 N个网络输出
⑥设定网络的目标函数, 将其转换为适应度, 对 N个网络进行评价
⑦依据适应度在遗传空间进行选择操作.
⑧依据选定的交叉、 变异及有关算法、 参数进行相应的操作, 得到新一代种群
⑨返回步骤④, 直到满足性能要求, 得到一组优化的权系数.
步骤三 PI控制
得到协调控制因子之后, 将协调控制因子输入给 PI控制器, PI控制器将协调控制因子 与其参数进行进一步的运算与结合, 得出控制变量, 从而对被控机构进行协调有效的控制。
根据本发明的一个实施例, 循环冷却水流量的控制方程为:
Gw = KcpZcvel + K,
Figure imgf000016_0002
( 14) 式中:
G -: 冷却水的质量流量 温差 循环冷却水水流量控制器的比例系数
Κ": 循环冷却水水流量控制器的 R分系数
Kcd : 循环冷却水水流量控制器的微分系数
电加热器加热功率的控制方程为:
Figure imgf000017_0001
式中-
Q": 电加热器加热功率
K'p : 电加热器控制器的比例系数 κ": 电加热器控制器的积分系数 κ'" 电加热器控制器的微分系数 步骤四 限幅处理
得到控制变量之后, 难免会有一些不精确性, 因此, 还要对控制变量进行有效的限幅处 理, 使控制变量最优化。 这里应用 S型函数进行限幅的处理, S函数公式为:
1
λ =■
CT 1 + 。v (16)
由上述公式得限幅处理公式为:
/(")="2
f(u) = u* lcv ux≤u< u2
f (u) = u{ u < «,
(17)
式中-
"'· 控制变量^或 ;
"' , "2 : 限幅阀值; ("): 经过限幅处理后的控制变量
总之, 本发明的原理是: PI 控制器由于其设计简单、 易于实现、 可靠性高等优点,被 广泛应用于机电、 冶金、 机械、 化工等行业的过程控制和运动控制中,尤其适用于可建立精 确数学模型的确定性控制系统。 但是,实际工业过程存在很多不确定因素的干扰,使用经典
PI 控制器进行控制系统的校正往往达不到理想的控制效果。 而神经网络具有非线性映射, 自学习能力, 分布存储能力及处理信息等特点, 因此将神经网络与 PI控制器结合在一起, 但又不同于传统的结合。在此处的结合是建立在协调控制基础上的结合, 由于协调控制的不 确定性和复杂性, 因此我们利用神经网络的非线性, 自学习能力, 在输入信号进入 PI控制 器之前进行协调控制的一个预判断, 得出一个协调控制因子, 确定在所有协调控制的控制机 构中, 哪些控制机构是主要控制机构, 哪些是辅助控制机构, 这样使得控制系统不仅具备处 理不精确性, 不确定性的能力, 同时还具有了协调控制稳定性的能力。 同时又通过自学习能 力, 不断修正神经网络连接权值, 调整协调控制因子, 使该因子为某种最优协调控制调节下 参数值, 以期达到系统性能指标的要求。 将这个协调控制因子送入 PI控制器, 进行相应的 控制操作, 在 PI控制器输出信号输出之后, 还要进行相应的限幅处理, 对控制变量进一步 做出精确化, 最终达到砂尘环境试验风洞的温度智能控制的协调性和可靠性。减少了砂尘温 度调节过程中的波动幅度。
基于神经网络 PI的温度智能控制系统的神经网络结构是由一个单隐含层的三层网络构 成。 隐含层单元的变换函数采用径向基函数中的 函数, 它是一种局部分布的对中心 点径向对称衰减的非负非线性函数。 此神经网络由输入空间到隐含层空间的映射是非线性 的, 而从隐含层空间到输出层空间的隐射是线性的, 这样的神经网络实现了加快学习速度的 特点, 以及避免了一定的振荡性和局部极小值问题。
需要说明的是, 学习训练后的神经网络能够很好的调整参数, 高精度的逼近温度智能控 制的输入输出信号的非线性函数,且具有很强的泛化能力。学习训练的过程为从初始化之后, 进入判断模块, 判别温度差值是否超过了一个限定值, 如果超过了这个限定值, 则进行离线 动态温度模型学习, 这样可以进行在线学习之前, 使误差量缩小, 缩短在线学习的时间。 离 线模型学习调整参数实质上是利用改进之后的 BP算法实现权值的离线全局寻优。 之后继续 进行判断, 直到这个差值在限定值范围内之后, 再进行在线学习训练阶段, 在线学习调整权 值参数主要是通过遗传算法, 在高维可行解空间随机产生多个起始点并同时幵始搜索, 由适 应度函数来指导搜索方向, 最终获得最优权值参数。 根据本发明的一个实施例, 定义离线学习时, 采用专家系统的动态模型-
Mtct = khnf 3 + Gaca α -0,)-ksFs* ψ, es) + Qd-Gwcw(0l +A0wo-ewi)
dt (18)
Μ ': 循环风道内的当量金属质量 °': 平均比热容 循环风道内的风机转速 k": 比例系数 k". 循环风道外表面的散热系数 F- 散热面积 , θ", °': 环境温度 冷却水的质量流量 ^, °": 比热 ^': 进口温度
L. 冷却水的出口端差 °": 循环风道辅助气流质量流』
Q . 电加热器的加热功率
本发明能够实现沙尘环境试验装置系统的温度控制, 以满足环境试验所需的温度条件, 不仅结构简单, 实现一体化, 而且能够快速高效的控制温度, 可用于结构经凑、 清理方便、 频繁试验的场合。

Claims

权 利 要 求 书
1、 一种砂尘环境试验系统的温度调节装置, 该砂尘环境试验系统主要包括- 循环风道, 为不规则结构的密闭风道, 用于提供吹砂和吹尘环境的场所; 该循环风道包 括有第二收缩段、 试验段及分离段等,
主风机, 用于驱动所述循环风道内气流的流动;
u型分离器, 设置于分离段, 用于对试验后的沙尘进行分离回收; 及
导流叶片, 分四组分别设置在循环风道的四个拐角处;
其特征在于: 该温度调节装置包括:
在所述的 u型分离器内部铺设热交换管;
所述四组导流叶片, 在其中的三组导流叶片中分别设置有热交换工质通孔; 另外一组导 流叶片中设置有加热装置;
在所述的第二收缩段设置有电加热器。
2、 根据权利要求 1所述砂尘环境试验系统的温度调节装置, 其特征在于: 所述的温度 调节装置进一步包括:
循环冷却水水源;
冷水机组; 及
电锅炉;
所述的循环冷却水水源、冷水机组及电锅炉分别通过管道与具有热交换功能的导流叶片 及 U型分离器相连; 在各管道上设置有用于调节冷热工质流量的阀门。
3、 根据权利要求 1所述砂尘环境试验系统的温度调节装置, 其特征在于: 所述的导流 叶片中的加热装置为一电加热器, 该电加热器外依次包覆有绝缘层及耐磨保护层。
4、 一种砂尘环境试验系统的智能温度控制方法, 该方法是通过基于神经网络的 PI智能 温度控制系统实现, 其特征在于: 该智能温度控制系统包括:
风速感应器, 安装于循环风道内,
温度感应器, 安装于循环风道内,
神经网络控制器, 该神经网络控制器具体包括一输入层, 一个隐含层和一输出层; 该神 经网络控制器的作用是得出一个协调控制因子, 以决定哪种控制设备是主要控制设备, 哪种 控制装备是辅助控制设备。
PI控制器, 用于接收来自神经网络控制器的输出 协调控制因子以及来自温度感应器所 测得的温度值, 并将经过处理的控制量传递给沙尘环境试验系统的温度控制装置。
5、 根据权利要求 4所述一种砂尘环境试验系统的智能温度控制方法, 该方法是通过基 于神经网络的 PI智能温度控制系统实现, 其特征在于: 所述的 PI控制器包括:
第一 PI控制器, 用于根据所述温度值和所述协调控制因子, 产生第一粗略控制量, 第二 PI控制器, 用于根据所述温度值和所述协调控制因子, 产生第二粗略控制量。
6、 根据权利要求 4所述一种砂尘环境试验系统的智能温度控制方法, 该方法是通过基 于神经网络的 PI智能温度控制系统实现, 其特征在于: 所述的智能温度控制系统, 进一步 包括:
第一限幅器, 用于对所述第一粗略控制量进行限幅和优化处理, 从而产生用于电 锅炉回水控制阀的优化精细控制量,
第二限幅器, 用于对所述第二粗略控制量进行限幅和优化处理, 从而产生用于循 环冷却水出水阀的优化精细控制量。
7、 一种砂尘环境试验系统的智能温度控制方法, 该方法包括如下步骤:
步骤一、 建立神经网络系统结构
本发明采用单隐层的三层前馈网络构成; 隐含层单元的变换函数采用正负对称的 S gww^函数, 它是一种局部分布的对中心点径向对称衰减的非负非线性函数, 此神经网络 由输入空间到隐含层空间的映射是非线性的, 而从隐含层空间到输出层空间的隐射是线性 的, 这样的神经网络实现了加快学习速度的特点, 以及避免了一定的振荡性和局部极小值问 题, 该三层前馈网络包括:
1 ) 输入层
输入层采用了特殊的 3个输入, 分别对应输入风速值 v, 通过安装在循环风道内的风速 感应器获得; 温度的误差 ec, 通过安装在循环风道内的温度感应器获得温度值之后, 进行减 法运算获得; 常量 1 , 常量在这里起一个干扰的作用, 则输入模式向量为 x = [v,e£,l], 比起 2 个的输入向量的结构 X = [v,ej更符合实际的工作环境;
输入层神经元的输入输出函数为 -
式中, 为输入层神经元的个数, = 123
2 ) 隐含层 ;
其神经元的输入为:
3
netJ(k) = j wJiOi
(2) 式中, ^'为输入层到隐含层的权值, 且所述隐含层神经元的输出为:
Figure imgf000022_0001
式中, 隐含层神经元的个数, 为隐含层的激活函数, 且所述隐含层的激活函数取正负对 称的 S g w'i 函数: exp( ) - exp(-x)
/ (JC) = tanh( ) =
Figure imgf000022_0002
3 ) 输出层 - 所述输出层的神经元的输入为: net,( = tWijO k、
J—-1 (5) 式中, 为隐含层到输出层的权值, 且所述输出层神经元的输出为-
Ol (k) = g(netl (k)) (6) 式中,
Figure imgf000022_0003
所述输出层的神经元的输出对应着协调控制因子 Zcv , 该协调控制因子决定哪种设备是主 要控制设备, 哪种是辅助控制设备; 步骤二、 神经网络参数的混合学习训练
在所述步骤一得到的神经网络模型的基础上,釆用在线训练和离线训练相结合的方式进 行, 经过预先判断之后, 决定是进行基于 BP算法的离线训练, 还是进行基于遗传算法的在 线训练; 步骤三、 PI控制
由所述步骤一和二最终得到协调控制因子之后, 将协调控制因子输入给 PI控制器, PI 控制器将协调控制因子与其参数进行进一步的运算与结合, 最终得出控制变量, 对被控机构 进行控制; 步骤四、 限幅处理
通过步骤三得到控制变量之后, 应用 S型函数进行限幅处理, 使控制变量最优化。
8、 根据权利要求 7所述的砂尘环境试验系统的智能温度控制方法, 其特征在于: 所述 步骤二中, 进行神经网络参数的混合学习训练, 要预先进行判断模块, 当温度差值好过规 定界限值值时, 进行基于 BP学习算法的离线动态温度模型学习训练, 实现权值的离线全局 寻优,使误差量缩小, 而缩短在线学习所需的时间; 相反, 当在界定值范围以内时, 利用遗 传算法进行在线调整, 使权值最优化。
9、 根据权利要求 7所述的环境模拟系统的智能温度控制方法, 其特征在于: 利用所述 神经网络控制器, 根据所述风速值,.得出一个协调控制因子; 根据该协调控制因子把至少两 个温度控制机构中的一个确定为主要控制机构,把至少两个温度控制机构中其余的控制机构 确定为辅助控制机构, 利用所述 PI控制器, 根据所述温度值和所述协调控制因子而得到所 述粗略控制量。
PCT/CN2010/000720 2009-07-15 2010-05-20 砂尘环境试验系统的温度调节装置及智能温度控制方法 WO2011006344A1 (zh)

Priority Applications (1)

Application Number Priority Date Filing Date Title
US13/318,327 US20120048953A1 (en) 2009-07-15 2010-05-20 Temperature adjusting device and an intelligent temperature control method for a sand and dust environment testing system

Applications Claiming Priority (4)

Application Number Priority Date Filing Date Title
CN200910089289.8 2009-07-15
CN2009100892898A CN101639397B (zh) 2009-07-15 2009-07-15 一体式吹砂吹尘环境模拟系统的温度调节系统和方法
CN201010034360.5A CN102129259B (zh) 2010-01-20 2010-01-20 用于砂尘环境试验风洞的基于神经网络pi的智能温度控制系统和方法
CN201010034360.5 2010-01-20

Publications (1)

Publication Number Publication Date
WO2011006344A1 true WO2011006344A1 (zh) 2011-01-20

Family

ID=43448899

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2010/000720 WO2011006344A1 (zh) 2009-07-15 2010-05-20 砂尘环境试验系统的温度调节装置及智能温度控制方法

Country Status (2)

Country Link
US (1) US20120048953A1 (zh)
WO (1) WO2011006344A1 (zh)

Cited By (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104729823A (zh) * 2015-04-09 2015-06-24 中国科学院电工研究所 用于建筑物和构筑物风沙实验的风沙两相流实验装置
CN107562035A (zh) * 2017-07-24 2018-01-09 芜湖赛宝信息产业技术研究院有限公司 一种新能源汽车电机及控制器沙尘试验方法
CN109186926A (zh) * 2018-11-19 2019-01-11 温州大学瓯江学院 一种模拟沙尘暴的风洞
CN110046412A (zh) * 2019-04-02 2019-07-23 合肥工业大学 一种基于优化bp神经网络的圆光栅测角误差修正方法
CN112084709A (zh) * 2020-09-04 2020-12-15 西安交通大学 基于遗传算法和径向基神经网络的大型发电机绝缘状态评估方法
CN112945511A (zh) * 2021-01-29 2021-06-11 西安科技大学 一种风洞模拟及环境适应性试验系统及方法
CN114323537A (zh) * 2022-01-06 2022-04-12 东北农业大学 一种风力机沙尘实验流动控制系统及方法
CN114623649A (zh) * 2022-05-17 2022-06-14 中国空气动力研究与发展中心高速空气动力研究所 一种连续式风洞气流温度冷却系统
CN114679881A (zh) * 2022-01-05 2022-06-28 南京航空航天大学 一种多单元均温的冷量分配智能控制电子散热系统

Families Citing this family (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101940198B1 (ko) 2010-01-15 2019-01-18 스카이벤쳐 인터내셔널 (유케이) 엘티디 풍동 터닝 베인 열 교환기
US9453779B2 (en) * 2013-09-06 2016-09-27 Tai-Her Yang Semi-opened hydrodynamic testing device having reverse pumping return unit
CN104714409B (zh) * 2014-12-31 2017-07-11 重庆科技学院 滞后校正器结构参数的获取方法及装置
US10387775B2 (en) * 2015-09-09 2019-08-20 Emerson Process Management Power & Water Solutions, Inc. Model-based characterization of pressure/load relationship for power plant load control
JP6516294B2 (ja) * 2015-11-24 2019-05-22 株式会社日立プラントメカニクス 環境風洞試験装置
CN108803570A (zh) * 2018-06-21 2018-11-13 中国民航大学 基于信号转接技术的飞机温度控制系统故障测试系统
CN109582057B (zh) * 2018-11-09 2021-10-01 中国航天空气动力技术研究院 一种环境风洞温度控制系统
CN109946098A (zh) * 2019-02-14 2019-06-28 江苏科技大学 一种闭式带中间冷媒的结霜工况下表面冷却器性能试验台
CN109884885A (zh) * 2019-02-25 2019-06-14 常州兰陵自动化设备有限公司 一种直流电动装置控制系统及其构建方法
CN109779736B (zh) * 2019-03-20 2023-08-11 山东交通学院 一种发动机电控硅油风扇的节能优化控制方法及系统
SE1950948A1 (en) * 2019-08-20 2021-02-21 Swegon Operations Ab Fan for Air Handling Unit (AHU)
CN112332686B (zh) * 2020-10-26 2021-08-27 湖南大学 一种含有电流限幅器的下垂逆变器能量函数构造的方法
CN112613603B (zh) * 2020-12-25 2023-04-07 西安交通大学 基于限幅器的神经网络训练方法及其应用
CN113095398B (zh) * 2021-04-08 2022-07-12 西南石油大学 一种基于遗传算法优化的bp神经网络的压裂数据清洗方法
CN114878639A (zh) * 2022-05-17 2022-08-09 重庆文理学院 一种基于神经网络的汽车尾气浓度检测系统

Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4751844A (en) * 1987-02-03 1988-06-21 Kabushiki-Kaisha Toyo Seisakusho Adjustable constant temperature and humidity control device for self-propelled vehicles
US4799390A (en) * 1987-03-11 1989-01-24 Kabushiki-Kaisha Toyo Seisakusho Snow-weathering test apparatus for self-propelled vehicle
CN1621797A (zh) * 2005-01-07 2005-06-01 北京航空航天大学 吹砂吹尘环境试验装置系统
CN101101239A (zh) * 2006-07-03 2008-01-09 珠海格力电器股份有限公司 用于测试空调的设备及方法
CN101140450A (zh) * 2006-09-08 2008-03-12 香港中文大学精密工程研究所 节能型热舒适控制器及控制方法
CN201075100Y (zh) * 2007-07-29 2008-06-18 朱建忠 程控模拟沙尘试验装置
CN201129823Y (zh) * 2007-11-21 2008-10-08 厦门立思科技有限公司 基于人工神经网络技术的中央空调节能控制装置
CN101392939A (zh) * 2008-11-18 2009-03-25 天津大学 一种建筑物独立供能温度非线性预测控制方法
CN101398343A (zh) * 2007-09-26 2009-04-01 北京师范大学 风沙环境与工程风洞模拟装置
CN201237396Y (zh) * 2008-05-30 2009-05-13 珠海格力电器股份有限公司 用于邻室环境模拟室的工况调节系统
CN101639397A (zh) * 2009-07-15 2010-02-03 北京航空航天大学 一体式吹砂吹尘环境模拟系统的温度调节系统和方法

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPS5845041B2 (ja) * 1973-04-20 1983-10-06 株式会社日立製作所 プラントの最適化制御方法
CH566086A5 (zh) * 1973-05-18 1975-08-29 Zellweger Uster Ag
WO2004084371A1 (ja) * 1998-08-31 2004-09-30 Kaoru Fujita 発電プラントの最適化制御方法及び最適化制御装置
US7117045B2 (en) * 2001-09-08 2006-10-03 Colorado State University Research Foundation Combined proportional plus integral (PI) and neural network (nN) controller

Patent Citations (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US4751844A (en) * 1987-02-03 1988-06-21 Kabushiki-Kaisha Toyo Seisakusho Adjustable constant temperature and humidity control device for self-propelled vehicles
US4799390A (en) * 1987-03-11 1989-01-24 Kabushiki-Kaisha Toyo Seisakusho Snow-weathering test apparatus for self-propelled vehicle
CN1621797A (zh) * 2005-01-07 2005-06-01 北京航空航天大学 吹砂吹尘环境试验装置系统
CN101101239A (zh) * 2006-07-03 2008-01-09 珠海格力电器股份有限公司 用于测试空调的设备及方法
CN101140450A (zh) * 2006-09-08 2008-03-12 香港中文大学精密工程研究所 节能型热舒适控制器及控制方法
CN201075100Y (zh) * 2007-07-29 2008-06-18 朱建忠 程控模拟沙尘试验装置
CN101398343A (zh) * 2007-09-26 2009-04-01 北京师范大学 风沙环境与工程风洞模拟装置
CN201129823Y (zh) * 2007-11-21 2008-10-08 厦门立思科技有限公司 基于人工神经网络技术的中央空调节能控制装置
CN201237396Y (zh) * 2008-05-30 2009-05-13 珠海格力电器股份有限公司 用于邻室环境模拟室的工况调节系统
CN101392939A (zh) * 2008-11-18 2009-03-25 天津大学 一种建筑物独立供能温度非线性预测控制方法
CN101639397A (zh) * 2009-07-15 2010-02-03 北京航空航天大学 一体式吹砂吹尘环境模拟系统的温度调节系统和方法

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104729823B (zh) * 2015-04-09 2017-06-30 中国科学院电工研究所 用于建筑物和构筑物风沙实验的风沙两相流实验装置
CN104729823A (zh) * 2015-04-09 2015-06-24 中国科学院电工研究所 用于建筑物和构筑物风沙实验的风沙两相流实验装置
CN107562035A (zh) * 2017-07-24 2018-01-09 芜湖赛宝信息产业技术研究院有限公司 一种新能源汽车电机及控制器沙尘试验方法
CN109186926A (zh) * 2018-11-19 2019-01-11 温州大学瓯江学院 一种模拟沙尘暴的风洞
CN110046412B (zh) * 2019-04-02 2022-10-25 合肥工业大学 一种基于优化bp神经网络的圆光栅测角误差修正方法
CN110046412A (zh) * 2019-04-02 2019-07-23 合肥工业大学 一种基于优化bp神经网络的圆光栅测角误差修正方法
CN112084709A (zh) * 2020-09-04 2020-12-15 西安交通大学 基于遗传算法和径向基神经网络的大型发电机绝缘状态评估方法
CN112084709B (zh) * 2020-09-04 2024-04-05 西安交通大学 基于遗传算法和径向基神经网络的大型发电机绝缘状态评估方法
CN112945511A (zh) * 2021-01-29 2021-06-11 西安科技大学 一种风洞模拟及环境适应性试验系统及方法
CN112945511B (zh) * 2021-01-29 2023-09-22 西安科技大学 一种风洞模拟及环境适应性试验系统及方法
CN114679881A (zh) * 2022-01-05 2022-06-28 南京航空航天大学 一种多单元均温的冷量分配智能控制电子散热系统
CN114323537B (zh) * 2022-01-06 2024-04-02 东北农业大学 一种风力机沙尘实验流动控制系统及方法
CN114323537A (zh) * 2022-01-06 2022-04-12 东北农业大学 一种风力机沙尘实验流动控制系统及方法
CN114623649A (zh) * 2022-05-17 2022-06-14 中国空气动力研究与发展中心高速空气动力研究所 一种连续式风洞气流温度冷却系统

Also Published As

Publication number Publication date
US20120048953A1 (en) 2012-03-01

Similar Documents

Publication Publication Date Title
WO2011006344A1 (zh) 砂尘环境试验系统的温度调节装置及智能温度控制方法
Zhang et al. Optimal design of building environment with hybrid genetic algorithm, artificial neural network, multivariate regression analysis and fuzzy logic controller
CN111815042B (zh) 计及精细化热网模型的电热综合能源系统优化调度方法
CN105571343B (zh) 空冷发电机组汽轮机运行背压连续优化控制方法及系统
Ke et al. Thermal-hydraulic performance and optimization of attack angle of delta winglets in plain and wavy finned-tube heat exchangers
Olabi et al. Application of artificial intelligence for prediction, optimization, and control of thermal energy storage systems
US20050192680A1 (en) System and method for optimizing global set points in a building environmental management system
CN111881505B (zh) 一种基于ga-rbf算法的既有建筑多目标优化改造决策方法
Huang et al. A new zone temperature predictive modeling for energy saving in buildings
CN107023966B (zh) 一种地铁站空调冷却水出水温度设定值优化方法
CN109764570B (zh) 一种基于神经元网络的跨临界二氧化碳热泵系统排气压力的控制方法
CN113255198B (zh) 一种含虚拟储能的冷热电联供微电网多目标优化方法
KR102148726B1 (ko) 공조 시스템의 이코노마이저 제어 방법 및 이를 수행하는 공조 시스템
CN116398994B (zh) 一种基于负荷预测的冷水机组群控优化方法
Liu et al. Optimization of intelligent heating ventilation air conditioning system in urban building based on BIM and artificial intelligence technology
CN109857177B (zh) 一种建筑电气节能监测方法
CN116717839A (zh) 供热控制方法、控制装置和供热系统
CN115691986A (zh) 一种变压器辅助风冷、水冷智能散热系统的设计方法
CN114251716B (zh) 一种供热管网调节阀门的调节参数确定方法和系统
Mei et al. Energy efficiency prediction of screw chillers on BP neural network optimized by improved genetic algorithm
CN115743504A (zh) 一种海洋平台多舱室通风多目标优化控制方法
Tan et al. Modeling and simulation of layered water tank based on MATLAB/SIMULINK
Fong et al. A robust evolutionary algorithm for HVAC engineering optimization
Zheng Research on energy-saving control and optimisation of air conditioning system based on genetic algorithm
Banjar et al. Design of fuzzy controller for a hybrid active/passive cooling system in smart homes with a windcatcher

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 10799342

Country of ref document: EP

Kind code of ref document: A1

WWE Wipo information: entry into national phase

Ref document number: 13318327

Country of ref document: US

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 10799342

Country of ref document: EP

Kind code of ref document: A1