通过python实现新能源汽车车辆事故定责报告的实现

目录

一，前言：

二，报告的效果展示

​编辑

​编辑

三，python生成pdf报告的实现过程

1，清洗can信号数据

2，清洗车道线数据

3，清洗环境数据（障碍物数据）

4，标签数据清洗

5，通过matplotlib生成报告需要的图片

6，生成doc报告文件，生成pdf文件

---

一，前言：

         新能源汽车行业涉及大量数据，包括销售数据、充电设施数据、车辆性能数据等。利用Python进行数据分析和可视化可以帮助企业更好地理解市场趋势、用户需求和产品表现，为决策提供有力支持。

        Python可以编写自动化脚本，从不同数据源获取数据并生成报告，节省人力和时间成本。这对于需要定期生成报告的企业来说尤其重要，可以保持报告的及时性和准确性。

        新能源汽车行业处于快速发展阶段，技术和创新是竞争的关键。利用Python进行报告生成不仅能提高工作效率，还可以通过引入新的分析方法和工具不断优化报告内容和质量，保持企业在行业中的竞争优势。

二，报告的效果展示

车辆状态

三，python生成pdf报告的实现过程

1，清洗can信号数据

#整车零部件状态
df2 = pd.DataFrame()
df2['DPB激活'] = np.where(df_func['IDB7_DynParkBrkByIDBActive'.lower()] == 1, 1, 0)
df2['ABS激活'] = np.where(df_func['IDB7_ABSActive'.lower()] == 1, 1, 0)
df2['TCS激活'] = np.where(df_func['IDB7_TCSActive'.lower()] == 1, 1, 0)
df2['ESC激活'] = np.where(df_func['IDB7_ESCActive'.lower()] == 1, 1, 0)
df2['ARP激活'] = np.where(df_func['IDB7_ROPACTIVE'.lower()] == 1, 1, 0)
print("================22222========================")
# df2['AVH激活'] = np.where(df_func['IDB7_AVHStatus'.lower()] == 1, 1, 0)
print("================3333========================")
df2['EPB激活'] = np.where(df_func['IDB6_EPBStatus'.lower()].isin([2, 3, 4, 5]), 1, 0)
print("================4444========================")
df2['HDC激活'] = np.where(df_func['IDB7_HDCAtive'.lower()] == 1, 1, 0)
print("================22222========================")

df2['ABS故障'] = np.where(df_func['IDB7_ABSFail'.lower()] == 1, 1, 0)
df2['EBD故障'] = np.where(df_func['IDB7_EBD_Fail'.lower()] == 1, 1, 0)
df2['TCS故障'] = np.where(df_func['IDB7_TCSFail'.lower()] == 1, 1, 0)
df2['ESC故障'] = np.where(df_func['IDB7_ESCFail'.lower()] == 1, 1, 0)
df2['ARP故障'] = np.where(df_func['IDB7_ROPFailure'.lower()] == 1, 1, 0)
print("================22222========================")

df2['ESC处于关闭状态'] = np.where(df_func['IDB7_TCSDisable'.lower()] == 1, 1, 0)
df2['制动盘温度过高'] = np.where(df_func['IDB5_BrakeDiscTempSts'.lower()] == 1, 1, 0)
df2['刹车踏板信号无效'] = np.where(df_func['IDB1_BrakePedalAppliedV'.lower()] == 0, 1, 0)
df2['档位信号无效'] = np.where(df_func['VCU5_ActGearValid'.lower()] == 0, 1, 0)
df2['油门踏板信号无效'] = np.where(df_func['VCU5_AccelPositionValid'.lower()] == 0, 1, 0)
df2['加速度信号无效'] = np.where((df_func['ACU2_LatAccSensorValueV'.lower()] == 0) |
                                      (df_func['ACU2_VehicleDynYawRateV'.lower()] == 0) |
                                      (df_func['ACU2_LongAccSensorValueV'.lower()] == 0), 1, 0)
df2['车速信号无效'] = np.where(df_func['IDB3_VehicleSpdValid'.lower()] == 0, 1, 0)
df2['车辆静止信号无效'] = np.where(df_func['IDB5_VehicleStanstill'.lower()] == 2, 1, 0)
df2['电子档位杆无效'] = np.where(df_func['CS1_GearPositionReqStValid'.lower()] == 0, 1, 0)
df2['方向盘信号无效'] = np.where(df_func['EPS1_SteeringAngleValid'.lower()] == 0, 1, 0)
df2['左方控ACC按键无效'] = np.where(df_func['Region_A_SwError'.lower()] == 1, 1, 0)
df2['轮速信号无效'] = np.where((df_func['IDB4_FLWhlVelocityValid'.lower()] == 0) |
                                     (df_func['IDB4_FRWhlVelocityValid'.lower()] == 0) |
                                     (df_func['IDB4_RRWhlVelocityValid'.lower()] == 0) |
                                     (df_func['IDB4_RLWhlVelocityValid'.lower()] == 0), 1, 0)

print("========================================")
df2['四门两盖任一处于开启状态'] = np.where(
(df_func['BDCS1_HoodAjarSts'.lower()] == 1) | (df_func['BDCS1_TrunkLockSts'.lower()] == 0) |(df_func['DDCU1_FLDoorAjar'.lower()] == 2) | (df_func['DDCU1_RLDoorAjar'.lower()] == 2) |(df_func['PDCU1_FRDoorAjar'.lower()] == 2) | (df_func['PDCU1_RRDoorAjar'.lower()] == 2), 1, 0)
df2['双闪开启'] = np.where(df_func['BDCS1_HazardLampSt'.lower()] == 1, 1, 0)
df2['EPS未标定'] = np.where(df_func['EPS1_HSTWCalibrated'.lower()] == 0, 1, 0)
df2['动力系统ACC扭矩未ready'] = np.where(df_func['VCU14_ACC_Ready'.lower()] == 0, 1, 0)
df2['IDB不支持ACC功能'] = np.where(df_func['IDB5_ACC_Enable'.lower()] != 0, 1, 0)

print("========================================")
df = df.reset_index(drop=False)
df2 = df2.reset_index(drop=False)
df = pd.concat([df, df2], axis=1)
print("df....")
print(df.columns)

2，清洗车道线数据

    try:
        df_lane['send_time'] = pd.to_datetime(df_lane['timestamp'], unit='ms').astype(str)
        df_cat = df_lane[~((df_lane['lane1_linemarkcolor'] == 0) & (df_lane['lane1_type'] == 0) & (df_lane['lane1_status'] == 0))][
            ['send_time', 'lane1_curvature']]
        print("df_cat:")
        print(df_cat)

        df = df_lane[['timestamp']].astype('str')
        df['send_time'] = pd.to_datetime(df_lane['timestamp'], unit='ms').astype(str)

        df['大曲率弯路'] = df_cat['lane1_curvature'] > 0.005
        df1 = df.groupby('send_time')['大曲率弯路'].apply(list).reset_index()
        df1['是否大曲率弯路'] = df1['大曲率弯路'].apply(lambda x: sum(x) > 0)
        df1.to_pickle(savepath + 'df_j3_lane_cleaned_' + vehicle_id + '_' + day + '_' + hour + '_' + bagname + '.pkl')
    except Exception as e:
        print('data cleaning lane error, ', str(e))

3，清洗环境数据（障碍物数据）

        df_obj_filtered = df_obj[df_obj['obj_id'] > 0]
        df['objCount'] = df_obj_filtered.groupby(pd.Grouper(key='send_time'))['obj_id'].transform('count')
        df['objCount']=df['objCount'].fillna(0)
        print("df_obj_filtered....")
        print(df['objCount'])
        df['position'] = (df_obj['obj_px'] ** 2 + df_obj['obj_py'] ** 2).astype(
            'float').apply(np.sqrt)
        condition_1 = [
            df_obj['obj_class'] == 0, df_obj['obj_class'] == 1, df_obj['obj_class'] == 2, df_obj['obj_class'] == 3, df_obj['obj_class'] == 4, df_obj['obj_class'] == 5,
            df_obj['obj_class'] == 6, df_obj['obj_class'] == 7, df_obj['obj_class'] == 8
        ]
        # value_1 = [
        #     '未知', '行人', '自行车', '摩托车', '三轮车', '轿车',
        #     '卡车', '客车公交', '专用车', 'SUV', 'Minibus', 'Ambulance',
        #     'Firetruck', '工程车', 'Animal', '小货车'
        # ]
        value_1 = [
            '未知', '轿车', '摩托车', '客车公交', '小货车', '专用车',
            '行人', '自行车', 'Tricycle']
        df['type_obj'] = np.select(condition_1, value_1)

4，标签数据清洗

        filter_acc_long = signal.lfilter(b, a, df_ego['纵向加速度'])
        df_filter_acc_long = pd.DataFrame(filter_acc_long, columns=['filter_acc_long'])
        df_ego = pd.concat([df_ego, df_filter_acc_long], axis=1).sort_values('timestamp')
        df_ego['lag_filter_acc_long'] = df_ego['filter_acc_long'].shift(1)
        df_ego['diff_filter_acc_long'] = df_ego['filter_acc_long'] - df_ego['lag_filter_acc_long']
        df_ego['lag_time'] = df_ego['timestamp'].shift(1)
        df_ego['diff_time'] = df_ego['timestamp'] - df_ego['lag_time']
        df_ego['jerk_longi'] = df_ego['diff_filter_acc_long'] / df_ego['diff_time'] * 1000
        df_ego['tap_brake'] = np.where(
            (df_ego['jerk_longi'].abs() > 3.65) & (df_ego['filter_acc_long'] < -1.34) & (df_ego['ACC状态'] == 'ACC激活'),
            1, 0)
        df.loc[0, '点刹'] = np.where(1 in df_ego['tap_brake'].unique(), 1, 0)

        df_ego['左变道'] = (df_ego['方向盘转角'] > 20) & (df_ego['车速'] > 15)
        df_ego['右变道'] = (df_ego['方向盘转角'] < -20) & (df_ego['车速'] > 15)
        df.loc[0, '左变道'] = np.where(any(df_ego['左变道'] == True), 1, 0)
        df.loc[0, '右变道'] = np.where(any(df_ego['右变道'] == True), 1, 0)

5，通过matplotlib生成报告需要的图片

def plot_adas(savepath,vehicle_id,day,hour,bagname,df,title,secondary_y =False,annotate=''):
    print("11111111111111111111111111111111111111111")
    #指定画布
    fig1 = plt.figure(figsize=(18,10), tight_layout=True)
    #在fig1上添加子图，行数为1，列数为1
    ax1 = fig1.add_subplot(1, 1, 1)
    # ax1.xaxis.set_major_locator(dates.SecondLocator(interval=1))
    #用于指定如何显示 x 轴上的日期时间信息
    ax1.xaxis.set_major_formatter(dates.DateFormatter('%H:%M:%S'))
    # nan_rows = df[df.isna().any(axis=1)]
    #     # print(nan_rows)

    #kind ： 类型为线型的图形
    #title: 设置标题
    #secondary_y: 这是一个布尔参数，用于指示是否应在图表的右侧创建一个辅助（次要）Y轴。如果设置为True，那么数据将根据右侧的Y轴刻度绘制。
    #mark_right： 这是一个布尔参数，用于指示是否应在图表的右侧标记数据。如果设置为False，数据点将不会在右侧的Y轴上标记。
    ax1 = df.plot(kind='line', ax=ax1, style='-o', title=title, secondary_y = secondary_y, mark_right=False)
    # global trigger_time, vehicle_id, trigger_package
    # ax1.axvline(trigger_time, color = 'r', linestyle='--')
    # ax1.text(trigger_time[0]+pd.Timedelta(200, unit='ms'), df[df.columns[0]].max(), 'Trigger触发时间点', color = 'red',
    #          bbox=dict(boxstyle='round,pad=0.3', fc='yellow', ec='k',lw=1 ,alpha=0.5))
    print("pilot=====================================")
    if secondary_y != False and secondary_y != '转向手力矩':
        #用于设置Y轴的标签文本
        ax1.set_ylabel(df.columns[0])
        #用于设置右侧Y轴的y轴的标签文本。
        ax1.right_ax.set_ylabel(df.columns[1])
    elif secondary_y != False and secondary_y == '转向手力矩':
        ax1.set_ylabel(df.columns[0] + ',' + df.columns[1])
        ax1.right_ax.set_ylabel(df.columns[2])
    else:
        #添加图例
        ax1.legend(loc='best')
    ax1.set_xlabel(None)
    #plt.legend(loc='best')
    #网格线将以虚线的形式绘制在图表上，线宽为0.5，颜色为灰色，以帮助视觉上更好地理解图表中的数据点和趋势
    ax1.grid(ls='--', lw=0.5, color='#4E616C')
    y_min, y_max = ax1.get_ylim()
    x_min, x_max = ax1.get_xlim()
    if title == '方向盘变化':
        ax1.annotate('*根据方向盘转角实际值和请求值的波动频率，可判断是否发生了方向盘抖动。', xy=(x_min, y_min),  xytext=(x_min, y_min-(y_max-y_min)/7), fontsize=18, ha='left')
    if title == '横摆角速率及加速度变化':
        #黄线
        # ax1.right_ax.axhline(y=-0.8, color='orange', linestyle='--')
        # ax1.right_ax.axhline(y=0.8, color='orange', linestyle='--')
        ax1.annotate('*此trigger' + annotate + '(NNP或PILOT处于激活状态、转向灯关闭, 且横向加速度存在2个波峰波谷, 波峰波谷的差值分别大于1和0.6且至少一个幅值大于0.5, 则疑似出现画龙).', xy=(x_min, y_min), xytext=(x_min, y_min - (y_max - y_min) / 7), fontsize=14, ha='left')

    plt.savefig(savepath + vehicle_id + '_' + day + '_' + hour + '_' + bagname + '_' + title + '.png', dpi=600, transparent=False)
    # plt.show()

6，生成doc报告文件，生成pdf文件

        doc=get_main_text(doc, df)

        #添加head...
        doc=add_head(savepath, doc, vehicle_id, day, hour, bagname)

        # 页眉页脚设置
        doc=add_header_footer(doc, catalog_p,savepath)

        report_name = '智驾相关事故定责分析报告'
        doc_name = report_name + '.docx'
        pdf_name = report_name + '.pdf'
        doc_file = savepath + doc_name
        doc.save(savepath + doc_name)
        # pdf_file = savepath + pdf_name  # 要生成的文件：不存在
        pdf_file = savepath
        doc_to_pdf(doc_file, pdf_file)

金三银四大数据面试黄金期，想要获取详细大数据面试题，请加入星球。