Amplitude modulation transmitter position sensor temperature compensation method, apparatus, device, and medium

Abstract

The application provides an AMT position sensor temperature compensation method, device, equipment and medium. The method comprises the following steps: acquiring temperature information, detection position information and actual position information of an AMT actuator under different temperatures and different working displacements, wherein the detection position information is obtained by an AMT position sensor; performing filtering processing on the detection position information to obtain filtered detection position information; taking a chromosome determined by a genetic algorithm as a first initial weight and a first initial threshold value of a neural network; taking the filtered detection position information and the temperature information as inputs and taking the actual position information as an output, and inputting the neural network for training to obtain a temperature compensation model. The method of the application can enhance the accuracy of position acquisition by performing temperature compensation on the sensor measurement results.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicles, and particularly relates to an AMT position sensor temperature compensation method, device, equipment and medium. BACKGROUND

[0002] An automatic transmission (AMT) relies on a position sensor to obtain the current position of an actuator and controls the actuator to reach a target position to complete gear shifting.

[0003] Therefore, obtaining an accurate position of the AMT actuator is very important to ensure that the gear shifting is accurately completed. The existing processing method is to establish a bilinear interpolation algorithm module based on the steady-state voltage, pulse width and actual position of the sensor. The bilinear interpolation module uses the actual steady-state voltage and pulse width signal value generated by the position sensor to obtain the specific position of the AMT actuator. Such a processing method requires the addition of extra hardware circuits, which increases the circuit volume, and does not fully consider the influence of temperature on the AMT actuator, resulting in the obtained position not being accurate enough.

[0004] Therefore, it is urgent to propose a more optimized AMT actuator position acquisition method. SUMMARY

[0005] The present application provides an AMT position sensor temperature compensation method, device, equipment and medium to solve the problem of too many circuit structures and inaccurate position acquisition in the prior art.

[0006] In a first aspect, the present application provides an AMT position sensor temperature compensation method, comprising:

[0007] Obtaining temperature information, detection position information and actual position information of an AMT actuator under different temperatures and different working displacements, wherein the detection position information is obtained by an AMT position sensor;

[0008] Filtering the detection position information to obtain filtered detection position information;

[0009] Determining a chromosome by a genetic algorithm as a first initial weight and a first initial threshold of a neural network;

[0010] Inputting the filtered detection position information and the temperature information as inputs and the actual position information as an output, and inputting the neural network for training to obtain a temperature compensation model.

[0011] In a possible implementation, the obtaining of the temperature information, the detection position information and the actual position information of the AMT actuator under different temperatures and different working displacements comprises:

[0012] The AMT actuator is set with an initial temperature, and an initial temperature value of the AMT actuator is obtained by a temperature sensor, an initial detection position of the AMT actuator under different working displacements is obtained by an AMT position sensor, and an initial actual position of the AMT actuator is obtained.

[0013] According to a preset temperature gradient, the temperature of the AMT actuator is changed N times, and the re-measured temperature value, the re-measured detection position and the re-measured actual position of the AMT actuator are obtained N times, wherein the temperature information comprises N+1 temperature values, the detection position information comprises N+1 detection positions, and the actual position information comprises N+1 actual positions.

[0014] In a possible implementation, the filtering processing of the detection position information to obtain filtered detection position information comprises:

[0015] Filtering coefficients are set for signal data at different time points of each detection position in the detection position information, each detection position is input into a low-pass filter for filtering to obtain each filtered detection position, and each detection position in the detection position information is a signal wave comprising a plurality of time points and having corresponding signal data at each time point.

[0016] In a possible implementation, the determination of the chromosome of the genetic algorithm as the first initial weight and the first initial threshold of the neural network comprises:

[0017] The filtered detection position information and the temperature information are input into the neural network as inputs of the neural network, the actual position information is input into the neural network as an output of the neural network, the first weight and the first threshold between the input layer and the hidden layer of the neural network are constructed, and the second weight and the second threshold between the hidden layer and the output layer of the neural network are constructed.

[0018] According to the number of the first weight, the first threshold, the second weight and the second threshold, the length of the chromosome of the genetic algorithm is confirmed, and a plurality of chromosomes are generated to form a population.

[0019] Each chromosome in the population is input into the neural network as the second initial weight and the second initial threshold of the neural network, a plurality of fitnesses are obtained, the second initial weight comprises a first value corresponding to the first weight and a first value corresponding to the second weight, and the second initial threshold comprises a first value corresponding to the first threshold and a first value corresponding to the second weight.

[0020] selecting part of the chromosomes from the plurality of chromosomes according to the fitness and a preset iteration number, and obtaining an optimal chromosome;

[0021] taking a value corresponding to the optimal chromosome as a first initial weight and a first initial threshold, wherein the first initial weight comprises a second value corresponding to the first weight and a second value corresponding to the second weight, and the first initial threshold comprises a second value corresponding to the first threshold and a second value corresponding to the second weight.

[0022] In a possible implementation, the selecting part of the chromosomes from the plurality of chromosomes according to the fitness and a preset iteration number, and obtaining an optimal chromosome, comprises:

[0023] obtaining selection probabilities of the plurality of chromosomes according to the fitness;

[0024] selecting chromosomes for crossover according to the selection probabilities, and selecting chromosomes for mutation according to the selection probabilities;

[0025] re-obtaining the fitness and the selection probabilities of the chromosomes after mutation according to a preset iteration number, and repeating the crossover process and the mutation process.

[0026] In a possible implementation, the taking the filtered detection position information and the temperature information as input, taking the actual position information as output, inputting the neural network for training, and obtaining a temperature compensation model, comprises:

[0027] one-to-one correspondence between N+1 temperature values of the temperature information, N+1 detection positions of the filtered detection position information, and N+1 actual positions in the actual position information, to construct N+1 samples;

[0028] selecting part of the samples as a training set, and inputting data in the training set after normalization processing into the neural network to obtain an output value and an expected value of the neural network;

[0029] updating the first initial weight and the first initial threshold according to an error between the output value and the expected value, until a trained temperature compensation model is obtained.

[0030] In a possible implementation, the normalization processing of the data in the training set comprises:

[0031] obtaining a first normalization parameter according to a maximum sample and a minimum sample in the training set;

[0032] According to each sample in the training set and the minimum sample, a second normalization parameter is obtained;

[0033] According to the first normalization parameter and the second normalization parameter, a normalized training set is obtained.

[0034] In a second aspect, the present application provides an AMT position sensor temperature compensation device, comprising:

[0035] An acquisition module is configured to acquire temperature information, detected position information and actual position information of an AMT actuator under different temperatures and different working displacements;

[0036] A filtering module is configured to filter the detected position information to obtain filtered detected position information;

[0037] A processing module is configured to determine a chromosome by using a genetic algorithm as a first initial weight and a first initial threshold of a neural network;

[0038] A training module is configured to input the filtered detected position information and the temperature information as inputs, input the actual position information as an output, and input the neural network for training to obtain a temperature compensation model.

[0039] In a third aspect, the present application provides an AMT position sensor temperature compensation device, comprising at least one processor and a memory;

[0040] The memory stores computer execution instructions;

[0041] The at least one processor executes the computer execution instructions stored in the memory, so that the at least one processor executes the AMT position sensor temperature compensation method as described above.

[0042] In a fourth aspect, the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the steps of the AMT position sensor temperature compensation method as described above.

[0043] The AMT position sensor temperature compensation method, device, equipment and medium provided by the present application acquire temperature information, detected position information and actual position information of an AMT actuator under different temperatures and different working displacements, filter the detected position information to obtain filtered detected position information, determine a chromosome by using a genetic algorithm as a first initial weight and a first initial threshold of a neural network, input the filtered detected position information and the temperature information as inputs, input the actual position information as an output, input the neural network for training, and obtain a temperature compensation model.

[0044] In the above method, the temperature information, the detection position information and the actual position information of the AMT actuator under different temperatures and different working displacements are detected, the temperature information and the position information of the real scene are fully utilized, the detection position information is filtered to reduce the influence of noise on information collection, the filtered detection position information and the temperature information are taken as the neural network input, the actual position information is taken as the neural network output, the neural network is constructed, and the chromosome determined through the genetic algorithm is taken as the first initial weight and the first initial threshold of the neural network, so as to ensure that the neural network is fully temperature compensated and the obtained position information of the AMT actuator is accurate enough. BRIEF DESCRIPTION OF DRAWINGS

[0045] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0046] Figure 1 A scene schematic diagram of AMT position sensor temperature compensation provided for the embodiments of the present application;

[0047] Figure 2 A flowchart of an AMT position sensor temperature compensation method provided for the embodiments of the present application Figure 1 ;

[0048] Figure 3 A flowchart of an AMT position sensor temperature compensation method provided for the embodiments of the present application Figure 2 ;

[0049] Figure 4 A flowchart of an AMT position sensor temperature compensation method provided for the embodiments of the present application Figure 3 ;

[0050] Figure 4 A flowchart of an AMT position sensor temperature compensation method provided for the embodiments of the present application Figure 5 ;

[0051] Figure 6 A diagram of an AMT position sensor temperature compensation device provided for the embodiments of the present application;

[0052] Figure 1 A hardware schematic diagram of an AMT position sensor temperature compensation device provided for the embodiments of the present application. DETAILED DESCRIPTION

[0053] In order to make the purposes, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some but not all of the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0054] The AMT is a mechanical automatic transmission, which comprises an AMT controller and an AMT actuator. The AMT controller can control the AMT actuator (such as a transmission) to perform corresponding work. When the AMT position sensor obtains a change in the working displacement of the transmission, the change can be fed back to the AMT controller to enable the AMT controller to perform a gear shifting operation.

[0055] However, the temperature of the AMT actuator can affect the position acquisition result of the AMT position sensor, resulting in inaccurate position acquisition of the AMT actuator and ultimately inaccurate gear shifting. The existing processing method cannot fully consider the influence of temperature on the position measurement of the AMT position sensor, and additional hardware structures (such as a bilinear interpolation algorithm module) are added to ensure accuracy. Further improvement is needed.

[0056] Therefore, the present application provides a processing method that does not change the hardware structure and fully considers the influence of temperature on position acquisition.

[0057] A method for compensating for the temperature of an AMT position sensor is provided below in combination with the drawings and specific embodiments.

[0058] Figure 1 A scene diagram for compensating for the temperature of an AMT position sensor is provided in the embodiments of the present application. As shown in Figure 2 the system comprises an AMT controller 101, an AMT actuator 102, an AMT position sensor 103 and a temperature sensor 104;

[0059] The AMT controller 101 is configured to control the AMT actuator 102 to achieve the purpose of shifting / gear shifting. The AMT actuator 102 mainly comprises three schemes: a motor scheme, a hydraulic scheme and a pneumatic scheme. The motor scheme is to directly install a motor to the AMT actuator 102, and the AMT actuator 102 controls the rotation of the motor to realize gear shifting. The hydraulic scheme and the pneumatic scheme realize gear shifting through hydraulic transmission and pneumatic transmission, respectively.

[0060] The AMT position sensor 103 can be used to collect the working displacement information of the AMT actuator 102, to obtain the detection position information, and different displacements correspond to different gear positions.

[0061] Figure 1 A flowchart of an AMT position sensor temperature compensation method provided by an embodiment of the present application Figure 2 As shown in Figure 3 , the method comprises the following steps.

[0062] S201, under different temperature and different working displacement conditions, obtaining the temperature information, detection position information and actual position information of the AMT actuator; wherein the detection position information is obtained by the AMT position sensor.

[0063] The different working displacements of the AMT actuator can correspond to different gear positions, and the working displacement can be adjusted; since the temperature can affect the working displacement of the AMT actuator, different temperature conditions are set, and the temperature information, detection position information and actual position information of the AMT actuator under different working displacements are collected under different temperature conditions; wherein the actual position information is obtained by a position survey tool.

[0064] For example, some AMT actuators have a paddle structure, and the paddle has different opening sizes (unit: degree), i.e., different positions; when the paddle is controlled to change, the opening of the paddle changes, but due to the temperature influence, the collection of the position of the paddle by the AMT position sensor can not be accurate enough, at this time, the temperature and the actual position of the AMT actuator are introduced as a reference.

[0065] S202, filtering the detection position information to obtain the filtered detection position information.

[0066] In order to reduce the influence of noise and other problems on the detection position information obtained by the AMT position sensor, a filter is used to process the detection position information to obtain the filtered detection position information; wherein the filtering method includes using a low-pass filter.

[0067] S203, using the chromosome determined by the genetic algorithm as the first initial weight and the first initial threshold of the neural network.

[0068] The filtered detection position information and the temperature information are taken as inputs of the neural network, the actual position information is taken as an output of the neural network, the neural network is constructed, and weights and thresholds required by the neural network are obtained; according to the weights and the thresholds, it is confirmed how many numerical values are required by the neural network, and according to the numerical values, the length (i.e., the number of bits) of a chromosome in a genetic algorithm is confirmed;

[0069] According to the values of different bits of the chromosome determined by the genetic algorithm, the values are taken as first initial weights and first initial thresholds of the neural network.

[0070] S204, the filtered detection position information and the temperature information are taken as inputs, the actual position information is taken as an output, the neural network is input for training, and a temperature compensation model is obtained.

[0071] After the first initial weights and the first initial thresholds of the neural network are confirmed, the neural network has a starting point for training, the filtered detection position information and the temperature information are taken as inputs, the actual position information is taken as an output, the neural network is input for training, the first initial weights and the first initial thresholds are updated constantly, until a prediction result reaches a preset accuracy or a training round number reaches a preset round number, then the training is ended, and the temperature compensation model is obtained; the temperature compensation model can be used to obtain position information of an AMT actuator in actual driving, and has high accuracy.

[0072] In the embodiment of the application, the temperature information, the detection position information and the actual position information of the AMT actuator under different temperatures and different working displacements are detected, the temperature information and the position information of a real scene are fully utilized, the detection position information is filtered to reduce the influence of noise on information collection, the filtered detection position information and the temperature information are taken as inputs of the neural network, the actual position information is taken as an output of the neural network, the neural network is constructed, and the chromosome determined by the genetic algorithm is taken as the first initial weights and the first initial thresholds of the neural network, so as to ensure sufficient temperature compensation for the neural network and to ensure that the obtained position information of the AMT actuator is accurate enough.

[0073] Figure 2 A flowchart of an AMT position sensor temperature compensation method provided in the embodiment of the application Figure 3 As shown in Figure 4 , the method comprises the following steps.

[0074] S301, an initial temperature is set for the AMT actuator, an initial temperature value of the AMT actuator is obtained through a temperature sensor, an initial detection position of the AMT actuator under different working displacements is obtained through an AMT position sensor, and an initial actual position of the AMT actuator is obtained.

[0075] The temperature of the AMT actuator in the ambient temperature state at the time of measurement can be taken as the initial temperature, and the initial temperature value of the AMT actuator is obtained by the temperature sensor at this time;

[0076] The working displacement of the AMT actuator is changed, and the initial detection position of the AMT actuator is obtained by the AMT position sensor and the initial actual position of the AMT actuator is obtained by the position survey tool under different working displacements.

[0077] S302, according to the preset temperature gradient, the temperature of the AMT actuator is changed N times, and the re-measured temperature value, the re-measured detection position and the re-measured actual position of the AMT actuator are obtained N times; wherein, the temperature information includes N+1 temperature values, the detection position information includes N+1 detection positions, and the actual position information includes N+1 actual positions.

[0078] According to the preset temperature gradient, the ambient temperature is changed N times, which can be increased or decreased; each change is re-measured to obtain the corresponding data of the AMT actuator by the temperature sensor, the AMT position sensor and the position survey tool, and N+1 temperature values, N+1 detection positions and N+1 actual positions are obtained.

[0079] S303, a filter coefficient is set for the signal data of each detection position at different times in the detection position information, each detection position is input into a low-pass filter for filtering to obtain each filtered detection position; wherein, each detection position in the detection position information is a signal wave including a plurality of times, and each time has corresponding signal data.

[0080] Each detection position in the detection position information is filtered by a FIR (Finite Impulse Response) low-pass filter, and the filter equation is:

[0081] y(k)=a0x(k)+a1x(k-1)+a2x(k-2)+…

[0082] Wherein, y(k) is a certain filtered detection position, x(k) is the signal data of a certain detection position k, x(k-1) is the signal data of a certain detection position k-1, x(k-2) is the signal data of a certain detection position k-2, and a0, a1 and a2 are filter coefficients.

[0083] In the embodiments of the application, specific temperature conditions and working displacement conditions are set, and the required data is obtained by the temperature sensor, the AMT position sensor and the position survey tool; the data obtained by the AMT position sensor is input into a low-pass filter for filtering to reduce the influence of noise and ensure that the subsequent processing process is more accurate.

[0084] Figure 3 A flowchart of an AMT position sensor temperature compensation method provided for an embodiment of the present application Figure 4 . As Figure 5 shown, the method comprises:

[0085] S401, the filtered detection position information and the temperature information are taken as inputs of the neural network, the actual position information is taken as an output of the neural network, a first weight and a first threshold value between an input layer and a hidden layer of the neural network are constructed, and a second weight and a second threshold value between the hidden layer and an output layer of the neural network are constructed.

[0086] The neural network has two inputs and one output, corresponding to the first weight and the first threshold value of the nodes between the input layer and the hidden layer of the neural network constructed:

[0087]

[0088] wherein α1, α2…α k are parameters of the nodes of the hidden layer; w 1,1 , w 2,1 …w k,1 , w 1,2 , w 2,2 …w k,2 are the first weights; x1 is the first input; x2 is the second input; b1, b2…b k are the first threshold values; the nodes of the hidden layer are f(α i ):

[0089]

[0090] wherein e is a natural logarithm.

[0091] Corresponding to the second weight and the second threshold value of the nodes between the hidden layer and the output layer of the neural network constructed:

[0092]

[0093] wherein w′ 1,1 , w′ 1,2 …w′ 1,k are the second weights; b′1 is the second threshold value; β is the output.

[0094] S402, according to the number of the first weight, the first threshold value, the second weight and the second threshold value, confirming the chromosome length of the genetic algorithm, and generating a plurality of chromosomes to constitute a population.

[0095] Data in the first weight, the first threshold, the second weight and the second threshold are arranged in order into a row, and the row of data is taken as a chromosome length, wherein the first weight has 2*k data, the first threshold has k data, the second weight has k data, and the second threshold has 1 data, and the chromosome has (2*k+k+k+1) data bits;

[0096] A plurality of chromosomes are randomly generated according to the chromosome length, and the data bits corresponding to the chromosomes have randomly generated values, and the plurality of chromosome clusters can be referred to as a population.

[0097] S403, each chromosome in the population is taken as a second initial weight and a second initial threshold of a neural network, and is input into the neural network to obtain a plurality of fitness, wherein the second initial weight includes a first value corresponding to the first weight and a first value corresponding to the second weight, and the second initial threshold includes a first value corresponding to the first threshold and a first value corresponding to the second weight.

[0098] Each randomly generated chromosome is taken as a second initial weight and a second initial threshold of a neural network, and is input into the neural network to obtain an output value and an expected value of the neural network;

[0099] The fitness g of the jth chromosome is obtained according to the output value and the expected value of the neural network. j The formula is:

[0100]

[0101] Wherein, y out_i is the output of the i th sample input into the neural network, y i is the expected value of the i th sample input into the neural network; N+1 temperature values, N+1 detection positions and N+1 actual positions can construct N+1 samples, from which I samples are collected to input into the neural network.

[0102] S404, according to the fitness and a preset iteration number, a part of the chromosomes is selected for iteration to obtain an optimal chromosome.

[0103] According to the fitness, the selection probability of each chromosome can also be obtained, and a part of the chromosomes is randomly selected from the cluster according to the selection probability; and according to the preset iteration number, the selected part of the chromosomes is subjected to crossover operation and mutation operation until the preset iteration number is reached, and in this process, the fitness and the preset iteration number are calculated again each time the iteration is performed once;

[0104] Further, according to the fitness, the selection probability of the plurality of chromosomes is obtained.

[0105] The selection probability P of the jth chromosome is obtained according to the fitness.j The formula is:

[0106]

[0107] Where p is the number of chromosomes.

[0108] Chromosomes are selected for crossover based on the selection probability, and chromosomes are selected for mutation based on the selection probability.

[0109] The formula for selecting chromosomes for crossover is:

[0110]

[0111] Among them, a mu a is a real number representing the u-th data bit of the m-th chromosome; nu is a real number for the u-th data bit of the n-th chromosome; rand is a random number between (0,1).

[0112] The formula for processing mutations on the selected chromosome is:

[0113]

[0114]

[0115] Among them, a wv a is a real number representing the v-th data bit of the w-th chromosome; max For a wv The upper limit of the value of a; min For a wv The lower bound of the value; rand′ is a random number between (0,1); s is the current iteration number; s max This represents the maximum number of iterations.

[0116] Based on a preset number of iterations, the fitness and selection probability of each mutated chromosome are re-obtained, and the crossover process and the mutation process are repeated.

[0117] Each iteration re-obtains the fitness and selection probability of each mutated chromosome using the above method, and repeats the crossover and mutation process until the preset number of iterations (maximum number of iterations) is reached.

[0118] S405. The value corresponding to the optimal chromosome is used as the first initial weight and the first initial threshold, wherein the first initial weight includes a second value corresponding to the first weight and a second value corresponding to the second weight, and the first initial threshold includes a second value corresponding to the first threshold and a second value corresponding to the second weight.

[0119] After the iteration ends, the value corresponding to each data bit of a certain best chromosome is taken as the first initial weight and the first initial threshold value; at this time, the data of each data bit of the chromosome has changed, and the specific value is the second value of the first weight and the second value of the second weight, and the second value of the first threshold value and the second value of the second weight.

[0120] In the embodiment of the application, through the specific processing process of the genetic algorithm, the neural network selects appropriate first initial weight and first initial threshold value, so that the neural network can subsequently accurately train the input data and obtain a temperature compensation model.

[0121] Figure 4 A flowchart of an AMT position sensor temperature compensation method provided in the embodiment of the application Figure 5 As shown in Figure 6 , the method comprises:

[0122] S501, one-to-one correspondence is established between N+1 temperature values of the temperature information, N+1 detection positions of the filtered detection position information, and N+1 actual positions of the actual position information, and N+1 samples are constructed.

[0123] Under the same temperature condition and the same working displacement condition, the measured temperature value, the detection position, and the actual position are one-to-one corresponding, and N+1 samples are obtained;

[0124] The N+1 samples can be expressed as: [d r T r D r ], r = 1, 2, …, N+1; wherein, d r is the temperature value of the rth sample; T r is the detection position of the rth sample; D r is the actual position of the rth sample.

[0125] S502, select part of the samples as a training set in the sample, and input the data in the training set after normalization processing into the neural network to obtain the output value and the expected value of the neural network.

[0126] Select part of the samples as a training set in the sample, select part of the samples as a test set, and train the neural network by taking the training set as the input of the neural network; before input, the data in the training set is normalized:

[0127] In an example, a first normalization parameter is obtained according to the maximum sample and the minimum sample in the training set;

[0128] A second normalization parameter is obtained according to each sample and the minimum sample in the training set;

[0129] According to the first normalization parameter and the second normalization parameter, a normalized training set is obtained.

[0130] The specific formula for normalization is:

[0131] X1=x-x min

[0132] X2=x max -x min

[0133] x'=X1 / X2

[0134] Wherein, X1 is the second normalization parameter; x is one of each sample in the training set; x min is the minimum sample in the training set; X2 is the first normalization parameter; x max is the maximum sample in the training set; x' is the normalized sample; after normalizing each sample, the normalized training set is obtained; the normalized training set is input into the output value and the expected value of the neural network.

[0135] S503, according to the error of the output value and the expected value, update the first initial weight and the first initial threshold, until the trained temperature compensation model is obtained.

[0136] According to the error formula:

[0137]

[0138] The error E of the output value and the expected value is obtained; wherein, y out is the output of the neural network; y is the expected value of the neural network, and the neural network can select BP neural network;

[0139] Based on the gradient descent strategy, the error is back propagated, and the new first initial weight and the first initial threshold are updated; wherein, the first initial weight adjustment value is η is the learning rate of the neural network, Δ is the differential symbol, is the differential symbol; the first initial threshold adjustment value is

[0140] When the error of a certain time is less than the preset error, or the training number of the neural network reaches the preset number, the training is completed to obtain the trained neural network, that is, the temperature compensation model.

[0141] When the temperature compensation model is actually used, only the real-time detection position information obtained by the AMT position sensor and the real-time temperature information obtained by the temperature sensor are needed as the model input.

[0142] In the embodiment of the present application, a training set is constructed, and the training set is normalized; the normalized training set is input into the constructed neural network for training, so as to obtain a temperature compensation model capable of compensating temperature difference and ensuring that the position information of the AMT actuator can be accurately acquired.

[0143] Figure 6 An AMT position sensor temperature compensation device provided by the embodiment of the present application is shown in FIG. 1, which comprises an acquisition module 601, a filtering module 602, a processing module 603 and a training module 604. Figure 7

[0144] The acquisition module 601 is configured to acquire temperature information, detection position information and actual position information of an AMT actuator under different temperatures and different working displacements.

[0145] The acquisition module 601 is further configured to set an initial temperature for the AMT actuator, acquire an initial temperature value of the AMT actuator by a temperature sensor, acquire an initial detection position of the AMT actuator under different working displacements by an AMT position sensor, and acquire an initial actual position of the AMT actuator.

[0146] According to a preset temperature gradient, the temperature of the AMT actuator is changed N times, and the re-measured temperature value, the re-measured detection position and the re-measured actual position of the AMT actuator are acquired N times; wherein the temperature information comprises N+1 temperature values, the detection position information comprises N+1 detection positions, and the actual position information comprises N+1 actual positions.

[0147] The filtering module 602 is configured to perform filtering processing on the detection position information to obtain filtered detection position information.

[0148] The filtering module 602 is configured to set a filtering coefficient for signal data at different time points in each detection position in the detection position information, input each detection position into a low-pass filter for filtering to obtain each filtered detection position; wherein each detection position in the detection position information comprises signal waves at multiple time points, and each time point has corresponding signal data.

[0149] The processing module 603 is configured to determine a chromosome by using a genetic algorithm as a first initial weight and a first initial threshold of a neural network.

[0150] The processing module 603 is further configured to input the filtered detection position information and the temperature information into the neural network as inputs of the neural network, input the actual position information into the neural network as an output of the neural network, construct a first weight and a first threshold between an input layer and a hidden layer of the neural network, and construct a second weight and a second threshold between the hidden layer and an output layer of the neural network.​

[0151] According to the number of the first weight, the first threshold, the second weight and the second threshold, the length of the chromosome of the genetic algorithm is confirmed, and a plurality of chromosomes are generated to form a population;

[0152] Each chromosome in the population is input into the neural network as a second initial weight and a second initial threshold of the neural network respectively, a plurality of fitness values are obtained, wherein the second initial weight includes a first value corresponding to the first weight and a first value corresponding to the second weight, and the second initial threshold includes a first value corresponding to the first threshold and a first value corresponding to the second weight;

[0153] According to the fitness values and a preset number of iterations, part of the chromosomes are selected for iteration to obtain an optimal chromosome;

[0154] The value corresponding to the optimal chromosome is taken as a first initial weight and a first initial threshold, wherein the first initial weight includes a second value corresponding to the first weight and a second value corresponding to the second weight, and the first initial threshold includes a second value corresponding to the first threshold and a second value corresponding to the second weight.

[0155] The training module 604 is configured to input the filtered detection position information and the temperature information as inputs, and the actual position information as an output, into the neural network for training to obtain a temperature compensation model.

[0156] The training module 604 is further configured to one-to-one correspond N+1 temperature values of the temperature information, N+1 detection positions of the filtered detection position information and N+1 actual positions of the actual position information to construct N+1 samples.

[0157] Part of the samples are selected as a training set, and data in the training set are normalized and input into the neural network to obtain an output value and an expected value of the neural network;

[0158] According to an error between the output value and the expected value, the first initial weight and the first initial threshold are updated until the trained temperature compensation model is obtained.

[0159] The application also provides an AMT position sensor temperature compensation device, comprising at least one processor and a memory.

[0160] The memory stores computer execution instructions.

[0161] The at least one processor executes the computer execution instructions stored in the memory, so that the at least one processor executes the AMT position sensor temperature compensation method.

[0162] Figure 7 A hardware schematic diagram of the AMT position sensor temperature compensation device provided by the embodiment of the present application is shown in FIG. 7. As shown in the figure, the AMT position sensor temperature compensation device 70 provided by the embodiment of the present application comprises at least one processor 701 and a memory 702. The device 70 further comprises a communication component 703. The processor 701, the memory 702 and the communication component 703 are connected through a bus 704. Figure 7

[0163] In the implementation process, the at least one processor 701 executes the computer execution instructions stored in the memory 702, so that the at least one processor 701 executes the AMT position sensor temperature compensation method as described above.

[0164] The specific implementation process of the processor 701 can refer to the method embodiments described above, which has similar implementation principles and technical effects, and will not be described here again in the embodiment.

[0165] In the above-mentioned ​ In the embodiment shown in the figure, it should be understood that the processor can be a central processing unit (English: Central Processing Unit, for short: CPU), and can also be other general-purpose processors, digital signal processors (English: Digital Signal Processor, for short: DSP), application specific integrated circuits (English: Application Specific Integrated Circuit, for short: ASIC) and the like. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor or the like. The steps of the method disclosed in combination with the application can be directly embodied as the execution of the hardware processor, or executed by the combination of hardware and software modules in the processor.

[0166] The memory can contain a random access memory (Random Access Memory, RAM), and can also include a non-volatile memory (Non-volatile Memory, NVM), for example, at least one disk memory.

[0167] The bus can be an industry standard architecture (Industry Standard Architecture, ISA) bus, a peripheral component (Peripheral Component, PCI) bus or an extended industry standard architecture (Extended Industry Standard Architecture, EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, the bus in the drawings of the present application does not limit only one bus or one type of bus. ​

[0168] The application further provides a computer readable storage medium, wherein computer execution instructions are stored in the computer readable storage medium, and when a processor executes the computer execution instructions, the AMT position sensor temperature compensation method is realized.

[0169] The computer readable storage medium can be realized by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer.

[0170] An exemplary readable storage medium is coupled to the processor, so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be an integral part of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor and the readable storage medium can also exist as discrete components in the device.

[0171] The division of the units is only a logical function division, and when actually implemented, there can be another division mode, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the coupling or direct coupling or communication connection between the units or components shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0172] The units described as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, that is, they can be located in one place, or they can be distributed on a plurality of network units. According to actual needs, part or all of the units can be selected to achieve the purpose of the embodiment scheme.

[0173] In addition, the functional units in each embodiment of the application can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit.

[0174] If the functions are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0175] It can be understood by those skilled in the art that all or part of the steps of the above-mentioned method embodiments can be completed by program instruction related hardware. The aforementioned program can be stored in a computer readable storage medium. The program executes the steps of the above-mentioned method embodiments when executed; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk, and various media that can store program codes.

[0176] Finally, it should be noted that: those skilled in the art will easily think of other embodiments of the present application after considering the specification and practicing the application disclosed herein. The present application is intended to cover any variations, uses or adaptations of the present application that follow the general principles of the present application and include common knowledge or conventional technical means in the art that are not disclosed in the present application, and is not limited to the precise structure described above and shown in the drawings, and various modifications and changes can be made without departing from the scope thereof. The scope of the present application is only limited by the appended claims.

Claims

1. An AMT position sensor temperature compensation method, characterized in that, The application relates to a temperature compensation method for an AMT (automatic mechanical transmission) actuator. The temperature information, the detection position information and the actual position information of the AMT actuator are acquired under different temperatures and different working displacements, wherein the detection position information is acquired by an AMT position sensor; The detection position information is filtered to obtain filtered detection position information; The chromosome determined by the genetic algorithm is used as the first initial weight and the first initial threshold of the neural network; The filtered detection position information and the temperature information are used as inputs, the actual position information is used as output, the neural network is input for training, and a temperature compensation model is obtained; The temperature information, the detection position information and the actual position information of the AMT actuator are acquired under different temperatures and different working displacements, including: An initial temperature is set for the AMT actuator, the initial temperature value of the AMT actuator is acquired by a temperature sensor, the initial detection position of the AMT actuator under different working displacements is acquired by an AMT position sensor, and the initial actual position of the AMT actuator is acquired; According to a preset temperature gradient, the temperature of the AMT actuator is changed N times, and the re-measured temperature value, the re-measured detection position and the re-measured actual position of the AMT actuator are acquired N times; wherein the temperature information comprises N+1 temperature values, the detection position information comprises N+1 detection positions, and the actual position information comprises N+1 actual positions; The detection position information is filtered to obtain filtered detection position information, including: Filtering coefficients are set for the signal data of each detection position at different moments in the detection position information, each detection position is input into a low-pass filter for filtering to obtain each filtered detection position; wherein each detection position in the detection position information is a signal wave comprising a plurality of moments, and each moment has corresponding signal data; The filtered detection position information and the temperature information are used as inputs, the actual position information is used as output, the neural network is input for training, and a temperature compensation model is obtained, including: N+1 samples are constructed by one-to-one correspondence of the N+1 temperature values of the temperature information, the N+1 detection positions of the filtered detection position information and the N+1 actual positions in the actual position information; Part of the samples in the samples are selected as a training set, data in the training set is normalized and input into the neural network, and the output value and the expected value of the neural network are acquired; According to the error between the output value and the expected value, the first initial weight and the first initial threshold are updated until the trained temperature compensation model is obtained.

2. The method of claim 1, wherein, The chromosome determined by the genetic algorithm is used as the first initial weight and the first initial threshold of the neural network, including: constructing first weights and first thresholds between an input layer and a hidden layer of the neural network by taking the filtered detection position information and the temperature information as inputs of the neural network and taking the actual position information as an output of the neural network; confirming a chromosome length of the genetic algorithm according to a quantity of the first weights, the first thresholds, the second weights and the second thresholds, and generating a plurality of chromosomes to constitute a population; inputting each chromosome in the population as second initial weights and second initial thresholds of the neural network respectively, and obtaining a plurality of fitnesses by inputting the neural network, wherein the second initial weights include first values corresponding to the first weights and the second weights, and the second initial thresholds include first values corresponding to the first thresholds and the second weights; selecting part of the chromosomes to iterate according to the fitnesses and a preset iteration number, and obtaining an optimal chromosome; taking values corresponding to the optimal chromosome as first initial weights and first initial thresholds, wherein the first initial weights include second values corresponding to the first weights and the second weights, and the first initial thresholds include second values corresponding to the first thresholds and the second weights.

3. The method of claim 2, wherein, The selecting part of the chromosomes to iterate according to the fitnesses and the preset iteration number, and obtaining the optimal chromosome, includes: obtaining selection probabilities of the plurality of chromosomes according to the fitnesses; selecting chromosomes to cross according to the selection probabilities, and selecting chromosomes to mutate according to the selection probabilities; re-obtaining the fitnesses and the selection probabilities of the chromosomes after mutation according to the preset iteration number, and repeating the crossing process and the mutation process.

4. The method of claim 1, wherein, The normalizing the data in the training set includes: obtaining a first normalization parameter according to a maximum sample and a minimum sample in the training set; obtaining a second normalization parameter according to each sample and the minimum sample in the training set; obtaining a normalized training set according to the first normalization parameter and the second normalization parameter.

5. An AMT position sensor temperature compensation device, characterized by, includes: an obtaining module, configured to obtain temperature information, detection position information and actual position information of an AMT actuating mechanism under conditions of different temperatures and different working displacements; a filtering module, configured to perform filtering processing on the detection position information to obtain filtered detection position information; a processing module, configured to take a chromosome determined by a genetic algorithm as first initial weights and first initial thresholds of a neural network; a training module, configured to take the filtered detection position information and the temperature information as inputs and take the actual position information as an output, and input the neural network for training to obtain a temperature compensation model; the obtaining module is specifically configured to: Setting an initial temperature for the AMT actuator, obtaining an initial temperature value of the AMT actuator through a temperature sensor, obtaining an initial detection position of the AMT actuator at different working displacements through an AMT position sensor, and obtaining an initial actual position of the AMT actuator; According to a preset temperature gradient, changing the temperature of the AMT actuator N times, and obtaining a re-measured temperature value, a re-measured detection position, and a re-measured actual position of the AMT actuator N times; wherein the temperature information includes N+1 temperature values, the detection position information includes N+1 detection positions, and the actual position information includes N+1 actual positions; The filtering module is specifically configured to: Set a filtering coefficient for signal data of each detection position at different time points in the detection position information, input each detection position into a low-pass filter for filtering to obtain each filtered detection position; wherein each detection position in the detection position information is a signal wave including multiple time points, and each time point has corresponding signal data; The training module is specifically configured to: Correspond one-to-one the N+1 temperature values of the temperature information, the N+1 detection positions of the filtered detection position information, and the N+1 actual positions in the actual position information to construct N+1 samples; Select part of the samples as a training set in the samples, input the data in the training set after normalization processing into the neural network, obtain an output value and an expected value of the neural network, and update the first initial weight and the first initial threshold according to the error between the output value and the expected value until a trained temperature compensation model is obtained. Comprise:

6. An AMT position sensor temperature compensation apparatus, characterized by, At least one processor and a memory; The memory stores computer execution instructions; The at least one processor executes the computer execution instructions stored in the memory, so that the at least one processor executes the AMT position sensor temperature compensation method according to any one of claims 1-4. The computer program is executed by the processor to realize the steps of the AMT position sensor temperature compensation method according to any one of claims 1-4.

7. A computer readable storage medium having stored thereon a computer program, characterized in that, ​

Images