With the development of science and technology, the birth of scroll compressors, the improvement of frequency conversion technology and the application of fuzzy control technology have made air conditioners gradually enter the era of electronic control. The characteristic of the variable frequency control air conditioner is that the compressor can automatically adjust the amount of cooling (heat) according to the amount of cold (heat) required in the room, which has many advantages and is being favored by the majority of air conditioner users. In general applications of air conditioners with ON/OFF control, due to the structural characteristics of the air conditioner itself, the output of the indoor unit tends to lag behind the output of the compressor, while the hysteresis of the indoor air parameters is greater, which affects the temperature control of the air conditioner. Increased precision and comfort. In addition, the actual operating conditions of the air conditioner are ever-changing. Most of the time, the air conditioner is not working in an optimal state, and the inverter control technology is the best solution to solve the above problems. Since the air conditioner is a complex control object with large lag, nonlinearity and time-varying, it is difficult to obtain good control quality by using the traditional control strategy. The artificial neural network provides a simple solution to such problems with its large-capacity parallel processing, distributed storage, intelligent fuzzy classification, and the ability to self-organize, self-learn and process the intrinsic and difficult to analyze the regularity of the expression. An effective way. This paper introduces a control method based on BP neural network model. It can quickly and accurately synthesize and analyze the data obtained from the actual environment and get the correct conclusion. Therefore, the compressor, the fan and the electronic expansion valve are adjusted by the control unit to react quickly according to the status quo, and the effect of intelligent control is achieved. 1Inverter air conditioner control technology The so-called variable frequency control air conditioner compares the actual environmental state measured by the sensor with the state of the air conditioner system and the set state that people expect to achieve. The fuzzy logic control technology enables the air conditioner control system to have a self-adjusting function. The characteristics, so as to control the variable frequency power supply of the air conditioner and each execution unit, so that the working state of the air conditioner automatically changes with the change of people's requirements and the change of the environmental state, quickly and accurately meet the requirements of the people, and the air conditioner The working state is kept in the most reasonable state. 1 msp430 microcontroller as the control core TI's msp430 series of microcontrollers is an ultra-low power mixed-signal controller, including a series of devices. The 16 registers and constant generators in the CPU enable the MSP430 microcontroller to achieve the highest code efficiency; a flexible clock source allows the device to achieve the lowest power consumption; and a digitally controlled oscillator allows the device to quickly wake up from a low-power mode Activate to an active mode of operation in less than 6μs. At the same time, it has the following characteristics: low voltage, ultra-low power consumption; powerful processing capability, which can produce high-efficiency source program; stable system operation; rich on-chip peripherals, which can integrate rich on-chip peripherals; Conversion rate; convenient and efficient development environment, development language is assembly language and C language. The msp430 MCU mainly completes the following functions: comprehensive indoor unit refrigeration data for variable frequency speed regulation of the compressor; respectively controls each electronic expansion valve according to the data transmitted in each room; monitors the outdoor ambient temperature, the condenser temperature, whether the compressor is normal, and determines the whole machine Working state, control compressor, electronic expansion valve. 1 2 Inverter air conditioner control unit Inverter air conditioner control unit can be divided into three relatively independent parts: remote control transmitter, indoor control unit and outdoor control unit. A Bluetooth wireless communication module is used for communication between the remote control transmitter and the indoor control unit, and the indoor control unit and the outdoor control unit perform two-way communication to exchange information. The indoor control unit controls the operation of the air conditioner according to the command issued by the remote control transmitter and the state of the system, calculates the operating frequency of the compressor, and controls the opening and closing of the outdoor power supply. The outdoor control unit controls the operation of the outdoor part of the air conditioner according to the instruction of the indoor control unit, and can perform the special state operation independently. The schematic is as follows: (1) The remote control adopts 4-bit single-chip microcomputer with keyboard input, infrared light-emitting diode and LCD display. The chip has LCD control drive circuit, which can directly drive the LCD display. (2) Indoor control unit: The indoor control unit is composed of a main control board, an L ED display board and a wireless receiving module. The main functions of the indoor control unit: measuring the indoor return air temperature and the evaporator outlet temperature and the evaporator inlet temperature; receiving and processing the signal from the remote control transmitter, combined with the state control state of the system; controlling the operation and speed of the indoor fan; Control the swing and position of the air deflector; calculate the operating frequency of the compressor according to the indoor temperature, target temperature and its changes; communicate with the outdoor control unit; control the power supply of the outdoor part; display the indoor temperature, target temperature and system operating status with L ED , fault information, etc. The core of the indoor control unit is the operational control and frequency calculation of the compressor. In this controller, the control method based on BP neural network is adopted to achieve the effects of fast cooling (heating) speed, small overshoot and accurate and stable temperature control. After reaching a steady state, the room temperature can be controlled within the target temperature range of ±0 15 °C. (3) Outdoor control unit: The outdoor control unit consists of the main circuit of the inverter and the control board. The main functions of the outdoor control unit: detecting the condenser outlet temperature and the compressor top cover temperature; communicating with the indoor control unit, receiving and analyzing the commands of the indoor control unit; controlling the outdoor part of the air conditioner according to the commands of the indoor control unit and the respective detection parameters Operating state; controlling the operation of the outdoor fan; controlling the action of the reversing valve; controlling the speed of the compressor and starting and stopping; performing the defrosting operation; performing the protection operation. The core part of the outdoor control unit is the drive control of the compressor. The outdoor control unit controls the inverter to generate an SPWM output at a modulation frequency of 3 1 6 kHz to drive the compressor. The outdoor control unit can perform voltage compensation according to the fluctuation of the DC side voltage, so that the operation of the system can not only adapt to the change of the load, but also adapt to the fluctuation of the grid voltage. While receiving the instruction of the indoor control unit, the outdoor control unit also independently analyzes the operating status of the outdoor part to determine whether to enter the defrosting operation or the protection operation. 2 BP neural network In this paper, the error back propagation (BP) algorithm in the neural network is used to control the compressor of the air conditioner. The basic idea of ​​the BP algorithm is that the learning process consists of two processes: the forward propagation of the signal and the back propagation of the error. In the case of forward propagation, the input samples are passed in from the input layer, processed layer by layer through each hidden layer, and then transmitted to the output layer. If the actual output of the output layer does not match the expected output (teacher signal), it goes to the backpropagation phase of the error. Error back propagation is to pass the output error back to the input layer layer by layer through the hidden layer, and distribute the error to all the units of each layer, so as to obtain the error signal of each layer unit, and the error signal and the correction unit The basis of the weight. The process of adjusting the weight of each layer of the signal-like propagation and error back-propagation is repeated. The process of continuously adjusting the weight is the learning and training process of the network. This process continues until the error in the network output is reduced to an acceptable level or to a predetermined number of learning times. In the application of the multi-layer feedforward network of the BP algorithm, the most common application is the single hidden layer network. A single hidden layer feedforward network is also called a three-layer feedforward network, including an input layer, a hidden layer, and an output layer. As shown. In a three-layer feedforward network, the input vector is X = ( x1 , x2... Xn)T, which can introduce thresholds for hidden layer neurons; the hidden layer output vector is Y = ( y1 , y2... Ym)T, which can introduce thresholds for the output layer neurons; the output layer output vector is O = ( o1 , o2... Ol)T; the expected output vector is d = ( d1 , d2... Dl)T. The weight matrix of the input layer to the hidden layer is represented by V V = ( v1 , v2... Vt... Vm)T, where the column vector vt is the weight vector corresponding to the tth neuron of the hidden layer; the weight matrix of the hidden layer to the output layer is W = ( w1 , w2... Wt... Wl)T, where the column vector vt is the weight vector corresponding to the tth neuron of the hidden layer. The following is a mathematical analysis of the relationship between the signals of the layers: for the input layer O k = f ( net k) k = 1 , 2... l net k =∑w jk yj k = 1 , 2... l , j∈<0 , m> for the hidden layer yj = f ( net j) j = 1 , 2... m net j =∑w ij xi j = 1 , 2... m , i∈<0 , m> In the above two equations, the transfer function f ( x) uses the unipolar Sigmoid function f ( x) = 1 - e? x / 1 + e? x Because f(x) has continuous and conductive characteristics, and has f'(x) = f(x) <1 - f(x) > 3 programmed air conditioner control process, room leakage, people, things The number of rooms, the number of door opening and closing, etc. are all uncertain factors. The refrigeration capacity is also nonlinearly related to the compressor speed. Therefore, the fuzzy control is more accurate than the traditional control, the transition process is excellent, and the comfort is high. The control system is as shown. The error E of the two is calculated based on the temperature value measured by the temperature sensor and the demand thermometer input by the user, and the error change rate d E/dt is calculated. These two sets of data are converted into input ranges as input values ​​for the network. Learning by BP The law trains the network, and the training process of the network is done independently on the computer. The input value is passed through the network to obtain a corresponding output value, and then the output value is converted into the range of the control domain, and the compressor is controlled in real time. 3 The basic programming steps of the 1BP learning algorithm are as follows. (1) Initialization "A random number is assigned to the weight matrix V, W, where V is a matrix of 2 by 2, and W is a matrix of 2 by 1. The sample mode counter p and the training number counter q are set to 1, and the error E is set. 0, the learning rate n is set to 0 - l decimal, the precision E min achieved after network training is set to a positive decimal; (2) input training sample pairs: the sample is given by experts and technicians based on experimental experience Data. How to select and process the sample is described in detail below. Calculate the output of each layer, assign the values ​​of the vector array X and d according to the sample, and calculate the components in Y and O; (3) Calculate the network output error: Set the common P For training samples, the network has different errors for the different samples E < p> , with the largest one E max representing the total error of the network; (4) calculating the error signals of each layer: G < k> , H < k>; (5) Adjust the weights of each layer to calculate the components in W and V; (6) Check whether all the samples are rotated once. If p < P - 1, the counters p and q increase by 1, return to step (2), otherwise turn Step (7); (7) Check if the total network error reaches the accuracy It is required that if E < E min, the training ends, otherwise E is set to 0, p is set to 0, and step (2) is returned. 3 1 2 Sample selection and processing Since the input and output data of the BP network are required to be within the < 0 , 1 > interval, the input data needs to be scale-converted and transformed into the interval. In this way, the network training can give the input components an equally important position from the beginning; after the transformation, the neuron output can be prevented from being saturated due to the excessive absolute value of the net input, and then the weight is adjusted into the flat region of the error surface; The training value adjusts the weight of the total error of the output, so that the relative error of the output component with small fraction of the total error is large, and the problem that the absolute error of the large-valued output component is large and the absolute error of the output component of the small value is small is solved. The typical response curve for a refrigeration air conditioning temperature control system is shown. We will summarize the requirements for refrigeration control at various stages of the dynamic process of the system by analyzing this response curve. AB stage: This is the lagging stage of the system. In order to obtain a faster descent speed, the cooling capacity should be increased, and the characteristics of the system are: E < 0, d E/ dt = 0. BC phase: This is the startup phase of the system. When the output starts to track, the cooling capacity should be increased to make the output have a shorter response time. When the output is close to the steady state value, the cooling capacity should be gradually reduced to reduce the overshoot. At this time, the system is characterized by E < 0 and d E/ dt > 0. CD stage: It is the stage in which the system develops toward the direction of deviation increase. The control should be strengthened to minimize the overshoot and make the output return to the steady state value as much as possible. The characteristics of the system at this time are E > 0 and d E/ dt > 0. DE: Stage: It is the stage where the system reduces the deviation to the steady state. It is necessary to appropriately reduce the control and avoid the callback. At this time, the characteristics of the system are: E < 0, d E/ dt < 0. EF phase: The system changes in the opposite direction of the steady state. At this time, it should be properly controlled to make the system return to the given value as much as possible. The characteristics of the system at this time are E > 0 and d E/ dt < 0. FG stage: The system presents a trend of gradually decreasing error. To be stable as soon as possible, the control effect should not be too strong, otherwise there will be another overshoot. At this time, the system is characterized by E < 0 and d E/ dt > 0. It can be seen that in different response stages, the system has different requirements for control strength. These need to be reflected in the training samples during the training process of the network. The suitability of training samples directly affects the efficiency of the network, so it is important to pay attention to the selection and processing of samples. 4 Conclusion This paper introduces a neural network based inverter air conditioner control system. The control system applies the current international advanced frequency conversion technology and neural network theory to the air conditioning field, and solves many problems such as large power consumption, unstable temperature control, and slow cooling (heat) speed in the ordinary air conditioner. Representing a trend in the future development of the air conditioning industry. It operates at three different levels of adjustments so you will find it great for your car as you can select the required temperature. The car seat also has an ergonomic design, which provides great com 24V Heated Pads,24V Seat Heater,Heated Pads Seat Heated,Seat Covers For Heated Seats JiLin Province Debang Auto Electric Co.,Ltd. , https://www.dbheatedpads.com
fy.
The use of inverter refrigeration equipment in the network