implement new temprature compensation

2024-01-28 21:52:57 +08:00
parent d43819eac8
commit 4ab46b3f9b
4 changed files with 442 additions and 346 deletions
--- a/arg_fit.py
+++ b/arg_fit.py
@@ -2,49 +2,37 @@
 from scipy.optimize import curve_fit
 import numpy as np
 import matplotlib.pyplot as plt
-import pandas as pd
-import warnings
-warnings.filterwarnings("ignore")
 class TempModel:
-    def __init__(self, amfg, tcc, tcfl, tctl, fmin, fmin_temp):
-        self.amfg = amfg
-        self.tcc = tcc
-        self.tcfl = tcfl
-        self.tctl = tctl
+    def __init__(self, a_a, a_b, b_a, b_b, fmin, fmin_temp):
+        self.a_a=a_a
+        self.a_b=a_b
+        self.b_a=b_a
+        self.b_b=b_b
        self.fmin = fmin
        self.fmin_temp = fmin_temp

-    def _tcf(self, f, df, dt, tctl):
-        tctl = self.tctl if tctl is None else tctl
-        tc = self.tcc + self.tcfl * df + tctl * df * df
-        return f + self.amfg * tc * dt * f
-
-    def compensate(self, freq, temp_source, temp_target, tctl=None):
-        dt = temp_target - temp_source
-        dfmin = self.fmin * self.amfg * self.tcc * \
-                (temp_source - self.fmin_temp)
-        df = freq - (self.fmin + dfmin)
-        if dt < 0.:
-            f2 = self._tcf(freq, df, dt, tctl)
-            dfmin2 = self.fmin * self.amfg * self.tcc * \
-                    (temp_target - self.fmin_temp)
-            df2 = f2 - (self.fmin + dfmin2)
-            f3 = self._tcf(f2, df2, -dt, tctl)
-            ferror = freq - f3
-            freq = freq + ferror
-            df = freq - (self.fmin + dfmin)
-        return self._tcf(freq, df, dt, tctl)
+    def compensate(self, freq, temp_source, temp_target):
+        if self.a_a == None or self.a_b == None or self.b_a == None or self.b_b == None:
+            return freq
+        A=4*(temp_source*self.a_a)**2+4*temp_source*self.a_a*self.b_a+self.b_a**2+4*self.a_a
+        B=8*temp_source**2*self.a_a*self.a_b+4*temp_source*(self.a_a*self.b_b+self.a_b*self.b_a)+2*self.b_a*self.b_b+4*self.a_b-4*(freq-model.fmin)*self.a_a
+        C=4*(temp_source*self.a_b)**2+4*temp_source*self.a_b*self.b_b+self.b_b**2-4*(freq-model.fmin)*self.a_b
+        if(B**2-4*A*C<0):
+            param_c=freq-param_linear(freq-model.fmin,self.a_a,self.a_b)*temp_source**2-param_linear(freq-model.fmin,self.b_a,self.b_b)*temp_source
+            return param_linear(freq-model.fmin,self.a_a,self.a_b)*temp_target**2+param_linear(freq-model.fmin,self.b_a,self.b_b)*temp_target+param_c
+        ax=(np.sqrt(B**2-4*A*C)-B)/2/A
+        param_a=param_linear(ax,self.a_a,self.a_b)
+        param_b=param_linear(ax,self.b_a,self.b_b)
+        return param_a*(temp_target+param_b/2/param_a)**2+ax+model.fmin
+        #print(-param_linear(ax,self.b_a,self.b_b)/2/param_linear(ax,self.a_a,self.a_b))
+        #param_c=freq-param_linear(freq-model.fmin,self.a_a,self.a_b)*temp_source**2-param_linear(freq-model.fmin,self.b_a,self.b_b)*temp_source
+        #return param_linear(freq-model.fmin,self.a_a,self.a_b)*temp_target**2+param_linear(freq-model.fmin,self.b_a,self.b_b)*temp_target+param_c
 def line_fit(x,a,b,c):
    return a*x**2+b*x+c
-def fit(data,tcc,tcfl,tctl):
-    result=[]
-    model.tcc=tcc
-    model.tcfl=tcfl
-    model.tctl-tctl
-    for j in range(len(datas)):
-        for i in range(len(data)):
-            result.append(model.compensate(datas[j][3000],35,data[i]))
-    return result
+def line0(x,a,c):
+    return a*x**2+c
+def line120(x,a,c):
+    return a*x**2-240*a*x+c
 def area_find(temp,freq):
    middle=int(len(temp)/100/2)*100
    i=j=100
@@ -54,7 +42,7 @@ def area_find(temp,freq):
        if(i_flag):
            i=i+100
            if middle-i>=0:
-                linear_params, params_covariance = curve_fit(line_fit, temp[middle-i:middle+j],freq[middle-i:middle+j],maxfev=100000,ftol=1e-10,xtol=1e-20)
+                linear_params, params_covariance = curve_fit(line_fit, temp[middle-i:middle+j],freq[middle-i:middle+j],maxfev=100000,ftol=1e-10,xtol=1e-10)
                minus=line_fit(temp[middle-i:middle+j],linear_params[0],linear_params[1],linear_params[2])-freq[middle-i:middle+j]
            if np.sum(np.square(minus))/len(minus)>threshold:
                i=i-100
@@ -62,73 +50,111 @@ def area_find(temp,freq):
        if(j_flag):
            j=j+100
            if middle+j<=len(freq):
-                linear_params, params_covariance = curve_fit(line_fit, temp[middle-i:middle+j],freq[middle-i:middle+j],maxfev=100000,ftol=1e-10,xtol=1e-20)
+                linear_params, params_covariance = curve_fit(line_fit, temp[middle-i:middle+j],freq[middle-i:middle+j],maxfev=100000,ftol=1e-10,xtol=1e-10)
                minus=line_fit(temp[middle-i:middle+j],linear_params[0],linear_params[1],linear_params[2])-freq[middle-i:middle+j]
            if np.sum(np.square(minus))/len(minus)>threshold:
                j=j-100
                j_flag=False
-    linear_params, params_covariance = curve_fit(line_fit, temp[middle-i:middle+j],freq[middle-i:middle+j],maxfev=100000,ftol=1e-10,xtol=1e-20)
+    linear_params, params_covariance = curve_fit(line_fit, temp[middle-i:middle+j],freq[middle-i:middle+j],maxfev=100000,ftol=1e-10,xtol=1e-10)
    return linear_params
 def data_process(path):
-    data=[]
-    file_path = path  # 替换为你的文件路径
-    with open(file_path, 'r') as file:
+    freq=[]
+    temp=[]
+    with open(path, 'r') as file:
        # 逐行读取文件内容
        lines = file.readlines()
        # 遍历每行内容
        for line in lines:
-            data.append(line.split(','))
-    file.close()
-    full_data=pd.DataFrame(data[1:-1],columns=data[0])
-    temp=np.array(full_data['temp']).astype(np.float32)
-    freq=np.array(full_data['freq']).astype(np.float32)
-    freq=freq[::100]
-    temp=temp[::100]
-    plt.plot(temp[10:],freq[10:])
-    linear_params=area_find(temp,freq)
-    plt.plot(temp,line_fit(temp,linear_params[0],linear_params[1],linear_params[2]))
-    data0=line_fit(np.arange(5,80,0.01),linear_params[0],linear_params[1],linear_params[2])
-    return data0
+            data=line.split(',')
+            try:
+                freq.append(float(data[3]))
+                temp.append(float(data[5]))
+            except:pass
+    dv=int(len(temp)/1000)
+    if dv>1:
+        freq=np.array(freq[::dv])
+        temp=np.array(temp[::dv])
+    plt.plot(temp[20:],freq[20:])
+    #linear_params=area_find(temp[20:],freq[20:])
+    param_bounds=([0,-np.inf,-np.inf],[np.inf,np.inf,np.inf])
+    linear_params, params_covariance = curve_fit(line_fit, temp[20:],freq[20:],bounds=param_bounds,maxfev=100000,ftol=1e-10,xtol=1e-10)
+
+    try:
+        plt.title("Range:"+str(int(np.max(freq[20:])-np.min(freq[20:]))))
+    except:
+        pass
+    axis=-1*linear_params[1]/2/linear_params[0]
+    if(axis>120):
+        linear_params1, params_covariance = curve_fit(line120, temp[20:],freq[20:],bounds=([0,-np.inf],[np.inf,np.inf]),maxfev=100000,ftol=1e-10,xtol=1e-10)
+        plt.plot(temp[20:],line120(temp[20:],linear_params1[0],linear_params1[1]))
+        return [linear_params1[0],-240*linear_params1[0],line120(120,linear_params1[0],linear_params1[1])]
+    elif(axis<0):
+        linear_params1, params_covariance = curve_fit(line0, temp[20:],freq[20:],bounds=([0,-np.inf],[np.inf,np.inf]),maxfev=100000,ftol=1e-10,xtol=1e-10)
+        plt.plot(temp[20:],line0(temp[20:],linear_params1[0],linear_params1[1]))
+        return [linear_params1[0],0,line0(axis,linear_params1[0],linear_params1[1])]
+    plt.plot(temp[20:],line_fit(temp[20:],linear_params[0],linear_params[1],linear_params[2]))
+    linear_params[2]=line_fit(axis,linear_params[0],linear_params[1],linear_params[2])
+    return linear_params
+def param_linear(x,a,b):
+    return a*x+b
+
 while(1):
    plt.figure(figsize=(25, 15))
    paths=['./data1','./data2','./data3','./data4']
-    datas=[]
+    a=[]
+    b=[]
+    freqs=[]
    num=241
-    threshold=int(input('threshold set(recommend start from 250):\n请输入阈值设置(默认推荐250):\n'))
+    #threshold=int(input('threshold set(recommend start from 1000):\n请输入阈值设置(默认推荐1000):\n'))
    try:
        for path in paths:
            plt.subplot(num)
            num+=1
-            datas.append(data_process(path))
+            temp=data_process(path)
+            a.append(temp[0])
+            b.append(temp[1])
+            freqs.append(temp[2])
    except:
        print("please make sure you have move the 4 data file to IDM folder\n请确认你有把4个文件拷到IDM文件夹内")
        break
    #反向求值
-    model=TempModel(1,-2.1429828e-05,-1.8980091e-10,3.6738370e-16,2943053.84,20.33)
-    p0=[-2.1429828e-05,-1.8980091e-10,3.6738370e-16]
-    params, params_covariance = curve_fit(fit,np.arange(5,80,0.01),np.hstack(datas),p0=p0,maxfev=1000000,ftol=1e-10,xtol=1e-10)
+    model=TempModel(None,None,None,None,2943053.8415908813,23.33)
+    linear_params, params_covariance = curve_fit(param_linear, np.array(freqs)-model.fmin,a,maxfev=100000,ftol=1e-10,xtol=1e-10)
+    model.a_a=linear_params[0]
+    model.a_b=linear_params[1]
+    linear_params1, params_covariance = curve_fit(param_linear, np.array(freqs)-model.fmin,b,maxfev=100000,ftol=1e-10,xtol=1e-10)
+    model.b_a=linear_params1[0]
+    model.b_b=linear_params1[1]
    for path in paths:
        plt.subplot(num)
        num+=1
-        data=[]
-        file_path = path  # 替换为你的文件路径
-        with open(file_path, 'r') as file:
+        freq=[]
+        temp=[]
+        with open(path, 'r') as file:
            # 逐行读取文件内容
            lines = file.readlines()
            # 遍历每行内容
            for line in lines:
-                data.append(line.split(','))
-        file.close()
-        full_data=pd.DataFrame(data[1:-1],columns=data[0])
-        temp=np.array(full_data['temp']).astype(np.float32)
-        freq=np.array(full_data['freq']).astype(np.float32)
-        freq=freq[::100]
-        temp=temp[::100]
+                data=line.split(',')
+                try:
+                    freq.append(float(data[3]))
+                    temp.append(float(data[5]))
+                except:pass
+        dv=int(len(temp)/10000)
+        if dv>1:
+            freq=np.array(freq[::dv])
+            temp=np.array(temp[::dv])
+        temp=temp[200:]
+        freq=freq[200:]
        result0=[]
        for i in range(len(temp)):
-            result0.append(model.compensate(freq[i],temp[i],20.66))
-        plt.plot(temp[10:],result0[10:])
-    plt.savefig('fit.png')
+            result0.append(model.compensate(freq[i],temp[i],50))
+        plt.plot(temp,result0)
+        try:
+            plt.title("Range:"+str(int(np.max(result0)-np.min(result0))))
+        except:
+            pass
+    plt.savefig('fit_output.png')
    print('fit result:')
-    print('tc_tcc:'+str(params[0])+'\ntc_tcfl:'+str(params[1])+'\ntc_tctl:'+str(params[2]))
+    print('tc_a_a:'+str(model.a_a)+'\ntc_a_b:'+str(model.a_b)+'\ntc_b_a:'+str(model.b_a)+'\ntc_b_b:'+str(model.b_b))
    break
--- a/idm.py
+++ b/idm.py
--- a/install.sh
+++ b/install.sh
@@ -12,12 +12,13 @@ fi

 # install idm requirements to env
 echo "idm: installing python requirements to env, this may take 10+ minutes."
-sudo apt-get install g++ gfortran
+sudo apt-get install g++ 
+sudo apt-get install gfortran 
+sudo apt-get install libopenblas-dev
 "${KENV}/bin/pip" install -r "${BKDIR}/requirements.txt"

 # update link to idm.py
 echo "idm: linking klippy to idm.py."
-sudo apt-get install g++ gfortran libopenblas-dev
 if [ -e "${KDIR}/klippy/extras/idm.py" ]; then
    rm "${KDIR}/klippy/extras/idm.py"
 fi
--- a/requirements.txt
+++ b/requirements.txt
@@ -2,4 +2,3 @@
 numpy>=1.16.6
 scipy>=1.10.0
 matplotlib>=3.7.0
-pandas>=1.4.2